League of Replays is primarily supported by these ads. Please consider disabling the AdBlock for our site so we can continue to add new features and improve the site. Thank you.

After Hours with Matchmaking and Lyte

Quote:
Originally Posted by Lyte View Post
Hm, you make some good points but I think there's a misunderstanding somewhere. I never said players end up at 1200 because they are average--we don't know what skill they represent until they play more games. Most of these players will probably still end up around 1200 after hundreds of matches.

The misconception that we're discussing spans a couple different points:

1) Changing the starting Elo for Ranked doesn't actually fix anything -- many players assume new Ranked players starting at 800 Elo 'fixes' matchmaking.

2) The average Elo of the system actually is 1200... but the fact that Bronze is the top 25% is actually a side effect of other issues.

3) After about 150-300 games, most players end up at their true Elo. We're trying to shorten the time it takes to get there because we feel ~300 games is too long.

4) The new players that enter Ranked represent about the same distribution as the whole Ranked population, which is why majority of people end up at 1200 after their placement matches.
It seems to me, and maybe I'm biased here as I, currently a 1200-1300 player, that there sure are a lot of new-to-ranked people in my elo range. I frequently see people with less than 10 games, unrankeds etc. I've checked match histories and it will have been people that play customs, normals, dominion etc. Can only check like the past 10 games or so, but still you can tell they don't play ranked that much or are just starting to.

My point is that 1100-1300 is a really hard range to get out of simply because the quality of your team mates (and opponents) greatly changes from match to match. These people are typically new to ranked and either drag the team they are on down, or are really good and carry the rest of the team to victory when they should be lower. Yet, again, there are other times when you just get a good team vs a horrible team. It's really hard to get consistently good games.


25 months ago

Recently, we collected a lot of community feedback on matchmaking.
[Thread] http://na.leagueoflegends.com/board/....php?t=1827767

I have isolated down a list of issues with matchmaking, and I would love to discuss these with the community as we drive towards some meaningful solutions. Please vote in the poll and join in the discussion!


DISCLAIMER
These are issues I have identified from dialogue with the community. Fixing many of these issues will take some time! I will update the discussion as we develop solutions and have better ETAs.


--- Completed ---

1) Free to Play Champions in Ranked Mode [5/22/2012]
Yegg explains this change here: http://na.leagueoflegends.com/board/...3#post24300123

2) Pick Order Bug in Normal Draft Mode [5/22/2012]
This bug was fixed. Players now have a random chance to be Team Captain in Normal Draft Mode.

3) Intentional Queue Dodging in Ranked to lower their Elo to stomp low Elo players [6/13/2012]
This has been changed. Players now receive a 30 minute then 1 hour penalty for queue dodging in Ranked Mode. This time suspension locks you out of all queues.

4) AFKs in Champion Select Lobby [7/19/2012]
Players who queue up for a match and then AFK force players to wait through the timers until the lobby boots the AFK players out. Alternatively, some players AFK and get assigned a Random Champion, which is also a negative experience for many players. In Ranked Games, Captains tend to see what the bans/picks are like and be able to react accordingly in the next lobby as the players will generally be the same.

5) Premade Matching [7/19/2012]
Currently, the matchmaking system tries to match Premade 5s with Premade 5s; however, we may consider prioritized matching for other Premade types. For example, prioritizing Premade 4s to match with Premade 4s, and finding a solo queue to fill out each team.

6) Duo Queue Prevalence in Ranked [7/19/2012]
Currently, it is possible to get two pairs of Duo-Queue Players per team in Ranked Mode, such that you have a team composed of Duo-Queue, Duo-Queue, and a Solo-Queue. Some data suggests that we should restrict the number of Duo-Queues per team to 1.

7) Level Disparities [7/19/2012]
Some players feel that as a low level, they should never be matched against Level 30s because they feel the match is unfair due to Level 30 rune/mastery advantages. Although these players are highly skilled low levels, they would prefer a different solution than facing an opponent with simply 'stronger stats.'


--- In Progress ---

1) Random Champions in Ranked Mode [6/13/2012]
Some players feel that you should not be able to "Random" a Champion in Ranked Play, and that this feature is used to intentionally troll their team in some use cases.

2) Skilled Ranked Players in Normal Modes [7/10/2012]
For a very small subset of highly skilled Ranked Players, their Ranked Elo is vastly higher than their Normal Elo. When these skilled Ranked Players play a Normal Game they generally play against opponents that are much lower in skill, resulting in lopsided matches.


--- List of Issues ---

[NEW] Extra noise in the ~1200 Elo Bracket
Because new Ranked players start at 1200, there is slightly more noise for the players who actually are 1200 Elo after ~200-300 games.

1) Duo-Queue Elo Disparities in Ranked
When players play as a Duo-Queue in Ranked, they generally have an advantage if their Elos are fairly close. However, when players play as a Duo-Queue in Ranked and their Elos are far apart, they generally are at a disadvantage.

2) Transitioning from Normal to Ranked Mode
Currently, we do not do a great job of educating players on the transition from Normal to Ranked Modes. What would players like to see here? What should the expectations of Ranked be? For example, Ranked could be "Bring your best, every single game!"

3) Provisional Matches in Ranked
When a player joins Ranked for the first time, the system starts them at 1200 Elo and begins their "placement matches." Unlucky strings of losses or lucky streaks of wins can propel a player into an Elo tier they do not belong in. Alternatively, players generally do not like playing with or against players in their placement matches and seeing they have only 1-9 wins.

4) Team Margin of Victory
Some players argue that in epic, close matches, teams should not gain or lose the exact same Elo as a lopsided match. What are good metrics for "team margin of victory" that are focused on promoting team play and not individual performances?



25 months ago

Quote:
Originally Posted by Strawhatxx View Post
man you should work on them all but I have to say the free champs in ranked and number 8 when a player gets higher elo by luck but really sucks. sick of these people who win like 7 games and have a 1400 elo and brag like they so good
This poll does not mean we will only work on the winning selection. The poll is just for us to gauge what is important to our players.


25 months ago

Quote:
Originally Posted by kucing View Post
My concern is none of these issues address the most important problem right now.

1. Afk-er during the game

2. Leaver
Those are in-game behavior offenses and we are working on them, but they are not necessarily related to the matchmaking process.


25 months ago

Quote:
Originally Posted by Sanman View Post
Hey Lyte,

My biggest problem with matchmaking actually isn't listed in the poll at all. I feel queue dodging in normals and in ranked is completely obscene. On a normal night my mates and I usually sit through 10-20 minutes of queues just to play a match because there is no penalty for dodging (I wouldn't call a five or fifteen minute wait a penalty). Honestly I'm not sure why one infraction of dodging warrants only five minutes of waiting. Generally a queue dodge results in 9 people wasting five minutes, give or take a few minutes, no? Why can't the dodger wait 9 times the amount of time he sat in queue for before dodging?

That way people seriously feel the burn of "I just wasted 9xX amount of time."

I've even gone so far as writing an article about it the other week for the website I work for.
I agree, and Yegg and I have been talking a lot recently about what we want to do with queue dodging. Generally, a lot of players admit to queue dodging to avoid what they perceive to be a toxic interaction in chat, so we need a penalty that fits the 'offense.' Currently, -Elo in Ranked Mode doesn't necessarily fit the offense, so we are revising it.


25 months ago

Quote:
Originally Posted by Slogra View Post
I really like this point, Lyte. A lot of people give up or play sloppy when the going gets tough. If the Elo loss would be less in a tighter game, I think players would give it their all more often. Also, thanks for creating the last mega thread, and salutations to this new epic thread as well.

Generally speaking, the length of a game can somewhat indicate how lopsided it is, or is not. Usually a lopsided match ends quickly in either a surrender or a few strong pushes. If a team holds out for 45 minutes or more, that means they at least have a fighting chance, unless the enemy's just dragging it out... but that's to their detriment, anyway.

However, taking away less Elo for longer games might encourage players to refuse to surrender for the wrong reasons. It should probably only play a small part... I was just suggesting time as a factor because it seems to be one of the most obvious factors in judging how close a match is.

The number of towers down and kills can also factor in. We're talking a loss reduction of maybe 1 or 2 Elo for both those factors, tops. It would take the sting off of losing a long round, or a round that the losers felt was close.
Yegg, Zileas and myself have talked about "team margin of victory" before, and like others we started with team differences in gold. Unfortunately, this might encourage teams to 'farm' the opposing team for more gold advantage or otherwise stall a game. A second idea was measuring number of towers destroyed and number of surviving towers. Of course, there's also length of time of a match and how it scales with all the other metrics, or time as an independent metric and winning faster means more gains.

The thing to keep in mind is... anything we 'measure' we would want to keep hidden, and ensure that they always promote team-oriented gameplay.


25 months ago

Quote:
Originally Posted by Indigo Kitsune View Post
Shouldn't this come with an "all of the above" option?
I could have done the poll a dozen different ways From the research, these are all issues; but, I just wanted to have a fun poll where players can be heard on their #1 issue and still post comments on other important issues.

If I did multiple choice, a high number of people would simply vote every option or 90% of the options. If I did an "All of the Above," many players would simply choose that instead. None of these approaches is necessarily 'better,' and all paint similar pictures in the end.


25 months ago

Quote:
Originally Posted by Guttler View Post
I want to vote for every one of the poll options. I hope something gets done eventually to remedy each of the categories.
We will probably tackle some of the easier items first while figuring out the design/coding of the more complicated ones. By no means will we choose a few items and ignore the rest.


25 months ago

Quote:
Originally Posted by Krshna View Post
3) Skilled Ranked Players in Normal Modes

This is the DEAD OPPOSITE of an issue. Why worry about something as trivial as this. The question should be, "Unskilled Normal Players in Ranked Modes", not the other way around. The biggest obstacle to trying to play this game in Ranked is that there is a huge percentage of players who are, plain and simple, unskilled. My last game a Caitlyn went 0/11. She gave up first Blood before minions spawned. She died 5 times by the time she was level 4. Why are players like this in Ranked? If a player plays poorly, consistently enough, they should be blocked from Ranked for a certain amount of time.
To be fair, a large majority of our players play Normal Modes, and it is an issue to them. Unskilled players in Ranked is a separate issue, and I think partially tied to the education on transitioning from Normal Mode -> Ranked Mode.

Secondly, why should a player be banned from competitive play because they perform poorly? To compare it to NBA for example, there is always going to be a team that is the 'worst' and loses a majority of their games. Is that team banned from play next season?

The goal should be to place these players in their proper Elo so they are playing against other opponents that are equally skilled.


25 months ago

Quote:
Originally Posted by Shadaii View Post
I was unsure which category my quandry fell into: 3 or 4. Regardless, my biggest issue is that my mates and I, (I play with between 3 to 5 of my mates at a time) when we play together, frequently get matched against teams of vastly more skill (and often wins) than us. We don't play ranked, we find it too...intense. We just wanna have fun. We occasionally play Draft but as many have said already you can sit there for 30 mins or more while people repeatedly dodge. Also Draft runs into the same problem as ranked mostly (too intense for us). So we like to play normals, much more chilled. The meta is not always played (we find the meta all the time very tedious) and that's ok in normals. The stress and intricate need to counter every single thing that happens from champ select to team fighting is not nearly as necessary.

So my issue is that when we group we are almost always pitted against teams we simply can't match skill wise. I assume this is because we get matched against other premades. The issue is that these premades are so much more advanced and seemingly experienced than we are and it almost always results in a lose for us (losing is not fun ><). Similarly we often get matched against people who have 500, 700, 1000+ wins. The largest amount of wins among my mates is arond 350. The lowest being around 80. I find it frustrating that our average of 300wins team gets matched against a team with an average of 600,700 or more wins. This problem isn't nearly as prevalent when we solo or duo que but...well we all like to play together . Sometimes after a long loosing streak we will break the group so we can duo/solo just so we can enjoy ourselves with a win.

It makes me sad that I cant group with mates because the matchmaking system seems unable to match our group with a similarly skilled group.

Thank you for asking us how you can improve this system though! =D
There's a few things I can think of that may help this issue. For example, we could better tune the matchmaker to prioritize other premade 5s to a degree, but if the gap in 'ratings' gets too large, to start prioritizing premade 4s, or premade 2+3s against you that may be more 'fair' than the premade 5s currently available.


25 months ago

Quote:
Originally Posted by vox1st View Post
Tl;DR?

8 and 11 are what concern me the most, 11 more so than 8


Thank you so much Lyte for taking the time and seriously addressing these issues with the playerbase.

I see you are working with zileas and one thing I picked up along the way, and I don't know where was this: Matchmaking tries to force players to win half their matches and lose half their matches, it achieves this by forcing lopsided teams in terms of skill(ELO) both in ranked and un-ranked games. Is this true? Or did I pick up on a lie?
This is false. The matchmaker does not 'force' players to win only half their matches. A matchmaker's goal is to be an accurate predictor; this means that in fair or lopsided matches, a matchmaker should be able to accurately predict your win percentages.

A lot of players have anecdotes that winning streaks are followed by losing streaks--this is simply how the games unfold and has nothing to do with the matchmaker.


25 months ago

Quote:
Originally Posted by Lonely Reaper View Post
Lyte I do have a curious question of something that happened to me.

I was gaining/losing 18-20 elo a match, at around 1300 elo with a 58% win rate i noticed i went from getting 18-20 to 11-13 (loosing as well, and this is a drastic change imo) and I proceeded to lose about 6 games in a row, a major loss streak.

I got down to about 1200 something and It felt nearly impossible to get out of but I eventually did, and it was a long painful way to get back up to 1300ish and i had to have like 4 more wins compared to losses (added onto the already 10 more wins over losses) to have the same elo i had at 1320 15 games previously, then i went right back down to the elo I am at now, and I really don't think I can keep 15+ wins over losses to keep my 1300 elo + rating that I honestly feel like I belong in while I know people with 10 wins over losses that are in higher elo than what I was before I got trolled down into lower elo and they have played the same amount of games as me roughly.
Basically, when you start Ranked, you have what is known as a high K-Factor. This means you will gain or lose ~40/40 Elo a match as the system tries to place you in an appropriate 'tier' range as quickly as possible. As you play more games, this K-Factor value drops lower until you start winning or losing about 11-13 Elo a match. We're looking into how K-Factor changes over time, and whether we need to smooth things out more.


25 months ago

Quote:
Originally Posted by ADGarner View Post
I'm not sure I really follow this line of thinking. It's not just predicting outcomes it is also "making matches". It's matching you with opponents around your skill level or so it thinks. If you start to win consistently when it thinks you should lose it has to make an adjustment until it's predictions are on track. Therefore you can experience streaks that are brought on by the matchmaker trying to adjust for your skill.

Consider the streams of people purposely tanking their elo just to climb back out. Big loss streak followed by even bigger win streak. Players make the biggest difference but the match maker most certainly has it's hand in streaks.

TLDR: I remember a time before matchmaking in online games. Ratios were rarely near 50/50 back then. Riot's matchmaker is managing ratios and causing streaks even if indirectly.
Oh, I agree. The matchmaker is always striving to find you 'fair' matches; however, for example there is nothing in the matchmaker that intentionally tries to break a winning streak. Sure, as you win several games in a row you will earn Elo and therefore be matched with tougher and tougher opponents.

However, let's say a player is winning several games in a row and goes from 1200 Elo to 1350 Elo. The matchmaker does not decide, "Hmm, time to test this player!" and pit him up against 1700 players 5 games in a row until he drops back down to 1200.


25 months ago

Quote:
Originally Posted by Los Dinosaurios View Post
firstly, as many have said, thank you so much for doing this and taking community feedback :]

secondly, though, this is a false analogy. there is always a worst TEAM in the NBA, but nobody is asking for a team to be banned from ranked, they are asking for the worst players to be banned. the worst players in the NBA are ALWAYS banned from play next season. . . they are either put into the d-league, ride the bench and get no important minutes (if they get any at all), or are cut entirely.

i DON'T think that people should be banned from ranked for poor play, elo eventually sorts them out, BUT trolls should get banned from ranked- like, much more quickly than an account is banned from the game.

if somebody is blatantly trolling or leaves a game, ruining it for 9 other people, and they get, say, 6 reports in the same game (two-thirds of the remaining players), they should be instantly and automatically banned from ranked for a week. they can still go play normals where their trolling doesn't kill anybody's visible elo and will probably still eventually get them banned, but they should have a temporary ban from ranked after an obvious incidence of trolling, and perhaps a much longer (season long?) ban if they continue. to prevent abuse, you could limit players to 2 or 3 reports per game and disable the feature in premades, but i have come across many many more leavers and trolls in ranked games than normals, which is why i am just now fighting my way above 1k elo >:[ (first 20 games (the ones that REALLY count) i had a troll or leaver on my side in 11 games, and accordingly went 8-12. imad.)


anyway, <3 u
We have talked a lot recently about whether trolling/toxic behavior in Ranked deserves harsher punishments. It turns out that players report far more often in Ranked Mode, so players actually end up in the Tribunal quicker being toxic in Ranked than in Normal.

Do we need to do even heavier punishments? It's possible, but we are still working it out and looking at the data here.


25 months ago

Quote:
Originally Posted by The Camel Rider View Post
So why do so many people have very close to 50% win rate?

Are these games we are playing literally just toss ups? Like a coin flip?

There is no way that I am so average that my wins/losses are exactly 50%. I mean, i guess I could troll my win rate down lower than 50%, but no matter how I try, theres no breaking that 50% number (higher).

Maybe im just bad?
A 50% win rate does not mean you are average, it just means the system believes your skill ratings are pretty close to where they should be. As you improve as a player your skill ratings will slowly increase, but your win percentage will rarely be higher than say 55%.

We realize that winning ~50% of your games is not a great way for seeing personal progress, and we are figuring out other ways players can see 'progression.'

The only players with high win percentages are professional players--this is because the matchmaker cannot reasonably find opponents for them that are their 'equal' and top players in the game will always be beating someone.


25 months ago

Quote:
Originally Posted by Sayath View Post
Seriously?

Could you please elaborate?

Me and my friend (Sayath & Ayer0 - NA & EUW, same names ~1600 ELO) have once again tried the great Duo-Queue (very close on NA, 100 ELO apart on EUW). We are once again utterly frustrated.

The problem:

Placement: ~55% chance to be fourth & fifth pick (regular)
Take a look at my dataset. This one is NA based. Being both last pick is truly charming. Especially if you are playing higher rated players (ELO boost for being Duo-Queue).
Furthermore, we tried some Duo on EUW. With a 100 ELO difference - we have been fourth & fifth the last three out of four games. This is anecdotal, not representative. But still.... Currently we see no point in playing together. It hinders both our efforts.


So what kind of advantage do we - a regular duo queue possess? We are not playing Ad & Support. We want to play mid & jungle. Being fourth & fifth does not help that.

Note: The xlsx / zip is outdated, but shows the data in a more meaningful way.
On average, if your Duo-Queue is far apart in Elo, you will end up first pick and last pick. On average, if your Duo-Queue is close in Elo, you will end up paired together near the middle or the bottom.

However, what I meant by 'advantage,' is that a Duo-Queue close in Elo typically wins more matches than the system expects. A Duo-Queue far in Elo typically wins fewer matches than the system expects.


25 months ago

Quote:
Originally Posted by Metaphisto View Post
Not to be the downer or a troll, Lyte, but I have an issue with this. Riot has been saying since I started playing (the week after Olaf was released) that there will be changes to things in game, but as of yet nothing has changed. Sure, you guys added the Tribunal, but it did not change anything. It was just a stunt for good PR, in my opinion.

So, instead of asking what we think should be fixed why not tell us when these things will be fixed and how they will be fixed. It is only beneficial to keep mushrooms in the dark. I can understand "trade secrets" and all that garbage, so why not just say something like: "Issue #1 will be fixed in 2 months." Instead of: "Issue #1 will be fixed soon (or what actually equates to over a year and won't really be a fix, just a PR stunt)."

Again, I'm not trying to be a troll or downer, but Riot's past practices have really jaded me.
Regarding the Tribunal, it prevents over 1 million toxic games a month--that's a meaningful number. However, given my background in Psychology, I also know there's a lot more we can do to shape player behavior more positively, and there are more things we can do to empower players and show them they are making a difference when using the Tribunal.

Regarding this matchmaking, I think I've been pretty clear throughout this process. I wanted to collect data from the community through screenshots, stories and data. Then, I was going to make a new thread discussing the issues we identified, and do some prioritization. Then, I will update the threads as we develop the solutions and have specific timelines for them.

Past practices are past practices. We are always trying to improve, so focus on the present!


25 months ago

Quote:
Originally Posted by Nostradamusss View Post
wow you didnt even posted the main problem...

#1 problem is people that queue dodge 100*time to 200elo, (they queue dodge 5second before the game start just to screw the maximum players) then pubstomp and disgust noobs, climb back to 1200elo, then queue dodge again to 200elo.
guys like this, end with completly irrealistic stats 1200win/150lose, almost never lose and make lose others players and deny their fun of the win.

every day in this forum i see people "look my stream 1700 real elo in the 300elo hell climbing back huehue"... just imagine the number of people doing that... game at 300-1200elo have 80%chance of queue dodger now and it's worse every day in 2010 there was barely 50% i would say. it need to be fixed.

queue dodge should also cost ip.
We are going to be changing this soon so that you cannot tank your Elo through queue dodging.


25 months ago

Quote:
Originally Posted by Ginga View Post
Just a quick observation from the games I've played and this comment, but...

Have you considered the possibility that something in Ranked is inherently causing the players to become more toxic rather than Ranked players themselves being toxic?

I've noticed that it's a lot easier to get allies to "shut up and play" in a Normal game than it is in Ranked. And another observation is roles in Normals and Ranked. In Ranked, no one wants to be Support, and that one person(often last pick) will always do a terrible job at supporting because his teammates forced him into a role he didn't want. In a Normal game, no one gives a **** that a person decides to support with Pantheon or whatever.

Juuust a weird little pattern I've noticed. Also noticed that Normals tends to be more enjoyable because players in Normal games tends to spice things up with team comps that you'll never find in Ranked.

If I were to venture a guess, the toxicity is 50% due to the stress of possibly losing ELO due to the perception that your allies are dragging you down, sometimes it's true, and in those cases it's not hard to see a person turn negative. The other 50% is likely due to the long game times. Players tend to get more toxic if a losing game is dragged over a long period of time. I even noticed this myself when it's possible to lose patience on someone who is repeatedly making bad plays even till the 50 minute mark. More importantly, for a lot of people, they can only play a game or two per day, so if they are having a bad game, they can't leave or else they get banned, so they are FORCED to play a really bad game for ~50 minutes and lose the game slowly and painfully if atleast two person refuses to surrender. That no doubt has a very negative effect on a lot of players who are slowly contemplating losing 10 ELO in a slow and painful game.
I agree, it's not necessarily that Ranked players themselves are more toxic (although this is part of it!). The other half is the fact that players have tangible (Elo) to win or lose in Ranked Mode, and often they feel like the reason they lose that tangible reward is out of their control.


25 months ago

Quote:
Originally Posted by Jixson View Post
I have to ask this then in regards to Elo and 50% win rates. I am currently at 49.8% win rate in ranked, and sitting below 1000 for the 4th time this season. I look at friends and various other people I play with, and they are at say 50.8% win rate and they are at 1300.

How is this possible? I know they may have gone 10-0 in their placement matches while I went 7-3(and started below the 1200 people normally start at as my season 1 Elo was SO bad), but it doesn't seem to be normal to have THAT big a difference between 1% in Wins/Losses.

Someone with a 55% win rate, sure 1300-1500 seems about right, but 49.8 and 50.8 shouldn't really be 300-500 difference.
Without going into too much detail, your win percentage is not directly linked to your Elo. It is perfectly possible to be a 50% win rate, and be 1900 Elo.

When your win rate decreases to about 50%, it just means the system feels your ratings and your current skill level are relatively close.


25 months ago

Quote:
Originally Posted by TeaSq View Post
People are saying #1 and aren't reading the entire thing.

It's talking about AFK and we have to wait until they get kicked... Not when they AFK and get a random champion. People complain about people AFK'ing and getting a random champion and then they are stuck with that on their team. This is saying otherwise, it's just saying that we have to waste a few minutes waiting on the person who never connected or who is dc'd to get booted after the selection of champions.
I've edited the main post to refer to both of those cases now. I didn't mean to exclude the cases where the player goes AFK and gets randomly assigned a Champion.


25 months ago

Quote:
Originally Posted by DoenVayne View Post
My biggest issues right now are:

Once you get low, the system is so ****ed noisy that it takes hundreds of games to level your Elo. A lot of it is K-Value related. A lot of it, also, I feel, is that you lose equal rating for both a win and a lose. I think the system needs to be weighted so that wins are more important than losses. If you got 20 elo for winning, and only lost 10 for losing, it'd be a lot less psychically damaging. I feel like there should be deprecating win/loss ratios for streaks, too. Like you always lose 10 for a loss, but as you streak higher, your elo gains slow (20-18-15-11-5). Essentially, the fact that losing a game is just as punishing as winning is rewarding spirals your elo out of control when you get a noisy streak of losses.

Also, I feel that player sentiment is ruining the game. Horrible people in matchmaking make me want to dodge regularly. They troll-ban and troll-pick. They don't pick champions that'll work with a composition, or say "I play this guy or troll all game," and stuff. I feel like, if you had the ability to thumbs-up and thumbs-down people at the end of the game, and said "If you got X amount of negative feeds, you have to wait X seconds/minutes before you can re-queue" or "You got X positive thumbs-ups. You get bonus IP or access to special skins or icons or stuff." Essentially, incentivize positive feedback and make it so people are also incentivized to punish players, as well. I think this goes a LONG way into letting people have better experiences and will remove a LOT of noise from the system as you don't get trolled nearly as much. Then I feel like I lost the game because of skill, not some ******* troll who was trash-talking and being a ***** in-game.
I agree, player behavior is one of the biggest problems in online games today. However, I am working on other projects that try to address this. Matchmaking can only affect player behaviors in a somewhat indirect way.


25 months ago

Quote:
Originally Posted by Xedrox View Post
what if a troll say he's going to afk in game during matchmaking and we cant do anything cuz if we dodge we lose 10 elo
If a troll says he's going to AFK in the game during matchmaking, then actually AFKs in the game, he is going to be punished by LeaverBuster.

Separately from this initiative, we are going to improve the feedback you receive for using systems like the Tribunal to better show you the impact of your actions.


25 months ago

Quote:
Originally Posted by GrantLewis View Post
Thank you for this immensely informative post into the matchmaking design process!

What do you think about the status of the normal elo ratings, which have not been reset, in relation to those of the ranked ladder?
Could you be more specific here? So far, we've seen a pretty decent correlation between your hidden Normal Elo and your Ranked Elo, if you play Ranked.


25 months ago

Quote:
Originally Posted by solidfake View Post
....normal draft pick order....
I've posted and followed up before in the other thread. I simply do not have anything new to update. Spamming this thread or other threads does not actually speed up the process In my last update, I mentioned that the bug is harder to fix than expected.

We also wanted to add some enhancements to the pick order problem in draft in general, so doing the bug fix + enhancements is taking longer than expected and we are not happy with it but also have nothing new to update yet. The normal draft pick order bug was also not in the scope of my matchmaking, as it is a current bug that needs to be fixed.

I highly encourage all of the posters from that thread to start contributing to this specific discussion, because the approach some of you have taken has been inappropriate.


25 months ago

Quote:
Originally Posted by Imclueless View Post
Would it be possible to make one effect the other in some way, to try and cement this relationship? It seems to me that keeping the two completely separate is the core cause behind some of the issues you've highlighted. High-ranked players with a low Normal Elo, and the frustration many players face trying to get their Ranked Elo up to their Normal Elo. I'm personally demotivated to play Ranked because: I already have (fairly) balanced matches playing Normal games (though it appears I get paired with worse players on average on the weekends... it's kinda strange). I thought it would be a quick and easy matter to get to that point in Ranked games. The experience was rather far from it. The Ranked games I did play were extremely frustrating. There were players clearly better than me, and there was a Ryze who rushed a Deathcap (saying Ryze is his best Champion). I realize that, with enough games, I could get my Ranked Elo to where I'll get balanced games like I do in Normals. But, at the moment, it doesn't seem like there's a point to it, because I already have them in Normal games. Why go through that headache when I already have balanced games available to me? It seems like there should be some sort of built-in correlation between your Ranked and Normal Elo's. It would get people playing in their proper skill bracket much faster, it seems to me.
We have old data on the relationships between Ranked and Normal Modes, and we are collecting some new data that will help us build an interaction between the two modes so that Elos are better normalized between the two modes.


25 months ago

Quote:
Originally Posted by TA Cake View Post
i have a problem. No necessarily towards matchmaking but the tribunal. Most people just look at people's matches and determine whether to ban them or not right on the spot looking at their kdr...people need to actually read what has been said in chat..
Actually, data suggests that most players read the chat and use that to decide their verdicts, instead of simply looking at the Report Reason, or game stats.


25 months ago

Quote:
Originally Posted by Btooom View Post
i noticed but that respone was months ago, i felt like they dropped the topic
I have not been a Rioter that long :] But now that I am here, I am working as fast as possible without losing my Santa Hat!!!


25 months ago

Quote:
Originally Posted by Imclueless View Post
Is this something we'll be privy to, or will it remain behind-the-scenes? I'd like to know how that system worked (at least in vague terms) if/when it was implemented. Also, would this relationship work one way or both ways? I realize you haven't actually come up with it yet, but I'm interested to know what your initial thoughts are.
This will most likely remain behind-the-scenes, but will work both ways.


25 months ago

Quote:
Originally Posted by Banuvan View Post
As much as I would really love to believe you Lyte there is two years of inaction on these issues. These issues have been known and posted on the forums since the game came out.
Well, I'd like to shed a bit more light to give you more context. For 2012, Tryndamere and Ryze made several statements about player behaviors being a core goal for Riot, and improving the League of Legends experience in general as a second core goal.

But did you know I only started in February? I was in contact with Riot in 2011, but wanted to finish my PhD and was entertaining offers from Ivy League institutions. I didn't decide to come to Riot until 2012. Others like Yegg (Senior Design), davin (UX Researcher), BamDragon (Director of Animation) and many others are all key players that started in the past few months. Looking a few years back, Riot wanted to do a lot of awesome things but they had limited resources. Riot is fixing the resource problem and not looking back.

So is it surprising that we are scrutinizing and fixing seemingly old problems "all of a sudden one day?" No! We all came to Riot to make League of Legends the best possible experience for our players. The amount of talent that has joined Riot even in the last 3 months is astounding, and 2012 is going to be a great year for our players.

You can blame me, you can blame our leadership or even blame Riot in general for not 'fixing' matchmaking sooner... or we can focus on the fact that I am making the effort now to connect with our players and you will get a better matchmaking experience because of it.


25 months ago

Quote:
Originally Posted by Lyte View Post
Well, I'd like to shed a bit more light to give you more context. For 2012, Tryndamere and Ryze made several statements about player behaviors being a core goal for Riot, and improving the League of Legends experience in general as a second core goal.

But did you know I only started in February? I was in contact with Riot in 2011, but wanted to finish my PhD and was entertaining offers from Ivy League institutions. I didn't decide to come to Riot until 2012. Others like Yegg (Senior Design), davin (UX Researcher), BamDragon (Director of Animation) and many others are all key players that started in the past few months. Looking a few years back, Riot wanted to do a lot of awesome things but they had limited resources. Riot is fixing the resource problem and not looking back.

So is it surprising that we are scrutinizing and fixing seemingly old problems "all of a sudden one day?" No! We all came to Riot to make League of Legends the best possible experience for our players. The amount of talent that has joined Riot even in the last 3 months is astounding, and 2012 is going to be a great year for our players.
I can with confidence say that Lyte, myself, and all the other folks involved in this are completely committed to making awesome changes--and working very hard to make those changes a reality. We are really, really excited to have the opportunity to shape matchmaking (and all the things around it) into something that makes playing the game even more rad.


25 months ago

Quote:
Originally Posted by XziisT View Post
Glad to see Riot listening to their customers. I am sure that in these 50 pages there have been many detailed writings about a player's personal thoughts on what they want changed and what is trying to be implemented; however, it seems like I speak for a lot of players when I say..

The ten issues you brought up in your poll is not the problem. The true problem is not in matchmaking itself. The bigger problem is in-game. The bigger problem is the player that leaves his/her team to go away from his/her computer for the remainder/part of the game. Sure, the issues you have mentioned are troublesome, but I would much rather wait through matchmaking when someone is AFK only to start over rather than play a 4v5 matchup. Players on a daily basis aren't complaining about AFK players in champion select mode or free to play champions in ranked mode, they are asking you to figure out a new system for in-game AFK players (trolls, feeders, etc., play a part in here, too, but does not cause as much pain as AFKers). In all honesty, a new system to rid the game of AFKers would take quite some time to develop if it can even be developed at all. I think that players are more concerned with in-game AFK players rather than the rest of the issues you have mentioned. It would be interesting to get another poll on this point because I gaurantee you it will quadruple the number of people who voted in this poll.

Thumbs up if you agree.
Although I agree AFKing in-game is an issue, this is probably not the thread for it. I am separately working on better AFK detection and ways to combat it, but this thread needs to focus on matchmaking which is the process before the game begins.


25 months ago

Quote:
Originally Posted by DangoxDango View Post
Also, what does it mean to "Tank ELO"
This is when players intentionally queue dodge repeatedly in Ranked Mode to drop their Elo down to a low number such as 300. Then, they stomp low Elo Ranked players over and over until their rating rises back up... and repeat the process by queue dodging again.


25 months ago

Quote:
Originally Posted by Futonlols View Post
Why is there a poll? All of these issues are important, just start from the top and work your way down. Lazy riot.
A poll is one way the community can have their voice heard.

Who said we would only fix one issue? 0_0


25 months ago

Quote:
Originally Posted by Rapier Napier View Post
Hi Lyte,

In your post in response to the prioritizing of premades in draft normals, you refer to the inappropriate approach players (such as myself) took in the "@Riot, draft mode is prioritizing premades" thread pointing out this issue.

I was just curious as to what types of behaviours you felt were inappropriate, or if the thread itself was inappropriate, as I personally feel the issue itself should be priority #1 for Riot (though this is admittedly biased due to draft normals being the only type of game mode I have fun playing). I don't think it was the goal of anyone to act inappropriately, minus the obvious forum trolls.

I think most people understand and realize that this is surely a difficult problem to fix, or else the staff at Riot would have already fixed it. I think what people are annoyed about is how drastically it effects gameplay for them if they mostly play solo queue in draft matchmaking, and the fact that it is not priority number 1, even though you state that the majority of players play normals.

Anyways, I know this issue was purposely not covered in this thread. I was just hoping you could clarify on the inappropriate behaviour, as there is a subtle line between pressuring Riot to try and make a change for the better vs being rude and arrogant while shouting for change.
The thread is fine, and bringing attention to an issue that greatly impacts your everyday experience is fine. There are often common misunderstandings though: some players thought that planning Remember Urf Day was somehow impacting the progress of normal draft pick orders when two entirely different departments at Riot are involved.

What was not fine was some posters constantly cursing in their posts to "get our attention," sending me personal e-mails with cursing, threats of leaving the game, and personal attacks. I think it is pretty obvious this approach is inappropriate


25 months ago

Quote:
Originally Posted by Dendrik View Post
Language barrier is another issue.
Language barriers are definitely an issue. I have thought a lot about matchmaking and trying to match players who have the same language in their League of Legends client together, but some pools of languages will not have a healthy number of players for good matchmaking.

We'll figure it out though, there are some other ways to reduce some of the problems with language mismatches.


25 months ago

Quote:
Originally Posted by Snafzg View Post
Hi Lyte, I hope you can grab a coffee and read over my thoughts. I really appreciate that Riot is trying to improve matchmaking, which I think will go a long way towards improving the community spirit. Sorry for the wall of text, but I am very passionate about these issues!

1) AFKs in Champion Select Lobby

Players who queue up for a match and then AFK force players to wait through the timers until the lobby boots the AFK players out. Alternatively, some players AFK and get assigned a Random Champion, which is also a negative experience for many players. In Ranked Games, Captains tend to see what the bans/picks are like and be able to react accordingly in the next lobby as the players will generally be the same.

AFK and disconnected players should be given a dodge with all its associated penalties from the champion select lobby, not assigned a random champion and allowed to troll their team mates (in normal or ranked). Once an AFK or disconnected player is booted from the lobby, the remaining players should not be automatically thrown into the queue either. Instead, give them a popup message box asking if they’d like to re-queue (yes or no). This solves the second issue you mentioned, and it also gives players an opportunity to avoid re-queuing with a potentially toxic person after someone purposefully dodged.
We have talked about similar implementations for solving #1.

Quote:
2) Duo-Queue Elo Disparities in Ranked

When players play as a Duo-Queue in Ranked, they generally have an advantage if their Elos are fairly close. However, when players play as a Duo-Queue in Ranked and their Elos are far apart, they generally are at a disadvantage.

You should only let people duo queue ranked if they’re within a certain ELO range of each other (e.g., ~100 ELO). Not only are the two duos at a disadvantage when there’s a large ELO spread, but they’ve put three other allies at a disadvantage too, which isn’t fair. This will also eliminate situations like Chauster carrying Missy to gold rank in season 1 (who actually had a negative win-loss score).
For the initial go, I want to try to not restrict Duo-Queues in Ranked Mode. There are some creative solutions we can try in the matchmaker first; however, if those do not pan out, I may look at restrictions for extreme differences in Duo-Queue Elo and slowly tweak those restrictions until Duo-Queues are 'just right' in Ranked Mode.

Quote:
3) Skilled Ranked Players in Normal Modes

For a very small subset of highly skilled Ranked Players, their Ranked Elo is vastly higher than their Normal Elo. When these skilled Ranked Players play a Normal Game they generally play against opponents that are much lower in skill, resulting in lopsided matches.

The best fix would be to tie someone’s normal ELO to their ranked ELO (at least average them to reduce the spread). Simply put, a 1900 ranked player should not be playing 1200 normal games. Conversely, a 900 normal player should start their ranked career (i.e., provisional matches) at 1200. More on this in #5 and #8 below.
We agree, we'll be doing some normalizing between Normal and Ranked Modes.

Quote:
4) Premade Matching

Currently, the matchmaking system tries to match Premade 5s with Premade 5s; however, we may consider prioritized matching for other Premade types. For example, prioritizing Premade 4s to match with Premade 4s, and finding a solo queue to fill out each team.

I can’t believe you guys don’t already do this… Please do so. I can’t think of any reason why you wouldn’t want to.
We already do for Premade 5s, but we will probably expand the preferential matching to other premade types.


Quote:
5) Transitioning from Normal to Ranked Mode

Currently, we do not do a great job of educating players on the transition from Normal to Ranked Modes. What would players like to see here? What should the expectations of Ranked be? For example, Ranked could be "Bring your best, every single game!"

This is easily my biggest gripe about matchmaking. There must be some sort of education process or barrier to entry for ranked. I’m so sick of fresh-faced/green level 30s jumping into ranked and polluting the matches. It’s gotten worse since you guys introduced Co-op vs. AI where someone could quite feasibly hit level 30 without ever playing vs. a real player. On top of this, allowing someone with a <1000 ELO in normal mode to start their provisional ranked matches at 1200 causes a huge problem in the 1000-1400 ELO range because the skill disparity between players in any given ranked match is huge. I realize you guys have a K-factor that supposed to quickly shuttle people towards their true ELO but that’s not fool-proof because of the L-factor (LUCK). It’s quite possible for a good or decent player to get unlucky and go on a losing streak in his early matches and it’s equally possible that a bad player will get lucky. This makes the 1000-1400 ELO range one giant cluster**** imho. The only people K-factor really helps is the top 0.5% of ranked players who want to rank up yet another smurf because they’re so good that the L-factor isn’t as big of an issue.

So now that I’ve got my rant out of the way, here are some suggestions!

Education: Produce a video that appears prominently in your client (preferably it will pop up and take over your entire screen when you try to queue up for your very first ranked match). This video would basically be a boot camp for newbies wanting to play ranked. It would bring them up to speed on expected behaviour, general team comps and strategies, and most importantly, it would reinforce the notion that ranked games are for people who want to take the game seriously and try hard to win.

Barrier to entry: You shouldn’t be allowed to queue for ranked until you’ve watched the video and signed off on a code of conduct. You also shouldn’t be allowed to queue for ranked until you’ve had at least 200 normal wins. People who have only/mostly played against bots are not ready for ranked no matter how nice and inclusive Riot wants to be. You guys need to get rid of the “everyone get’s a fresh start” in ranked mentality because starting everyone at 1200 ELO no matter what their normal ELO is has a toxic effect on the ranked community, especially in the 1000-1400 ELO range. Whatever your normal ELO is, that’s where you start your provisional matches. This is how the system should have been designed from the start. I realize it’s going to be a huge implementation issue, but you’re still a young company and the sooner you fix this, the stronger and more accurate your ranked ladder will be. You guys did an ELO reset at the start of S2 and it seems to have worked out well. The best time to reinvent your system would be at the start of S3. As a side note, this will allow you to tune down your K-factor or possibly get rid of it altogether.
We definitely could do a better job 'introducing' players to Ranked Mode, and a short video PSA might help, but wouldn't be the full solution.

Previously when we talked about Normal Elo affecting starting Ranked Elo, a lot of players voiced complaints that they play the different modes for different reasons and different intentions. So players use Normal Mode to test champions, try out fun builds, or mess around and they did not want this to affect their starting Ranked Elo.

I agree we could raise the requirements to play Ranked, but 200 Normal Wins might be excessive. Right now, it takes about 200 Wins to get Level 30, so I assume you mean 200 Wins after Level 30? Basically, that is about ~100 hours of games which is longer than it takes to hit "cap" in most other games these days!

Quote:
6) Free to Play Champions in Ranked Mode

Related to the transition from Normal to Ranked Mode, some players feel that you should only be able to play Champions you own in Ranked Mode. The argument is that a lot of players who use Free to Play Champions are inexperienced with those champions, adding a lot of noise in the matchmaking system.

You shouldn’t be allowed to play champions you don’t own in ranked because you should bring your A-game to ranked matches, not your troll-game. If you don’t own 16 champions (6 bans + 10 picks), how serious are you about LoL anyway? Not serious enough to be playing ranked imho!

7) Random Champions in Ranked Mode

Some players feel that you should not be able to "Random" a Champion in Ranked Play, and that this feature is used to intentionally troll their team in some use cases.

This is a total troll move in ranked plain and simple. Disallow it. It’s still pretty a pretty troll move in normal games, but normal games aren’t serious business imho. Also, AFK players shouldn’t be assigned a random champion in any game mode. They should be kicked. See #1 above.
Agree, I think part of the problem is educating players that Ranked is "bring your best, every single game."


Quote:
9) Duo Queue Prevalence in Ranked

Currently, it is possible to get two pairs of Duo-Queue Players per team in Ranked Mode, such that you have a team composed of Duo-Queue, Duo-Queue, and a Solo-Queue. Some data suggests that we should restrict the number of Duo-Queues per team to 1.

Does this exacerbate the issues described in #2 above? If so, sure, do it, especially if solo-queue players consistently land in the last-pick slot due to your draft mode bug (or is that bug restricted to normal draft?).
The pick order bug is only in Normal Draft Mode.

Quote:
10) Level Disparities

Some players feel that as a low level, they should never be matched against Level 30s because they feel the match is unfair due to Level 30 rune/mastery advantages. Although these players are highly skilled low levels, they would prefer a different solution than facing an opponent with simply 'stronger stats.'

This is a low priority for me, and here’s why. I let a friend play my level 30 account that’s loaded up with full runes and masteries and I logged into his level 20 account and stripped myself of all runes and masteries. We then played three 1v1 matches in mid lane: first-kill/tower wins. I wrecked him every single time because I have much more experience than he does (almost +1000 more games played). Runes and masteries are meaningless if there’s a skill disparity. On top of this, I almost exclusively grouped with level 30s (duo queue up to 5-man pre-made) as I ranked my account from levels 1-30 and I rarely felt at a statistical disadvantage. Sure, I came up against people that wrecked my face all the time, but that’s because they were BETTER than me.

One problem I have with matchmaking is WIN disparity. I’ve gone up against teams that have over 3000 more combined wins than our team’s combined wins. How is that fair? Are you factoring in their hidden ELO? I’m not sure, because we got pretty owned… Then again, I play with and against people with 1000 more wins than me all the time and often feel like they’re pretty bad players, so maybe wins don’t matter either...
I agree, if there's a huge skill disparity then arbitrary stat boosts do not matter. But remember, the matchmaker is trying to find you fair matches, so the skill disparities are not large and that is why low level players feel the games are unfair (even if they win!). They are playing relatively similar skilled players but are at a stat disadvantage.

Quote:
11) Team Margin of Victory

Some players argue that in epic, close matches, teams should not gain or lose the exact same Elo as a lopsided match. What are good metrics for "team margin of victory" that are focused on promoting team play and not individual performances?

Personally, I don’t feel like this system is necessary, but here’s a proposal for you to consider that values team play over individual performance:

Margin of victory could be the difference in team scores based on number of factors such as team kills/assists + buildings destroyed + dragons + barons. So let’s say kills are worth 2, assists are worth 1, towers are worth 3, inhibs are worth 4, dragons are worth 3, and barons are worth 5. The closer the teams are to each other, the less ELO the losing team loses! So let’s say your max ELO loss is -12, maybe you could reduce it to only -9 in a very close match where the margin was near or at 0. A margin score of <10 or less could give a 3 ELO boost to the losing team (-9 instead of -12 ELO), a score of 10-19 could give a 2 ELO boost (-10), 20-29 could give a 1 ELO boost (-11), and 30+ gives you nothing (-12).

Here is a practical example showing a pretty close match:

Team A (Winners): 29 kills (58 pts), 50 assists (50 pts), 4 towers (12 pts), 1 inhib (4 pts), 1 dragon (3 pts), and 0 barons (0 pts) = 127 points

Team B (Losers): 20 kills (40 pts), 50 assists (50 pts), 3 towers (9 pts), 0 inhibs (0 pts), 3 dragons (9 pts), 1 baron (5 pts) = 113 points

Margin of victory = 14 points (+2 ELO to losing team)

In the game above, Team A was slightly ahead in kills but behind in map objectives when Team B tried to sneak baron. Team B got baron but Team A swept in for an advantageous team fight that they won 4 kills to 0! Team A then pushed right up the middle and dropped the enemy nexus mere seconds before Team B respawned. EPICSAUCE!
I agree, it has to be a combination of metrics that all promotes team play, and the overall adjustments have to be small (a few Elo points in either direction). We do not want to promote stall tactics, we do not want promote prolonged 'suffering' or poor experiences. We want teams to play their best, every single game, and our metrics have to support that mentality--heck, even promote that mentality.

Quote:
I HOPE YOU MADE IT THIS FAR. IF YOU DID, THANKS FOR READING!
I drink tea, not coffee


25 months ago

Quote:
Originally Posted by Omortag View Post
Thanks for taking the time effort not only to fix these problems, but to keep the community up to date through this thread!

Personally, I feel like a lot of people rage in ranked because they feel like others are keeping them from their 'true elo,' meaning that they want to climb and won't be truly satisfied until they reach plat or 2.2k plus or whatever. That's because all the attention in ranked is focused on the top 0.001% of players who play. What I would really be interested in seeing is more elo specific tournaments and prizes. For example, have Riot host a 1 v 1 tournament for bronze players, or a 5 v 5 tournament for silver players. Award prizes every month for the best Silver rated Anivia (or any other champion) player. In short, give incentives for people to be happy at the elo where they are, instead of leaving it as a bloody fight to the top.
I like the idea of mini-tournaments for all demographics. A Weekly Bronze Showdown would be fun.


25 months ago

I've been interested in the matchmaking system ever since Zileas started talking about what Riot was doing back in Beta. Couple of thoughts on the discussion so far,

1) AFKs in Champion Select Lobby
This came up in the discussion already, but in regards to dodging AFK’s vs. randoming, I thought the system was intended to always dodge someone unless there’s a bug? If there is some sort of exception, doesn’t seem like it should be there.

Given what’s required for this to happen, (manually queue up the game and be AFK 1-2 minutes later.) Unless you’re at high ELO, the level of derp that causes champ select AFKs probably deserves a higher punishment.

In terms of the Dodge to -ELO point that's been floating around. I'd be worried what changing that system would do. The reason it was put into place in the beginning was because the level of queue dodging was so high and it disrupted matchmaker mechanics. Basically reducing the incentive not to dodge because a few players lower their ELO for fun doesn't seem like the best idea. A better idea might be to permenantly remove Gold and Platinum badges for players who do this


2) Duo-Queue Elo Disparities in Ranked
I completely understand where you’re coming from not wanting to restrict friends playing with friends, but this doesn’t quite mesh with focusing ranked on “bring your best, every single game.” Going back to your basketball analogy, a middle schooler might be great friends with Derrick Rose, but that doesn’t mean they’ll let him play in the NBA.


3) Skilled Ranked Players in Normal Modes
Normalizing sounds like a win 


4) Premade Matching
Especially given the data you’re mentioning that talks about premades having either benefits or disparities based on their ELO shift, it seems like there might need to be a multi-part piece to this. Both matching pre-mades to pre-mades and changing team average ELO bonuses for premades depending on the ELO gap between those players.


5) Transitioning from Normal to Ranked Mode

Quote:
Previously when we talked about Normal Elo affecting starting Ranked Elo, a lot of players voiced complaints that they play the different modes for different reasons and different intentions. So players use Normal Mode to test champions, try out fun builds, or mess around and they did not want this to affect their starting Ranked Elo.
I think this is the right idea just the wrong way of looking at how to implement this system. You don't need to permanently link Normal and Ranked ELO here. If you think about a player getting to level 30 through normal games, their ELO is more a picture of their skill level and evolution as a player rather than individual games where they tried something new. EVEN if they screwed around in some percentage of their games, their normal ELO whether it’s 400 or 2,000 is still going to be a way better indicator of where they should be initially seeded in ranked then just lumping everyone into a 1200 group.

For any player already in ranked, ELO’s are already established, new season ELOs are based off of soft resets, so any games they played in Normal wouldn’t affect their initial ELO seed.

Given how separate Ranked play is from Normal play, it would probably also be worthwhile to have a newbie island repeated at this level. It’s not a great analogy, but harkening back to DCUO, the game is basically level 1-30 are training, and then the disparity between new level 30’s and high level 30’s is night and day. A system that keeps these “new” players together and lets them fight it out would help to better adjust initial skill levels without throwing them in with 300+ game winners, especially if you’re not using normal game ELO seeding.


6) Free to Play Champions in Ranked Mode
Related to #5, I think eliminating these champions would be a great addition to ranked play. It encourages players not to try unknown, ELO skewing champions in “bring your best” games and also guarantees players will have at least played enough to earn the number of champions required to participate in Ranked.


7) Random Champions in Ranked Mode
Same as 6, education and options are two different things. It’s great to be able to educate someone on “this is what you should do for ranked.” But sports have rules and these are easy ways to visually and literally enforce more skilled game play. There’s no point in saying “bring your A game every time - don’t random, and oh by the way, we’re still going to let you random.”


8) Provisional Matches in Ranked
Addressed in #5, let new players play on Ranked Newbie Island. Normal ELO seeding would also help address this, starting closer to where you should be and moving around will put you closer to the right place no matter how your initial games worked out.

I suppose an alternative here could be to start everyone at 1200 ELO but give players different modified K-factors for wins and losses based on their Normal ELO.


9) Duo Queue Prevalence in Ranked
If the data says so, it might be worth a shot, I’d be more worried about what that would do to match finding times.


10) Level Disparities
Given the data we’ve seen all over the place, every one of these complaints has seemed more like “OMG I lost 1 game, time to shift blame” rather than any substantive issue with matchmaker. The system already gives slight ELO adjustments based on level, it seems like that works just fine. THIS is more of an education issue. We still see a lot of people who don’t understand MM in terms of win/loss ratio, level and ELO.


11) Team Margin of Victory
Dominion is easy, ending points :P

For Summoner’s Rift, I think the biggest single identifying factor is time. While certainly a team could “play” with an enemy just to farm out an “epic win” there’s a couple things that make this ok:
  • Epic wins tend to come from the fact that “winning” teams don’t finish fast enough and give their opponents too much time to farm and turn the game around for a comeback.
  • Even if this happens, both teams get an “epic game” bonus making it more worthwhile.


25 months ago

Quote:
Originally Posted by NihilisticGuy View Post
Hypothetically, let's say I'm doing this, dodging to 300 elo to stomp.

OK, you change it. So now if I'm doing this, and want to continue, I just pick Eve repeatedly and try to play roaming. It's fun, it's not against any rules anywhere, and I'll lose more games than I win simply because I'm not a good roaming Eve. I'm suddenly tanking not only my own elo, but my team's as well. How exactly is this better? Note that if I was not forced to actually lose the game to lose elo, my team would be spared the experience of playing with me when I'm not trying to win.

And if that's not enough there's always revive/smite jungling. Again, breaks no rules in the summoner's code, and is a lot of fun. People who want to tank ELO will do so. You can't stop it. As is, their teammates don't have to feel the frustration of playing with someone who is not trying to win.

People who want to do this will continue to do it. They'll just have to lose. There's plenty of ways to lose without breaking the summoner's code. However, it penalizes the team, because one person on the team is not trying their best. How is this an improvement over the current system?

TL;DR
People who want to tank their Elo will still do so. They'll just play subpar and lose games. They will intentionally cost their teammates elo and the chance of a fun experience, solely because they are unable to cost themselves elo in any other way. And it's impossible to prove they're breaking the summoner's code unless they flat out admit they're playing poorly intentionally in an attempt to fall elo.
Right now, it takes players a few hours to drop their Elo to 300 and begin stomping newbies. If they win 100% of their games, they ruin about 100 games before they reach the 1400-1500 range again. Then, they drop their Elo back to 300 and start over.

If players want to drop their Elo to 300 on purpose in the future, they may troll games and intentionally lose; however, each drop in Elo takes a much longer time and they have to actually try their best and play to win every single game (but still end up losing). If they do not try their best and 'hide their intentions' every game, they may be vulnerable to the Tribunal. It will basically be much, much more effort to do the same Elo dropping in the future, and the # of actual games affected by "Elo tankers" will in fact be lower.


25 months ago

Quote:
Originally Posted by nomadd View Post
Lyte, could you shed some light on how win ratio plays into the current match making system?

I know the system attempts to match you against players of a similar rating, however, my understanding is that the system also attempts to level your win percentage to near 50%.

Why is it, that at 1700 rating, I can have a better win percentage than a player in the 2000 range?

Unfortunate for me to be just beneath the wall of text
Win percentages actually don't really play a role in your matchmaking and the matchmaking does not try to force you into a 50% win/loss ratio. Think about it this way, the matchmaker is always trying to find you fair matches. When you eventually start facing similar skill opponents, your win percentage slowly drops to 50%.

You could have a 1700 rating with a higher win% for several reasons. One, you are still moving up, and will face harder and harder opponents as you move towards 2000. Two, you started lower than you should have from Provisional Matches, and are quickly clawing your way back up.


25 months ago

Quote:
Originally Posted by MouthØfMadness View Post
Lyte,
As you can see, Lyte, there is a detailed history of common misunderstandings on this issue. Perhaps you can excuse my frustration on this topic, as a Ph.D with a background in psychology, you can certainly understand how it originated.
We are always trying to improve our communication channels, and we have learned from our mistakes communicating about this particular issue. My first response on this issue was February 18th. That was basically around the time I started at Riot, so that can give you a sense of how quickly I picked it up considering I was still getting settled in to a new environment.

This doesn't excuse how this issue was handled, but I can assure you communication between Rioters and players is constantly being improved.


25 months ago

Quote:
Originally Posted by Lonely Reaper View Post
lyte, do you personally play ranked? if so, why, if not, why? What makes you decide your choice, with regard to your job actually being to work on the game itself.
I play Ranked quite often. I used to be a competitive gamer, even played in some WCG regionals/nationals in the early 2000s. In Everquest/WoW/other MMOs, I used to be in server- and world- leading guilds, so I guess being competitive has always been a part of me.

I started League of Legends Fall of 2011, I have 3 accounts in the 1500ish range (one account I started from 800 Elo to get some personal experiences going up), and 1 account in the 1800ish range.

I just love playing games in general, and every spare moment I have, I'm playing video games. I also can clearly see the frustrations our players face by simply putting myself in all sorts of shoes and wearing all sorts of hats. Even though I work on LoL, I still love playing it. Around the offices, Rioters are constantly playing League of Legends until 2-3 AM in our PC Cafe even though other games are available.


25 months ago

Quote:
Originally Posted by EFG View Post
People who complain about free champs in ranked are missing half the problem. Its not just free champs, its a summoner using ANY champ that they haven't played before. Unlocking something does not improve your skill with it and if you do do something about free champs in ranked the people who complained about that will just start complaining about this instead.
I agree, there's no 'perfect' solution here that addresses every user. If a player buys a Champion, there's a pretty decent chance they play that Champion a few times; however, some players buy a Champion just for trading.

If we force players to play a Champion X times before playing Ranked, some players will have to play 200+ games just to 'unlock' their Champions for Ranked. Many high level players do not need to play Caitlyn 5x, Ashe 5x, Graves 5x, and Vayne 5x to be able to compete with them.

The list of potential issues goes on and on But, some solutions address 80% of the problem and are easy to implement. Nothing here will solve 100% of the problem.


25 months ago

Quote:
Originally Posted by TheAncalagon View Post
A question - why aren't we seeing easy fixes being implemented now? Why weren't they implemented last patch?

For example: The random button in ranked games. You agree, and we agree, that it should be removed. Why hasn't it been removed?

As a computer science major, I know roughly the amount of work that it would take to remove the button. Having no knowledge of how the game's source code is set up, I could probably do it in less than two hours. Which means that your experienced programmers at Riot could do it in less than ten minutes, since they have much more experience, and much more knowledge of the source code.

Why isn't the fix to this released tomorrow? Why wasn't it released two weeks ago - since you all have known it's been an issue for so long?
A few changes have actually occurred to matchmaking already since I've been here. Zileas and I have been working pretty hard at it, and most of the changes have been 'behind the scenes.' These changes ultimately provided more value to the player, so we did them first.

Removing the Random Button in Ranked is obvious and relatively simple, but data suggests the actual value to the players is pretty small as well because very few players actually use the button. We will do it eventually, but we were working on some big value items first that unfortunately were 'hidden' to players.


25 months ago

Quote:
Originally Posted by Sypher76 View Post
Lyte, hopefully you'll forgive me if you've already answered this but can you tell us some about your background and reasons for coming to Riot/being hired? It sounds like Riot recruited you almost specifically for this job (dealing with matchmaking) - is that the case?
I have a BA (Psychology), MS (Cog. Neuroscience), and a PhD (Cog. Neuroscience). My education was funded by the Howard Hughes Medical Institute. During my PhD, I also worked at a top-notch game developer in Bellevue, WA. Post-PhD, I was deciding between a few Ivy League institutions and a few game developers.

It turns out my close friends Geeves and kitae worked at Riot and I visited Riot around the end of summer in 2011 to see them. I was very impressed by Zileas (one of the smartest designers I have ever met), and so impressed with the level of talent converging at Riot in 2011, that I knew it was the best place to be. I planned to go to Harvard first and come back to Riot at the end of 2012 or Summer '13, but Zileas told me to come in Feb '12 and I said OK

Here's my team's core goal at Riot: Ensure League of Legends has the most sportsman-like community in online games. This means matchmaking, Tribunal, and anything else you can think of related to player behaviors. In the end, I don't care what feature/project I work on, as long as it results in a better experience for the players.


25 months ago

Quote:
Originally Posted by Alyssa Riel View Post
I apologize in advance for turning this into a wall of text. I've been poking at these problems on the forums in the past and never really gotten any feedback. So here is what i would like to comment and add after a year and a half of play.
Thanks for the thoughtful post Aly. I think we agree on many of the issues, and yes, hold me accountable for change The only thing I want to add is that there is a team dedicated to making new champions but that doesn't mean Riot as a whole is focused on just profits through new champions.

It's up to the other teams at Riot to improve your everyday experiences and make League of Legends even more awesome.


25 months ago

Quote:
Originally Posted by Sypher76 View Post
I think the idea is more that there is an absolute minimum Elo of zero. *Everyone* gets started at zero, and the better a player is, the higher they can lift themselves off the elo "floor". There'll be a lot of noise at first, of course, but eventually unskilled players will be left behind, and players will play only people who have carried themselves to the same level.

More specifically, there won't be any noise band around 1200 where starting ranked people confuse things.
IF we put all players starting Ranked at 0 Elo, with a floor of 0, what stops troll players from starting Ranked Play and intentionally trolling every game? They know they can't go below an Elo of 0, so it is a free troll card until the Tribunal punishes them.


25 months ago

Quote:
Originally Posted by TheAncalagon View Post
Thanks for the response.

I suppose it isn't that huge of an issue after all - just one that kinda lurks in the background. If there are higher priorities, then it's good to hear that there is progress being made!

Another question if you will. When these changes are implemented, will we be notified of them in the patch notes? Like, "Changes to PVP.net - Queue dodging in ranked games no longer drops your elo. It now does ...."

Will we be notified of these changes? I understand that many of the implementation details shouldn't be released - in order to prevent abuse of the features. But will we know when major, more influential updates to the matchmaking system are released? For example - the queue-dodging issue?
Yes, changes will most likely be in patch notes (and in this thread) unless they are 'behind the scenes' and is something we don't want to reveal to the public.


25 months ago

Quote:
Originally Posted by FumIsChum View Post
Well if matchmaking is not causeing the win/lose streaks what is?Every big winning streak I have had was followed by a losing streak. Think I had a 13 game streak one time followed by around the same amount of loses. This happened to me multiple times and from what I hear many other people.

Edit:Also what are the chances of hitting a 15 game streak directly followed by a 15 game lose streak
The chances of a 15-game winning streak followed by a 15-game losing streak is extremely low.

The matchmaking is definitely not causing win/lose streaks, there is nothing in the code that breaks your winning streak intentionally. The human brain in general loves seeing patterns where there are none. For example, flip a coin 1000 times and you will see dozens and dozens of 'streaks.'

When some music apps first created 'random' shuffles for music lists, users often complained that the random generator was broken--they heard songs repeat more than they thought they should. Actually, the random generators were random. However, to appease these customers, many music apps start making pseudo- (fake!) random generators that would rarely repeat on purpose, to give customers the illusion the random generation was actually 'random.'


24 months ago

Quote:
Originally Posted by mojodarkale View Post
I actually don't think any of these are the main concern, Lyte. It's player behavior. That is by far the #1 reason for terrible ranked games. Your list would be a nice touch-up to the current system, but I think you Rioters should spend a lot of resources on finding ways to make it so that 90% of games don't end up with at least one person threatening to report someone. I don't actually think I'll get banned, but the lingering thought that I might, just because people threaten to report me for no apparent reason, is uncomfortable.

The environment in a typical ranked game is just disgusting (although it's getting slightly better at my current Elo); it usually begins at champ select, where people don't get to play the champ they want to or they're forced into a role they can't play well. I don't know that there necessarily needs to be harsher punishments in ranked, but there certainly has to be better and more tangible rewards for good behavior/play. It almost feels as if there's no punishment system considering how frequently games just turn into rage-/troll-fests. Just today I had someone intentionally announce that they'd AFK at around the 10-minute mark because another person on our team started arguing with him whether or not LoL used an Elo system. Yes, that actually happened, you can check my second to last game.

I'd really like to see some effort from Riot to control the quality of games by reducing the stress of the environment, rather than completely focusing on the numerical issues.

Edit:
Also, I totally agree with this guy's idea ^ about normal Elo translating into starting ranked Elo; judging by some of my friends' normal W/L and their ranked W/L + Elo, there's a strong correlation.
Just because we are working on matchmaking does not mean we are not working on other things!

I agree, player behavior is one of our most important problems. But, I am doing research on several things right now related to player behavior, and wanted to at least update the community about the matchmaking projects first.


24 months ago

Quote:
Originally Posted by SkÿNet View Post
Whenever you get the chance....

http://na.leagueoflegends.com/board/....php?t=1982126

A perfect example of flawed matchmaking. It's all there, you don't have to give me a definitive answer about what happened right now, but I'd at least appreciate a response saying you've seen it.
You are a 2000+ Elo player, so the matchmaker is trying to find you opponents that actually are challenging. The problem is, you are on a smurf, and there are no players below Level 30 that actually are 2000+ Elo.

Even though you lost that game, you were easily competitive with them because of your skill. We are adjusting level parameters a bit so that you are less likely to face Level 30s when you are < Level 30, but for you in particular, there will probably never be opponents < Level 30 who can match with a 2000+ Elo player.

The matchmaker isn't broken in this case, the top players in the game are simply always going to have some issues like this.


24 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
Lyte, why not remove duo queue from solo queue? And let it just be solo queue? What would be the problem there? Sure people wanna play with their friends, but there are other modes for that.
Well, it's surprising, but most people play Ranked Mode as a Duo-Queue, and for most people that is their preferred approach to Ranked Mode.

People just focus on the games where a Duo-Queue ruins their game, but they do not notice all the games where a Duo-Queue was on their team and they win. Duo-Queues are far from 'rare' in Ranked Mode... players just don't notice it, but we can see the data

Segmenting Ranked Mode into several different queues is not a feasible idea. Even now, with the current player pools, matchmaking has difficulty finding 'fair' matches sometimes. We have to be very careful segmenting the Ranked population into more and more queues.


24 months ago

Quote:
Originally Posted by DreamsOfGrandeur View Post
Isn't it arguable that most people now consistently duo-queue in ranked solo queue BECAUSE going solo is the non-feasible way to go?

So, you're sitting in a bit of circular logic, Lyte.

If duo-queue was removed, the people who ride on others' coattails would drift downward to their intended ELO and we would eventually see improvement in matchmaking as ELOs are closer to their true levels on everyone instead of half of them.


It is inherently flawed for a solo queue to allow duo-queue parties to join in where others are working with strangers.
In normals, this could be left alone because there's no relevancy, but in ranked this is a very serious matter as people play ranked so show their skill in hopes of getting equal opportunities.

I don't think anyone is really asking for segmented queues. They just want duo-queues gone.
People only duo-queue because it's the best way and any other way is less feasible. It's the meta of solo queue.
Solo Queue is very feasible to climb up the ladder, neither Solo nor Duo-Queue is clearly 'easier' to climb up the ladder, it depends on other factors. In fact, in many cases, it is easier to climb up as a Solo Queue, but players still want to Duo-Queue.


24 months ago

Quote:
Originally Posted by NettikRellik View Post
Hey Lyte, quick question (about queue dodging). Say you get a duo-queue troll in your ranked and they go double AD Carry AFTER Someone else has selected it and said they were going to, knowing that they are trolling and will make the game less fun for the other three people on the team, you decide to queue dodge to avoid the horrible trolls. Should there be punishment for this even though they were the ones 'forcing' you to dodge to avoid the inevitable loss?
There should be some penalty, or else players would simply queue-dodge every single lobby until they got the 'perfect' composition they were happy with, or until they got the champion or role they wanted. If every single player did this, it would take you 20 minutes just to actually get into a game due to excessive queue dodges.

However, I don't believe that -Elo in Ranked for queue dodging is the right penalty, which is why we are changing that.


24 months ago

Quote:
Originally Posted by Czxr View Post
Actually almost all random number generators are technically pseudorandom number generators. For a random number generator to not be a pseudorandom number generator is has to be measured from a random physical phenomenon such as radioactive decay or thermal noise.

Computer algorithms are predictable and therefore pseudorandom, the old shuffles would have been pseudorandom.

On topic: I can agree with most of the issues being important, except for removing free champions in ranked. I think having free champions gives a great opportunity for swapping champions. Also, if they were removed that would mean someone who had played 50 games on one champion in their free weeks but did not own it wouldn't be able to play a champion when another person who just bought one and never played it could play it in ranked.
Well yes, all computer random generators are pseudo-random, but you got the main point

We are still talking about the Free Champions in Ranked issue. Yes, it is possible that you can play a Champion 50x every time it appears in Free Rotations; however, that is less likely than a player who buys a Champion but never plays it.


24 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
That would make sense.

But then still, why not remove duo queue entirely? I mean, that would seem to really fix some things with people feeling like the system is wrong. It does seem odd that a team can have a duo lane which gives them an advantage and the other side might not right? Or does the system try and match teams up so that there are an equal number of duo queues against each other?
The problem with removing Duo Queue entirely is that the majority of the players in Ranked Mode prefer Duo Queue, and play as a Duo Queue...

The forum population is a very small, very vocal population. But the data of all the players playing Ranked Mode suggests that players prefer playing as a Duo-Queue in Ranked Mode. This is why it doesn't make sense to remove it


24 months ago

Quote:
Originally Posted by SkÿNet View Post
So you're saying matchmaking knows that I'm not just a smurf, but I am SkÿNet's smurf? Are you suggesting matchmaking is actually self-aware?

The issue wasn't particularly about me, it's regarding my entire team. If I'm a 2k player, why should I be getting matched against 2k players, but not with them? If matchmaking wasn't flawed, my team would have reflected their team too, not just me. You completely neglect to answer that matter. I was duo queued with a level 19 player who is roughly 70/80 in normal w/l, so why doesn't matchmaking take that into account, not just my elo?

I'm not sure if you were just unobservant to the rest of my team, or you just didn't want to answer it, but that is the underlying issue here. Not my own elo, but the elo difference between teams. Please, enlighten me on that.
Matchmaking did take the rest of your teammates into account. The problem is, anytime there is a 'top' player in the match, it is very difficult to find 9 people who fit into that match, especially if the 'top' player is actually on a smurf account and the player is smurfing with a friend who is completely different in skill.

It doesn't mean matchmaking is flawed, it is just extremely hard to find good matches when 1 player is a top player and that player is on a smurf account and is queued with a friend who is NOT a top player.

Answer this: What team would you have liked to see? What are the odds that team exists in the matchmaking pool, when your team queued up?


24 months ago

Quote:
Originally Posted by ncrwhale View Post
Then why not keep track of how many times a champion has been played instead of ownership? You don't want someone buying a champion and playing him for the first time in ranked, do you? That's just as toxic.

Also, Lyte, you are doing a phenomenal job.
Free to Play Champions in Ranked has no clearcut solution. At least, there hasn't been a solution we thought of, or suggested by players, that solves all the different use cases.

We could keep track of how many times you play a Champion, and 'unlock' that Champion for Ranked Mode. However, what about people who buy Champions just to trade to teammates and who do not actually want to play the Champion?

A player could buy a Champion and play him for the first time in Ranked; however, on average, the majority of players 1) play the Champions they buy and 2) play those Champions in a non-Ranked mode.


24 months ago

Quote:
Originally Posted by SexyTomato View Post
FINAL SUGGESTION:
Include an option that allows players to mutually agree to increase the risk-reward. That way, players get the opportunity to live up to their claims that they could easily win when they have a semi-competent/mature team.
Explanation: This suggestion is not targeted at any of the issues mentioned, but rather at matchmaking as a whole. Many players comment that they lose because of trolls and baddies. Furthermore, it is much less frustrating to lose because of bad play than it is to lose because of trolls.
What does this actually mean? Players can choose to risk more Elo in a specific match?


24 months ago

Quote:
Originally Posted by HKBS Shiros View Post
Far too often are solo queue games mostly a stomp one way or the other. I don't think it's right to penalize people who do poorly, but, the people who do exceptionally well, like going (high kills/low deaths or low deaths/high assists) and still losing because their teammates aren't up to par with the other team don't deserve such a drastic lose of elo as they tend to receive.

What stuff like that brings me to believe, is that being able to play in ranked is far far too easy. I don't know how you'd go about it, but, there needs to be more limitations to when you can start. Far too many people are not prepared for ranked and tend to bog down a team forcing others to lose, because, no matter what people say, you can't really carry a game by yourself. This game is far too team based, especially with summoner spells and the plethora of escape/disengage abilities on champions, making one person very unable to pick people off and push lanes by themselves. I'm on the boat for saying that 1) free champions shouldn't be allowed to be played in ranked unless you own them and 2) 450 IP champions should not count for "champion amount you must own to play ranked". Those two things may help slightly, but I still think more would need to be done (again I don't really know what).

Also, you value duo-queue far too highly. Constantly pairing duo-queue with players higher elo than them, forcing them to be the bottom two picks, should not be how it is. Place them with their proper elo ratings. If they end up being the last two, so be it, but they should still have the ability to be otherwise. It makes duo-queue too discouraging to me, constantly forced to play bot lane, what if neither you or your partner specialize in bot lane? It's not like your team will listen to you at all (normally).

Finally, I'd like to support this fellow's idea. I personally enjoy SC2's ranked system. Once you reach a bracket, you're in that bracket for the rest of the season (can't drop out of it, only become last place in the bracket) unless you progress out of it. It's really an accomplishing feeling to know that once you fight to a certain bracket, no jackass can tank a game for you and push you out of that bracket, only down in the ratings for said bracket.

I ranted far too much for most people to read this and I highly doubt it will be touched by Lyte at all, but I've had a need to get my thoughts out for a while. Sorry, no TL;DR.

Tidbit: I played a game with you a few weeks back, Lyte, you were my first Rioter. Happy to have played with you!
I have played a lot of other games in my time (surprise!) and I do enjoy some systems like the ones you mentioned. On the side, I am currently researching every ranked system used in online games to see the best features of each one and what works well/what doesn't.

I agree that the transition from Normal to Ranked is poorly executed right now, but would like to avoid hard restrictions to Ranked Play such as playing 300 Normal Games, or 5 Games with each Champion. We want more players to play Ranked, and making the barrier so high that it requires 200-300+ more hours of play just to open Ranked will probably lower the number of players playing.


24 months ago

Quote:
Originally Posted by Dazrix View Post
I don't think this would be practical I'm afraid

Lyte, is there any other way we could help you in your decision/solution?

EDIT: Also Lyte, would it be possible to incur suspensions on ranked mode itself? For example, if your team hits the report function on you unanimously, you incur a say, 24 hour suspension from ranked. This could work similar to a queue dodge penalty timer. It'd just be for 24 hours instead of like 10 minutes.
Yes, we're working on ways to do time penalties for specific game modes.


24 months ago

Quote:
Originally Posted by Cyndikate View Post
It doesn't change the fact that Duo Queue contributes to bad matchmaking, and problems with autolocking squishies and never wanting to lane anywhere but with each other, especially when Annie is suppose to go mid and Garen top lane. We call out Annie for feeding and Garen starts defending her.
I agree, Duo-Queue adds a lot of noise; however, that's why we are going to try to reduce the noise and try to patch up Duo-Queue situations in Ranked rather than remove it entirely.

If things still do not work and Duo-Queue simply adds too much noise, I will start considering some Elo restrictions to Duo-Queues.


24 months ago

Quote:
Originally Posted by ncrwhale View Post
Hi Lyte,

Do you have any thoughts on being able to swap picking order?

This would help even out competitive play for people who don't have as many champs (because, as is, they can't swap champs they don't have).

It would also allow for a minimum number of character plays before being allowed to select that character in ranked without penalizing people who bought champions just to swap.

Thanks much
It's in the discussions, but I am concerned about the timing of Champion Select Lobbies. Now when you enter a lobby, players have to debate what Champions they want, what lanes those Champions go, whether someone wants to trade Pick Orders... it seems like a lot to discuss in the limited timers you have in the lobby.

I would prefer to see some work done on improving communication channels in Champion Select Lobbies first to enable our players to communicate efficiently with each other about their preferences for the game.


24 months ago

Quote:
Originally Posted by Wendek View Post
What about being able to decide picking orders when you're premade ? That is, in the lobby before the game starts so there's no communication and timing problem.
I often play normals with 2 to 4 friends and it's annoying when I am firstpick if I wanted to solotop, etc. Of course trading is possible but some of my friends have quite a limited champ. pool and can't really trade much.

This is kinda off-topic with matchmaking, now that I think about it... but meh, I didn't bring the subject. x)
This has been brought up before, particularly with Ranked Teams and allowing players to pre-select their pick order. It's something we may work on, but separate from the matchmaking initiative.


24 months ago

Quote:
Originally Posted by Interceptor402 View Post
It is worse now *without* the ability to do this! WORSE!

So much time is wasted in the lobby because we are not allowed to swap pick order, and instead have to negotiate mutual champions (or worse, a 3-way swap) for trades. The way it works now, a bunch of time is blown in communication between players trying to pick for each other, because the nature of LoL means that two players owning the same wanted champion doesn't actually happen very often, especially for the expensive 6300IP picks. It's fine if Ashe wants to swap with Annie, but not so much if Wukong and Kog'maw want to trade.
Well, let me present you a scenario.

Player 1: "I'll trade first pick if someone wants to choose a strong AD. I'd like to counter top."
Player 2: "Sure, I'll choose AD first, you can take my slot."
--- swap happens ---
Player 2 chooses Warwick and says, "I'm going top."


24 months ago

Quote:
Originally Posted by Dazrix View Post
As much as I like these little ideas, I think fixing the ranked ELO system would be much better.

I suggested this in another thread previously Lyte, but perhaps having like.. an ELO offset factor (similar to SC2's 'bonus pool') that may assist in placing people more appropriately.

In particularly for the newer players to ranked who haven't yet played many games of ranked (or any) where there team mates are given a +/- of say, 3 elo per player on there team that is determined by the system to be 'inexperienced' to ranked.
IE. People that have played less then a set number of ranked games, an example might be, 40 ranked games(or wins if you prefer but this might be a bit extreme) will determine you to be 'new' to ranked. So when you play ranked each season when everyone starts off at 1200~ alot of scaling will occur for the first 40 matches everyone plays. This will also help the 'better' players rise to the correct ELO's they should be in, as they will rise much faster. And the lower ELO players will eventually settle into there place, but will allow for the 'average' player to level out also.

As it stands now, I feel like an average player will probably sink more then he will 'mellow' because with the new players that come in, he'll drop in ELO eventually but slowly. It's not about belonging there, it's about the quicksand thats only got a hold of your feet, so when you try to pull your feet out, it creates a vaccuum suction, and you can't get them out... and so you slowly start sinking into mud again.

Having said this, it should raise the % of players ABOVE the 1250 mark from 25%

cap the ELO loss/gained at say.. +/- 6 which would effectively accomodate 2 new players on a team. So the ELO loss would still be felt, but it wouldn't be so unforgiving.

Btw, Lyte whats your degree in? (just out of curiosity)
I like the idea of a bonus pool; in fact, our current system allowing you to gain +/- 40 per win/loss when you first start playing Ranked is very similar to a bonus pool. But, it might be worth considering showing our players what their remaining bonus pool is.

My PhD is in Cognitive Neuroscience (a combination of Psychology and Neuroscience).


24 months ago

Quote:
Originally Posted by timtwins View Post
Let me present you a scenario that happens in ranked too.

Player 1: "Pick me Janna, she is a highly contested pick and I'm afraid the other team wants her. You can have whatever champ you want.
Player 2: "Sure, I'll choose Janna, get me Akali."
-- Player 1 picks Janna --
Player 2 chooses whatever champion they want and leaves first pick with support.
Yes, these scenarios happen too.

But, fixing 1 case and introducing 2 more isn't solving the problem.


24 months ago

Quote:
Originally Posted by Triscadeca View Post
Hey Lyte, I have an idea that may help this that at least seems like it could be implemented fairly easily. What if, during draft pick, after bans, you could select a champion from the roster when it was not your pick but you couldn't lock it in? Selecting a champion would only show to your team. It could also show swaps when you select a champ. This way you could have at least some idea of what everyone wanted to do fairly quickly. I don't think this should block the person who's pick it is from locking that same champ though.
This is a good idea we've discussed recently and want to try to do.


24 months ago

Quote:
Originally Posted by Soup is lame View Post
While I understand Riot not wanting to "alienate" the general LoL population by restricting ranked play to, say, those who have won 300+ games, you have to consider why most people who care about competitive play (and, if I were to guess, most of the contributors to this thread/poll) are in ranked in the first place.

It's because we are COMPETITIVE players who want to take the game seriously with other COMPETITIVE people.
We are all making an assumption that experience 'solves' the problem. Let's say players need to play 300 games before they can unlock Ranked Mode. Even if this was true, there will still be players who are not 'good' in Ranked Mode, or are the 'worst' players in Ranked Mode. Why should we ban these players from 'competing?'

Even 700 Elo players may want to play in a more serious environment and test themselves and try to 'progress' up the ladder. We should improve the matchmaking so that all players are always placed appropriately <-- that is the key problem. When players are not placed appropriately at their appropriate tiers, then we have arguments that they are 'not ready' for Ranked, and don't 'belong' at 1200 Elo, etc.

We could make you play 1000 games before you could play Ranked, and players would still be arguing about the same issues of players who do not belong in Ranked.


24 months ago

Quote:
Originally Posted by Sypher76 View Post
Hey Lyte,

I'm still hoping you'll talk about the 'elo floor' concept a bit more. All we know is that you see some opportunities for trolling in it - not whether you consider it even possible, or any further thoughts on the matter.

I'm just hoping you'll briefly elaborate on where Riot stands on the idea.
I have been testing several different kinds of rating floors--various algorithms and ways to do rating floors; however, none of the systems I've researched and tested have been satisfactory to me

To be a bit vague, rating floors solved particular issues that when I broke it down further, would be better solved by something else.


24 months ago

Quote:
Originally Posted by Jaceive View Post
Lyte: what if you had a "ready up" room before you got into champion select in ranked? Here's what I'm thinking:

Match-making finds you a team of the appropriate elo: Team goes into a room where they can discuss lanes, pick order, swaps, etc... when they are ready they hit "go" and Match-making finds them an appropriate enemy team. Then they go forward with picks and bans.
I like the idea of a "ready up" system, but we would have to figure out ways to avoid it taking 20 minutes from hitting "Play" to actually getting into a game.


24 months ago

Quote:
Originally Posted by Alyssa Riel View Post
Lyte i know you are probably getting tired of hearing from me, but for rotation champs in ranked couldn't you cut the middle ground by saying they do not count towards your champion count, but they are available for selection and trade that week. This has the appeal of increasing player experience or riot revenue before a player enters ranked, but it also does still maintain that week to week reliability that every players will have at least some champions of each type.
Never tired of hearing from our players I even read every single post in this thread, just unfortunately cannot spend time replying to every single post.

There's been a lot of discussion on how we should execute the "Random Champion in Ranked Play" issue, and I agree with many of the concerns and use-cases. However, I think what we ultimately will do with this issue is go with the simple, straight forward solution that solves 80% of the problem, then work on other high priority items first, and re-visit the issue in the future and expand upon the solution to cover maybe another 10-15%.

I would love to include a "trade-enabled" feature where you can trade Free to Play Champions but cannot play unless you own them, but this would add a lot more dev time to the feature. The data suggests that if we first just removed Free to Play Champions from Ranked play, we can solve a majority of the problem--this also means we would have to revisit this issue again in the near future.


24 months ago

Quote:
Originally Posted by Soup is lame View Post
Exactly.

Believe me I completely understand that two different people will sometimes have differing opinions on their definition of "fun."

But when one person's "fun" is derived from not taking a game mode seriously that is, by definition, intended for those who want to take their games seriously, those people should not be allowed to play that mode until they are willing to be team players. There should be an iron-fist rule in play for ranked. I honestly don't care if people feel alienated by it. If you don't want to take the game seriously, don't play ranked. If it doesn't bother you that you are 3-8, don't play ranked. If you have "elo to burn," don't play ranked. If you don't care about your teammates, and only want to play teemo top, and never show up for team fights, don't play ranked. If you would rather lock in revive promote eve than play support because you are throwing a 5th pick tantrum because mid lane was already taken, don't play ranked.

I don't understand why there isn't a zero tolerance policy for this stuff. If you are trying to seriously grow league of legends as a pioneer and leader in the competitive e-sports scene, then you have to start correcting the problems at the source. If your kids are smearing poop on the walls of your office, lock the door.
Part of this is our fault. We have not executed a great transition from Normal to Ranked Mode, so many players may not actually know that Ranked Mode is "bring your best, every game!" mode. To them, Ranked Mode is just another mode that has Elo tracking instead.

Once we have better tools in place to educate players on what Ranked actually means, I would be much more confident implementing a zero-tolerance policy to negative behavior in Ranked Mode. In saying this, players in Ranked Mode generally report more often, so toxic players in Ranked tend to reach Tribunal faster already.


24 months ago

Quote:
Originally Posted by VeriNasti View Post
Wouldn't not having queue dodging lowering your elo lead to more 'trolls' in game so they can purposely lower it?
Not necessarily.

1) It takes longer to drop your Elo through games, and it takes far, far more effort.
2) Intentionally trolling games to drop your Elo leaves you vulnerable to Tribunal cases.

Yes, if players want to lower their Elo, they can still intentionally troll in a game and drop their Elo over hundreds of actual games. When players are trolling games and intentionally dropping their Elo, they still have to "act" like they are trying their best and put on a 'troll disguise,' or they are vulnerable to Tribunal punishments. Our players are generally pretty good at detecting troll disguises like this, and this is actually a lot of 'effort' on the troll's part.

3) Data suggests that the players who will actually 'fake try' and troll enough games to get low in Elo is pretty low; so there might be a few people who troll a few games on purpose to drop their Elo but overall, we are preventing a majority of the attempts to drop Elo and stomp lower level players.


24 months ago

Quote:
Originally Posted by CupcakeTrap View Post
Hey Lyte,

I made an ideas thread a while ago to compile solutions to the "new players being driven off" problem.

re: "I invited friends to play. They quit." (idea thread)
http://na.leagueoflegends.com/board/....php?t=1872516


I know this thread is about something subtly different, but your comment about "the transition from normal to ranked" reminded me of the transition from tutorial/bot games to normal play.

I think there are some good ideas in that thread, which I've tried to summarize in the OP. Some are extremely simple, like a pop-up message before your first matchmade game.
2 of the simpler things on that list have been cooking... and I think you'll be pretty happy.


24 months ago

Quote:
Originally Posted by RenegadeTQ View Post
Lyte, what do you think of my idea for a new item "Weighted Boots"? http://na.leagueoflegends.com/board/...weighted+boots

The thread started off well but lost.....traction. *pinky to corner of mouth*
I am mainly involved with all player behavior initiatives such as matchmaking, Tribunal, etc. Live Balance and champion/item changes are more Geeves domain.


24 months ago

Quote:
Originally Posted by Alyssa Riel View Post
My previous post based on the numbers they gave us at the end of season one is that the average elo is not where players are starting. 1080 is over a full hundred points below the starting point, and that is a problem. i know it doesn't feel like a 100 point elo change is that huge, but jsut playing through it a couple times you can really tell the difference.

If i am mistaken and the average is not 1080 then i'd be willing to concede the point, but by my understanding that is the case.

Also sorry i asked you to get that data i could have gotten that alone, i had held onto the hope that you had more accurate numbers below the 1200's.

Lyte do you think we could get some player percentages and averages for the current elo system so we can make some more informed decisions?
Unfortunately, we cannot release exact numbers for many of the things you guys are discussing. It's not because we want to hide data from our fans, it's a business decision we have to respect. Once in awhile, like at the end of a season, some limited stats will be released.


24 months ago

Quote:
Originally Posted by isobold View Post
About matchmaking: The biggest improvement I could imagine on match making would be to have it guess which players could harmonize as a team. The paper I already linked before: PlayerRating: A Reputation System for Multiplayer Online Games tries to allow for guesses about players harmonizing.
However it has the potential to impact on Elo-distribution. That's why it would need extensive simulation and testing, before introducing it to the actual matchmaking.

I would really like to know what Lyte thinks about this paper btw. . I'm pretty sure he knows the paper and if he doesn't, I think he would really like the reading.
Been super busy lately but wanted to jump in on my break. I know the paper and think an adaptation of it can work; however, it is not within the scope of this matchmaking initiative.

Aside from the mechanical issues of how it changes the Elo distribution, from a social perspective, if you put positive players with positive players, that means indirectly negative players are more likely stuck with negative players. Research suggests that the more negative you more, the more negative the people around you because they reciprocate your toxic behavior... creating a situation where some games are many magnitudes more toxic then they need to be. We think about these general issues around the offices a lot, and I have confidence we'll figure out a good solution.


24 months ago

Quote:
Originally Posted by isobold View Post
IMO the "noob-island" should last longer and the elo-variance should be smoothed out of the course of the matches, like it is now. Ex:
1: +/-50Elo
2: +/-45Elo
3: +/-40Elo
4: +/-36Elo
5: +/-32Elo
6: +/-29Elo
.
.
.
50: +/-12Elo
Those numbers being averages ...
Only after those 50 matches the player should be able to see his Elo at all ...
Yegg and I are thinking about this right now, and how long "Newbie Island" should be, and how long before a player actually sees their Elo in Ranked. Personally, I am leaning towards Newbie Island needs to last a bit longer, and players won't see Elo until it ends.


24 months ago

Quote:
Originally Posted by kidco View Post
So I just got banned for the second time today for 3 days and I realized pretty much the process of banning for Riot. The first time I got banned I admit I deserved that one cuz I actually did troll feed and afk but this time I realize what riot actually looks for when banning you.
You were not banned for playing a prank.

You were banned for screaming in chat in all caps , berating other players, telling them they are tards, that they suck, etc. In the first few games I counted 17 really inappropriate lines in all caps.

I'd really work on changing your behavior


24 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
What about while you're doing the "newbie island" games (why is it called that? >.>) you can't duo queue. That is, atm it's 10 games. The first 10 games you CAN NOT duo queue. If you make it longer, say, 20, then it'd bet the first 20 games. This way you can be more likely properly placed by the system and if you enforce a duo queue must be within X (say +/- 50) Elo of you, then you'd get better matchmaking right? Then you wouldn't have to artificially place duo queues in higher Elo groups or whatnot.
Yeah, we have talked about restricting Duo-Queues during "Newbie Island," however, if we are increasing the length of Newbie Island, that's more games where a player is forced to play alone when they prefer to play with a friend. There's a good compromise somewhere


24 months ago

Quote:
Originally Posted by Keyra View Post
Hi Lyte i'd like to bring something to your attention,

First off I would like to say thank you for your hard work. Most of the problems I wanted to address were already mentioned. Now the suggestion I wanted to address to you is to be weary of rating / voting systems at the end of games. The reason why is because some unskilled player will down vote a player due to inexperience.

For example, a top lanner who is super aggressive and has been pushing his lane for 7/8 minutes. Dies the first time and continues to be aggressive. Dies once again and complains to jungler that he never ganks top lane. In this case the top lanner will downvote the jungler because he didn't gank. BUT, in fact the top lanner should not have down voted the jungler because junglers are "technically" not suppose to gank lanes that are already pushed.

Second situation is bottom lane, AD's get A LOT of heat when doing bad, and at times it IS partially due to the support. For example a soraka support who constantly harasses with her "e" rather than giving mana. The AD dies a couple times and gets all the heat for the team losing when the AD fed bottom lane when actually the support could have helped often but just let his/ her AD die. Therefore the AD would get a downvote at the end of the game when they shouldn't have received a down vote.

I've noticed this often enough as I hover between the 900-1200 elo where people just don't understand how to play their character / role or just don't understand the concept of the game.

So just a warning with the voting system with unskilled / unknowledgeable players down voting at times when they shouldn't be.

Thanks again for all your hard work ^_^

TL;DR:
if a voting system is implemented, unskilled / unknowledgeable players would down vote others due to wrong reasoning
Hi Keyra, thanks for the comments. We are not moving towards any type of system where a player can rate another player's performance or skill at the end of a game because of many of the issues you listed; I don't feel it provides that much value to the game from a player's perspective, or a matchmaking perspective.


24 months ago

Quote:
Originally Posted by Ryuzaaki View Post
Alright Lyte, I'm sure this has been mentioned somewhere in this thread or the other, but I'm going to say it again just in case. Are you guys considering doing any work on elo disparity between normal and ranked? I play almost exclusively ranked, and not just because I want to climb elo there. My normal elo feels so far behind in player skill level, that it sucks all desire to play there. When the games are even possible to be carried through, which is not often, it's not as rewarding as ranked so there's not much point.
Yes, there are some plans to link Normal and Ranked Elo, such that if your Ranked Elo is boosted super high it will 'drag' your Normal Elo upwards a bit.


24 months ago

Quote:
Originally Posted by MrAngels View Post
Hey Lyte,

Zileas says we're allowed to troll you about having a PhD in Cognitive Neuroscience. Just thought you should be prepared.

In other news, how's your mind-control project coming?
Quote:
Originally Posted by GoldVicious View Post
I was in a game with Lyte the other day, I think he must be working out the kinks, he was controlling his character but very poorly.
It's a work in progress. My brain could be failing me, or the new mind-control cap is...


24 months ago

Quote:
Originally Posted by aaFen View Post
Will this work the other way at all (normal influencing ranked)? When I first tried ranked in season 1, I couldn't have played more than 15 ranked games total before the season ended (and always in small spurts of one or two games separated by weeks) because of how bad the experience was. I had already toughed out bad matchmaking for a long time in normals to get to a place I was relatively okay with (better than what I seemed to get in ranked at least), so there wasn't much motivation for me to do ranked. Same thing with season 2. Played some games at the very beginning since it was basically a grab-bag for who anyone would get in a game and the total chaos was sort of fun for a few games. Stopped playing for a while, tried it in one or two-game spurts again and was annoyed enough with the people in the games that I haven't really cared since. What motivation do I have to play ranked when I get more players that are more competitive and have better attitudes in my normal games?
I can't speak too much about this except I generally will not want to give players the ability to play Normal to up their Ranked Elo. In addition, some players want to make sure Normal Elo does not influence Ranked Elo because they play 'differently' in Normal or only play with lower skill friends or only mess around and do not want to be punished and started at a low Elo when they eventually try out Ranked.

We're thinking on the various ways we can allow the modes to interact that would provide value though.


24 months ago

Quote:
Originally Posted by ncrwhale View Post
Can you bring down your normal elo with a lower ranked elo?
Nope.


24 months ago

I honestly feel that there should be no penalty on the first que dodge of the day, especially in Ranked. Lets say you are about to go in a game, but then Janna decides to go Smite Revive. No one should have to put up with that and suffer the ELO loss. I admit I dodged that one, but that is the first one I have dodged in a while.

Another instance for ranked, an afk banner or non-communicative team. How are people supposed to get their ELO higher if they get paired with players that obviously don't belong in Ranked.



24 months ago

Quote:
Originally Posted by Kai Talakki View Post
I feel like this is a significant problem. Even with a premade team, even one level 30 can be devastating. The group I do pvp with does not have a high win-lose ratio, and most of our losses are from matches in which there was a level 30 who turned an even match into a hopeless game. Our levels are at a range (between 14 and 21), and those of us that are higher level are not as skilled as a level 30 with 309 wins...Someone who has played that many games should not be pitted against a team with no one that is on par with that. I understand premade teams have an advantage in that the people in that team have better communication with each other and work together better than five randoms, but someone who has that high of a skill level compared to everyone in the premade team is devastating.
I'm working on a few things that will help smooth out the experiences for premades that are greatly different in skill and levels.


24 months ago

Quote:
Originally Posted by Alyssa Riel View Post
The problem with League in a nutshell ^_^

http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect


Our brains are simultaneously amazing and silly at the same time.


24 months ago

Quote:
Originally Posted by Alan MQ View Post
Why Twisted Treeline is being ignored?????
Is there a matchmaking concern you had with Twisted Treeline?


24 months ago

Quote:
Originally Posted by Ko Hakoo View Post
I know I've posted this a few times without a response so I appologize for spamming, but it seems that TT and SR have the same normal Elo (anecdotal).

Some skills transfer, but throwing a 1400 Elo SR player into their first TT game at 1400 Elo spells doom so fast. If it is the same number, simply playing TT (and likely losing) will negatively impact your matches when you return to SR.
TT and SR Elos are tied a bit too closely at the moment and we would like to separate it. Currently it is not on this initiative because a different team owns this task.


24 months ago

Quote:
Originally Posted by DarrkWolfe View Post
hey Lyte, I have more to submit to the Ranked Problem but I don't know what the correct Email address is.

Is this it?

Matchmaking @riotgames .com


I'm getting frustrated as my findings are getting worse that I am about to put forth in confidence.
Yes, that is the correct e-mail address.


24 months ago

Quote:
Originally Posted by Merluza00 View Post
Lyte, how often do you update normal ELO for matchmaking?
Every often, if I have a 3 in a rough wins and stop playing, the next day I have a lose streak, then I get pissed off and dissapointed with myself, but the next day I get a win streak of several games and so on until the circle is broken for some reason and all come back to the typical 1 lose 1 win streak. I have the feeling that my ELO is updated once per day so the team with who I'm matched is sometimes unbalanced respect me (I admit I love win streaks even though they are not very frequents). Is this true or is it that I am just too bipolar and somedays I do good and others I suck? The weird thing is that If I have winning streaks, they are ALWAYS followed by a lose streak :/

inb4: I am not sure if it spells 'streak' or 'strike'
Elo is updated after every game.


24 months ago

Quote:
Originally Posted by exe3 View Post
No it doesn't but when I can identify someone who causes us to lose our games and is clearly of a significantly lower skill level and they all consistently have a low win rate when I check after the game ends then well, yeah...bit of a theme going. Sure i've seen good players with 200 wins (probably smurfs) but those are very rare exceptions.

Also love getting downvoted when I said nothing wrong. Such a stupid system. I just want matchmaking to provide fair matches.
There's a few unfair assumptions you are making here. For example, you assume that when a 200 player loses, they are 'just bad players.' However, when a 200 player wins, they are 'a rare smurf.'

We have been looking at the data related to number of wins and how it translates to player performance or skill. This is one of those issues where we cannot rely on anecdotes or human pattern recognition


24 months ago

Quote:
Originally Posted by Imclueless View Post
Hey, Lyte. I was wondering if there's anything that can be done about this:

So, myself (among many others) have noticed that, at certain times during the week or day, the skill level of players within an Elo bracket shifts. To be less vague: when you queue up on the weekend, players are worse, on average.

This isn't a rule or anything, but it's a trend that my friends and I have noticed. We've all got our Normal Elo's up high enough where games are competitive and pretty standard every time. But, when we queue on the weekends (or every day around the time of Spring Break), we find ourselves in the middle of players who had no idea what they are doing. They had a decent mechanical grasp of the game, but people don't know why a jungler's a good idea or what Support is.

Like I said, this isn't an absolute rule, but it's a trend we've noticed over time. It's like a different set of players plays on the weekends, so the Elos don't quite match up between them. Which, having said that, probably has a grain of truth in it.

This isn't exactly a big issue, and I'm not even sure there's anything to be done about it. Though, winning against these players feels almost dirty when we run into them. We're taking something like Soraka/Graves against people just running whoever they want in a lane together, and we can take things like Dragon with no contest most of the time.

Do you have any comments regarding this? Now, what I don't know is if this is also found in Ranked. If the forums are any indication, it very well could be.

Also, quick question. Does queue dodging in Normal games lower your Normal Elo like in Ranked?
There are certainly trends due to time of day, or day of the week that alter what the average skill level is for any Elo bracket; however, given the number of things currently on our plate, we are not going to try to tackle this issue for now

Queue dodging in Normal does not lower your Normal Elo at this time.


24 months ago

Quote:
Originally Posted by MowiOwi View Post
As I am not level 30 and cannot play ranked I hope my input is still appreciated.

I just got out of a game where my team was made up of a level 30, 19, 25, 28, and another 25 (me). The opposing team had 30, 30, 30, 17, and 20. I realize levels don't mean much, but when you have 3 people on the opposing team with full rune pages and several won games above the rest of my team it is a bit of a joke. The sad part is I was doing really well until my teammate's inexperience (or lack of rune pages) began to show itself. We could have won the game except for the huge disadvantage my team had. I have around 18 runes and 1 quint and I was destroying the other team. My back was not strong enough to carry 3 players who died often and how could it be with the level and rune page differences? Normally when my team and I lose a game it's because the other team has more level 30s than we do. Every once in a while we win, but mostly lose because the other team has 3 or 4 level 30s and our highest is normally me, currently at level 25.

It is very upsetting when teams like this are created with the matchmaking system. It feels like teams are put together so only one team wins. I've been told it's because I have a high summoner level and there are a lot of terrible level 30 players. If I'm stuck with the terrible level 30 players I'm also stuck with the terrible or good players who are around my level that simply will not win because of previously mentioned facts.

All I am asking for is a fair match. I do not care if I win or lose I just want a fair game.
Generally, matches where you are lower level and face Level 30s are pretty fair--you do win those games more often than not; however, I do realize that players see Level 30s in their game and perceive the matchmaker to be broken, or the match to be unfair. Even when lower levels beat Level 30s they still feel the game was not as fun!

We are going to tighten up the level requirements so that there's a lower possibility that lower levels will face Level 30s unless you queue up with a Level 30 friend; however, this will increase queue times slightly for lower level players who are highly skilled.


24 months ago

Quote:
Originally Posted by An Angry Janitor View Post
Lyte; I recently thought of an idea to improve ranked and remove the toxic of first pick what I do goes. What if the bans were done by the first pick slot and the first pick was done by the last pick slot? Or bans done by last pick first pick done by; well first pick. I think it would bring more insight for the 'team' aspect of the game that riot is seeking. It should not be under the control of one person to get their choice of bans and their first choice of pick over 4 other players (mainly last pick since they are essentially bound to support play or flat out have 0 input with bans considering 1st slot has all the control). Where at least picks 2,3 and 4 each can control the champion that they desire and even counter the original first pick (since no one even does trades in ranked in order to counter an early ap or top etc), Just seemed like a good idea.
It's an interesting idea. Yegg and I have been playing around with how pick orders should be in Ranked Mode because the lowest Elo player being last pick is not necessarily the best approach.


24 months ago

Quote:
Originally Posted by DarrkWolfe View Post
Dear Lyte,

I have tried sending you an email but I have gotten an error:

Delivery to the following recipient failed permanently:

Matchmaking@riotgames.com

Technical details of permanent failure:
Google tried to deliver your message, but it was rejected by the recipient domain. We recommend contacting the other email provider for further information about the cause of this error. The error that the other server returned was: 552 552 Error: message too large (state 17).


Ideas of what to do? :|
Hmm, I'm not sure! I tested the e-mail and it worked fine and I am still getting e-mails from players. Maybe try a different e-mail or client/browser?


24 months ago

Quote:
Originally Posted by Imclueless View Post
Hey, Lyte. Please don't hate me for this (I love you for responding to me in this thread and others).

http://penny-arcade.com/patv/episode/harassment

I know that this thread is about the matchmaker, and this is off-topic, BUT, I know that you are working on player behavior and such.
I don't hate our players for wondering about the future of player behavior in League of Legends. As one of the people leading the player behavior movement at Riot, I am constantly looking to the future and figuring out what we can do to improve the player experience in League; however, we are not ready to talk about any of the other projects I am working on.

PS - I saw that episode of PATV early this morning


24 months ago

Quote:
Originally Posted by My Small Puppy View Post
Unfortunately what's most important to me isn't even on there. The only thing wrecking it for me are crybabies who can only play one champion or lane, or else they feed, afk, etc. Or the instant locking ******s who then qq about everyone elses champ picks.
We have been thinking about ways to reduce the frequency of these situations; it is especially bad in Ranked Mode at the moment.


24 months ago

Quote:
Originally Posted by rivanor View Post
And try to do something about losing streaks. If I've lost 3 games in a row, put me in a game where I've got a better than 50% chance of winning (maybe even better than 60% chance). Give me some wins mixed in with those losses while you are trying to figure out what my ELO ought to be. Give people some _hope_!
I have been thinking about losing streaks a lot, but I do not like the solution here which is something other games have done in the past.

By giving you a ~60% win chance or better, we are intentionally giving the opposing team a ~40% win chance. If you extrapolate this across the entire LoL population... that means almost any given game could be a team who is getting a 'freebie,' and the opposing team is getting forced into an unfair game. I never want to be in a matchmaking situation where a team is intentionally put into a 'losing' scenario.


24 months ago

Quote:
Originally Posted by Alan MQ View Post
Lyte can we look into the elo of the player as well as their number of wins? A player have 500 wins 1200 elo is obv higher skill/experience than a 200 wins 1200 elo player. Why you would put these two player together? Clearly speaker the one who have 500 wins play 2.5 times more than the other. This is an unfair situation of the current making making. How could you said you"" never want to be in a matchmaking situation where a team is intentionally put into a 'losing' scenario""? this is what is happening now.

Me 320 wins play with a 180 wins and 220 wins friend premade, and our enemy always has around 1100 wins player and the other is 300 + wins player. Do you think this is a fair situation that you never want to put player in? At least it always happen on me play with my friend.
I am doing more analyses on this, but so far we have not seen that "number of wins" is a super strong factor when doing matchmaking. I've mentioned this before, but you probably do not notice all the times you beat players who have 1000+ wins.

To approach this problem from a different angle, if Elo was more accurate, than it wouldn't matter if you were 1200 Elo with 25 wins, or 1200 Elo with 200 wins because both are accurately judged to be 1200 Elo. The interaction here is that the more games you play, the more 'confident' we are that Elo represents your current skill. To solve the core of this problem, we have to make Elo more accurate and improve the system so that fewer games are required to have higher confidence in your Elo.


24 months ago

Quote:
Originally Posted by AdmiralWarron View Post
The adjustments dont need to be huge. Example: When somebody joins a queue every won game in his history of the same type adds 4 elo and every loss detracts 4 elo for the purpose of matchmaking. Somebody who has a bad string of games will be rated up to 40 elo lower or higher. This doesnt affect the win chance of somebody, it only gives them other types of games faster
What you are suggesting here is changing the K-factors and how they change over time which will directly influence what Elo you gain/lose in a game. Yegg and I have been talking about this and right now we are figuring out what the right K-factors should be.


24 months ago

Quote:
Originally Posted by Imclueless View Post
Hey, Lyte. Can we get a temporary fix for the transition to Ranked?

I just saw a thread on GD that said: If you aren't playing to win, don't play Ranked.

Sounds pretty reasonable, right? It was downvoted. And people were insulting the OP for being a "tryhard". You said that you want Ranked to be where people go to try and win. I believe your words were "bring your best every game" or something similar.

This is a problem, then. People in the community need to be made aware of what YOU want Ranked to be. They need to be made aware that Ranked is where people play seriously, and play to win.

You said you're going to work on the transition to Ranked. That's great, and I can't wait to see what you come up with.

But, we need a fix in the mean time. Because it's clear that your intentions are not making it across to the playerbase currently.

Would it be possible to get a "warning" message pop up when you try and queue for Ranked? It could say something like:

"Warning: Ranked play is intended for serious play from all involved. It is where our most competitive players test their mettle against one another. If you do not intend to play competitively and seriously, do not continue and instead play Normal games."

This seems (to me, who knows NOTHING about this, I understand that) like it could be relatively simple, and it could make an impact until bigger fixes come through.

If you've addressed this already, I apologize. Its been a while since I read through all of your posts, and I have a terrible memory.
Yegg is design lead of a team that is working on the competitive modes and figuring out the best ways to improve them. We talk a lot about the transition to Ranked and work on solutions together, but his team will be the one that implements the changes.

The actual work is not too difficult, but whatever we decide to put in becomes the 'official' stance so we have to take extra care in how we explain the game mode and whether we want to improve the experiences of playing Ranked first, or change the messaging first. This is where the complexity comes in as we try to model and predict the order effects of doing one before the other, or whether we must do both at once.


24 months ago

Quote:
Originally Posted by Well Whatever View Post
Being ignored suxx.
Cry-cry.

Well, at least I got +2 there!..

EDIT: Okay, thinking about it a bit more, I see that the silence is okay, since the response may look like would take an immense amount of time, or wouldn't befit for a rioter or somethign else... I'm just afraid that it is nice idea indeed and it gets forgotten. It's would be ok if it is not a good one, but I haven't seen a single counter to it, nor a -1...
I read every post, but am pretty choosy about responses

We talked about a similar idea a long time ago; Yegg and I were pondering whether solo queuers should be slightly higher Elo when getting paired with a premade of 4.

However, there are many concerns with using this idea in general for all team compositions. Who is the 1 player that is designated carry and slightly higher Elo? Does he know he is the designated carry, considering Normal Blind Pick orders are not Elo-dependent? What if he does not want to play a carry? If the designated carry plays a support, then the slight Elo advantage he has above his teammates is not enough to carry the team.

Let's assume the above issues are non-issues and resolved through design. Let's then assume the Blue Team's designated carry has higher Elo and in fact chooses a carry. If the Purple Team's designated carry also chooses a carry and they both meet bot lane, then maybe we have a 'fairer' match but realistically all we have done is slightly bumped the Elo of players in bot lane. If the worst player on the Purple Team decides to choose carry and is matched against the Blue Team's designated carry (and due to the Elo bump, is theoretically their best player), we have actually made the match far worse. Now, this feature is actually adding more noise to the system.


23 months ago

Quote:
Originally Posted by vox1st View Post
http://na.leagueoflegends.com/board/...4#post23946774
This is the thread I am referring to. Just ideas, not that I'm saying this is THE way. But the fact that nothing is done to rectify ranked issues and player behavior in general really bothers me.
We are actively working on fixing Ranked issues and player behavior issues! Regarding your comments on Ranked, I think there is a common misconception that some players are just not 'ready' for Ranked.

For example, let's say we have 4 tiers of players:
- Peasant
- Footman
- Knight
- Champion of League

Right now, players are suggesting ideas like "Get 300 Wins before you can play Ranked" or "Get 20 Champions before you can play Ranked." This is similar to saying, "Peasants should not play Ranked! They are not skilled enough and not ready! Make it so that only Footman and above can play Ranked!"

However, how does this solve the problem? If we do this then in the future, the Footmen become the worst players in the Ranked Ladder and arguably players will then suggest that Footmen should not be able to play Ranked. This cycle will just perpetuate over and over until only Champions of League can play Ranked Mode.

This goes against the very belief of competitive gaming: all players should be able to participate in Ranked Ladder if they want! However, as I have mentioned before, there are lots of things we still want to do to help transition players from Normal to Ranked Mode. But, I think restricting players from Ranked Mode is not the solution.


23 months ago

Quote:
Originally Posted by Cae vida loka View Post
Change the name of thread to "After Months with Matchmaking and Lyte"...
I think we are pretty on pace so far. It takes a bit of time to collect the data and run the analyses to identify where the problems are.

Then, we took some weeks to talk to our community about the issues we identified and get some feedback on potential solutions.

Now, we're coding up some of the solutions...


23 months ago

Quote:
Originally Posted by Epok2 View Post
I agree.

Lyte I heard that in SC2, the ladder is made of multiple 'stages' and once you reach one you cannot fall farther down than the bottom of it. What do you think about that ?
This is untrue; for example, if you are Gold in SC2, you can still fall down to Silver and Bronze.


23 months ago

Quote:
Originally Posted by mojodarkale View Post
I've said it before and you've even responded, but I'm going to say it again. I think player behavior is about 70% of the problem, not "bad" matchmaking necessarily (most of the time). I had three. THREE games today that were lost completely because one person didn't get the role they wanted, so they picked it anyway and forced players in other roles to compensate for it; this creates an annoyingly tense atmosphere immediately and everyone on the team just seems to hate each other.

A fourth loss today happened because we had some guy call top, but then he DC'd for the entire match. That's another issue entirely... but I beg you guys to find a way to make the people less horrible. Either the tribunal isn't harsh enough for extremely serious offenses in ranked, or it just doesn't spot/discern between what players are doing to create a horrible environment.

I've had several people this week say stuff like "I don't care about Elo," "Elo doesn't mean anything," "I don't care about this game anymore I'm just waiting for D3 to come out." I don't want these people in my ranked games. If they don't care to be competitive and do their best, then they shouldn't even push the ranked button.
I agree, player behavior is probably indirectly touching all these facets of the game negatively; but, tackling the challenge of player behavior takes time.

For example, we recently implemented the Show [All] Chat toggle, which has been very successful in curbing some player behavior. We are working on a lot more, but little features like these will start cropping up more often. In the past, many developers thought that player behavior was beyond the realm of our control--after all, how can we influence how players play the game? However, we are making a huge effort to design features for League of Legends that can actually shape and encourage positive behaviors... but because this is completely new territory, it takes some time.


23 months ago

Quote:
Originally Posted by Isamichi View Post
Lyte:

I wanna thank you for all your work on matchmaking to make it an even and awesome environment for all players! i hope that everything slowly smooths out since this problem has so many variables that are almost incomprehensible to us without the ability to mind-control things/electronics/chihuahuas/nyan cats.
Thanks

Quote:
A few concerns:
1. I know in normal, I typically queue up with friends and people that are considered lower elo. Please be merciful.
Actually, we are trying to make this a bit more lenient. We see now that players do enjoy playing with friends who are dramatically lower in Elo, and want to make this a smoother, more positive experience.

Quote:
2. How many pounds of tea have you gone through (and gallons of water flushed) for this huge task?
3-4 cups of tea a day... mainly because I am also working on a dozen other projects related to Player Behavior.

Quote:
3. Concerning the transition from normal to ranked. I know that a video or a message to all players telling them to "try their best" isn't bad. However, I feel that a good majority of players will ignore that or find it annoying. What do you have in mind to encourage players to "tryhard"-er?
Yegg is thinking about this a lot lately and might chime in. I think we just have to think about the players and their motivations for playing the game. If Ranked has rewards that are very desirable and require that you remain positive, I can predict that players will want the rewards more than they want to be toxic.

For example, as a random thought experiment, consider if playing Ranked Mode had special rewards that were 'lost' if you were toxic and/or punished by the Tribunal. You could not get these rewards anywhere else and the rewards were cooler than the Silver/Gold/Platinum medals or forum icons. I am willing to bet that a percentage of the currently toxic population would behave better in an attempt to grab these rewards.

Quote:
4. I can't believe you read through all of this. Stop brain-washing your lackeys to read them and wireless-ly send them from their brain to yours.
I actually just spent a week in Asia meeting and talking with fans on our other platforms We do this because we want to make awesome for our fans.

Quote:
5. I know that lower elo players simply don't have the "game sense" (an understanding of objectives, values of things, etc) that higher elo players do. Is it possible to have a battle training-like thing with quest to help educate players concerning objectives and game sense?
This is possible, Brackhar is very interested in these sorts of ideas.

Quote:
6. If I slay baron in game and take his santa hat, does that mean i'm slaying you?
Yes, but it's a good thing I am just a few lines of code and some pixels of art. Riot just re-creates me


23 months ago

Quote:
Originally Posted by Cae vida loka View Post
Lyte I dont wanna sound rude but by judging your twitter picture you seem to be so young

what makes you think you understand human behaviour?

are psychologists perfect human beings? my mom always told me that they cant even fix their own problems
Well, I have a PhD in Cognitive Psychology and Neuroscience, which means I studied human behaviors for at least a few years. No one fully understands every facet of human behavior yet which is why it is still such a popular field of study; but with data, with experiments, and through collaboration with many brilliant people like Yegg, davin (another PhD!) and Zileas, we can build confidence that we are always working in the right direction.


23 months ago

Quote:
Originally Posted by Darkwahn View Post
Games shouldnt start if someone in the lobby has afk'd. No one wants a 4v5, and no one wants someone to pop in with a random champ they've never used before.
We agree with you. Rjcombo is working on some potential ideas here.


23 months ago

Quote:
Originally Posted by Imclueless View Post
I know you said in response to one of my previous posts that you don't want to talk about future projects.

But, I'd like to know, do you at least have ideas for dealing with negative behavior that the Tribunal is blind to? There are a million ways to grief your teammates without it being obvious in Tribunal cases. I don't need to know what those ideas are, and I know you can't tell me.

But, it'd be nice to know that the more subtle means of griefing your teammates won't be devoid of consequence forever.
I just presented a report recently that while the Tribunal has been wildly successful, we still need to design and build more features to truly tackle player behavior problems.

In our research, we generally find that the League of Legends community is awesome; however, everyone has different 'tolerance' levels. For example, a player might have had a bad day at work. He might decide to play League a bit to relax but because he is so exhausted, he plays terribly that night. Then his teammates yell at him for feeding, possibly scream over [All] chat for the enemy team to report him... and he just goes toxic. The player then screams back and curses and is promptly reported for offensive language.

Millions of players play League of Legends everyday, each with their unique personalities, lives, and circumstances that brought them to League. The challenge now is, how do we prevent the above toxic situations from happening? How do we shield our players so that even in their worst days, they do not become toxic?

I look forward to the challenge


23 months ago

Quote:
Originally Posted by Cae vida loka View Post
Lyte I dont wanna sound rude but by judging your twitter picture you seem to be so young

what makes you think you understand human behaviour?
The nice thing about science is that you don't have to rely on just what you know--you draw on a giant body of knowledge from everyone that's worked on similar problems before. It's super collaborative on multiple levels (interpersonal and via things like journals).

SCIENCE!


23 months ago

Quote:
Originally Posted by Imclueless View Post
What about the people that just enjoy griefing others? We all know they exist. That's why you see people insta-lock Eve and say "Have fun with your free loss" in Champion Select. It's an unfortunate, but very real, situation. Hell, I saw a post on GD today about a guy who's been intentionally losing games for two weeks now, just because he can.

If they all were obviously griefing, it wouldn't be a big issue. It would be unfortunate, but they would get banned by the Tribunal eventually. But, these people are evolving their strategies. They've learned that just feeding 0-30 gets them banned. Griefers are becoming more subtle. Intentionally losing your lane, and watching it snowball to a loss. Refusing to gank as a jungler, when the opportunity shows itself, and watching your team lose. My favorite teammate danced in the fountain for the last 5 minutes of the game, because we wouldn't surrender. He pretended that he was still playing, in the chat, so the Tribunal has zero evidence of his misbehavior. He played the victim, claiming we were a premade that was abusing him.

Are there plans to win this arms race? Because the "best" griefers have beaten the Tribunal.
The Tribunal is constantly evolving as well. Every so often, I'll randomly sample a bunch of forum cases or reports where players are 'subtly' toxic and knowingly skirting what they believe to be the edge of the Tribunal's reach.

However, because the Tribunal constantly evolves and because our community decides the standards for player behavior, these toxic players have actually been punished fairly accurately by the Tribunal still.

I believe that the Tribunal will always be effective for identifying the players that are intentionally griefing other players or the players who enjoy being toxic. However, to tackle other player behavior issues we will need to do more than the Tribunal.


23 months ago

Quote:
Originally Posted by Deeve View Post
as for answer #5 on the poll that starts this thread, a detailed tutorial for ranked play when you hit level 30 (or just before you try to enter a ranked game for the first time) might be beneficial.

for champion select afk's randomly selecting a champ for the player when half of their time has run out can at least give teammates an opportunity to balance their team in blind pick if the champ select does not count as a dodge. for ranked play? I think that at the very least, if the #1 person on the team is afk the choice of bans should go to the #2 player and the #1 should go to the bottom to be last pick (in case he comes back so he doesn't have a random character)
If the players want it, I am sure Phreak will do it (I am going to get clubbed for this statement).

As I mentioned a few pages back, Rjcombo is working on some ideas that will attempt to solve AFKs in Champ Select Lobby.


23 months ago

Quote:
Originally Posted by HayApples View Post
If you compare this matchmaking system against more freeform systems (CS, Tribes, TF2, etc), the freeform systems have one critical advantage... player control.
I want to highlight this point of your post. I completely agree that our players deserve more control over their experiences in League and we are trying to figure out where we can offer players more control without interfering with the integrity of the game or hurting the player experience for someone else.


23 months ago

Quote:
Originally Posted by LLod View Post
Lyte
i just spent a ridiculous amount of time reading through every one of your posts, up or down voting alot of them.
do you collect data on how players vote in the forums?
for instance if one player voted up on one idea, but down on the other.
i know the forums are the vocal minority, but im a BS in math/economics and i get excited about this sort of stuff.
Indeed, our forum bosses collect and analyze data like this.


23 months ago

Quote:
Originally Posted by Grind The Ocean View Post
Changing it mid season would throw off quite a few numbers...
Not necessarily, many changes can benefit all queues or just Normal queues and improve experiences without doing crazy things to Elo.


23 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
It's just a pity that the voting system is shutting down conversation :\ having a system that automatically closes a thread due to downvotes is horrible. Say something that the community doesn't agree with, even if you provide substantial evidence and calm reasoning to back it up? Sorry, still get insta-downvotes and closes thread. Hell, even if the majority of the community WOULD like the thread, if the right # of people see it first and downvote, insta-closed.

Terrible terrible system

Gonna repost this
I know a lot of players loved the SC2 Challenge System and these are ideas we have been considering. Designers like Brackhar, and researchers like davin have been interested in these types of ideas.


23 months ago

Quote:
Originally Posted by Clerigon View Post
hey lyte im sick of play last pick every single game, i quit normal games for this, i queue again 1 month later and is the same ****...
Are you playing Normal Draft?


23 months ago

Quote:
Originally Posted by Doonhijoe V View Post
That begs the question: Is anything going to be done about that in normal draft? I know it's suppose to be random, but it isn't. Is there a fix anytime soon?
There's a fix going out soon.


23 months ago

Quote:
Originally Posted by SomaLOL View Post
Please, Please Lyte, remove the new champions in ranked... THIS is what happen....

Varus went 0/9 with 39 CS
This is a 1300 elo game but still.....

I mean, disable them for at least 1 week.... like 300 posts back there we were talking about not letting people use free week champions in ranked because this happens, they are playing champions they are not used to in the "tryhard" game....
New champions should be disabled from Ranked at least for 1 week. no one can jump in a Ranked game and say.... "Oh I completly dominated my first and only game with this new champion in bot game, he is OP I will easy win"

EDIT: Also.... we know new champs are really volatile. I know Riot try to do their best balancing champions, but sometimes they come either OP or UP, so having an insta OP champion or UP in ranked is not good either, 1 week off ranked games should be good to avoid that.
Free2Play Champions are being removed from Ranked Solo/Duo Queue soon unless you own the champions. We may revisit this policy in the future, but we are going to test this out.


23 months ago

Quote:
Originally Posted by 1wheel View Post
Why? Getting a win with all the champs is gonna be impossible now.
You'll be able to use free to play champs in Ranked Teams. If you have a team who's cool with you trying for this, you can still do it.


23 months ago

Quote:
Originally Posted by 1wheel View Post
Why? Getting a win with all the champs is gonna be impossible now.
In brief, we are exploring various hypotheses we have about sources of noise in Ranked Ladder. Removing F2P champions ultimately raises the barrier of entry into Ranked Ladder slightly (such that you must now own 16 champions to play Ranked); secondarily, players tend to be more experienced with Champions they own, so that might help reduce the fluctuations in skill we are observing.


23 months ago

Quote:
Originally Posted by diegoSven View Post
About free2play champions: You should consider to disable champions for players that haven't played enough games with them.

For instance, Vlad is free right now, and I've never tired him. I should not be able to play with him in ranked. On the other hand, I've played Malphite a considerable amount of times. I think I should be able to play him in ranked, even though I don't own him.

TLDR: If you are gonna disable free2play champs, for each player disable only those champs that he has not played with enough times.
We considered a "play X games to enable a champion for Ranked" approach, but that was more difficult to implement. We may re-visit that approach in the future, but we wanted to test out this solution first and analyze what happens.


23 months ago

Quote:
Originally Posted by Nyekon View Post
@Lyte when do you plan on doing this? within the next few months or at the end of the year?
Things are starting to trickle in. Removal of F2P Champions in Solo/Duo Queue and the fix to Normal Draft Pick orders are going in first.


23 months ago

Quote:
Originally Posted by Warewulf View Post
If you do this, I hope you implement the whole "picking for the team" mechanic, because it's hard enough to do champion swaps that I often have to resort to free week champions to find something we have in common. I like swapping because it lets me reserve my pick for later, especially if I want to play something that is easily countered. This worries me, slightly.

That said, I like the direction but not sure I like the angle.
We will keep a close eye on this change and see how it plays out. Many of the concerns pointed out by our players were discussed thoroughly internally, but we still want to move in a direction where we can constantly experiment and try new things. If it works, great! If it doesn't, we will go immediately back to the drawing board.


23 months ago

Quote:
Originally Posted by Lo4fer View Post
Want to reduce the noise and fluctuations in skill observed due to matchmaking?

Restrict the elo gap between duo-partners!

Quoting myself from another thread on the matter: Ranked duo queue - Huge game breaking imbalances?
This is something we are exploring. I am going to tweak how Duo-Queues are processed in the matchmaker first--if that is not successful, I may consider some restrictions to Duo-Queue ranges in Solo/Duo Queue.


23 months ago

Quote:
Originally Posted by Devare View Post
Lyte you guys need to remove the ability for people to duo queue in ranked with people that are a MUCH higher elo than the person who is just starting. I've seen players like Athene and Destiny and now Neilyo basically duo with 1900-2.3k elo players and just climb the ladder to 1500-1600 and get out of "nooby island" without knowing the game or 90% of the champions.

Duo queue really should just be listed under arranged team and left there because you are choosing at least one of your own team mates. On a team of 5 you can only have 2 duos meaning at least 1 person cannot choose their own team mate. Also chances are there is going to be only 1 duo in your group of five meaning three others cannot choose their team mates. It just is not fair that the duos can hand pick their 1 team mate and the others cannot. It should be so that no one is allowed to hand pick their team mate in ranked because it skews the fairness of a 50/50 shot of winning in reality vs numerically.
We are working on many of these points about Duo Queue in Ranked. For now, we are going to tweak the matchmaking when people of greatly different Elo queue together. If that does not have a strong effect, we may consider restrictions to Duo Queue in Ranked.

We are also working on a few others things related to Duo Queues in Ranked, but not ready to talk about them just yet.


23 months ago

Quote:
Originally Posted by Ginga View Post
Bugger. It's like every good idea can be shot down by unseen factors.

So...any discussion regarding the instant reward of raging like a chimp(the reward is blowing off steam) vs. the subtle and sometimes unseen reward of playing nicely?

Like offer monthly reward for those who received absolutely no successful report (1 game IP boost, or 1 RP? Would 1 RP per month for those who play consistently nice hurt profits? That's 12 RP lost per year)
Conceptually, you are on the right track and these are ideas we are bouncing around. We'd like there to be better (and more immediate) feedback in the game in general because that's how players learn what is positive versus negative behaviors.

Re: your first point, it is difficult on the forums sometimes when players suggest the same cool idea many times and call Rioters idiots for not thinking of it or implementing it already. Truth is, in many case the ideas have been discussed thoroughly in the past and there are still complicated details that need to be worked out before the idea is feasible.


23 months ago

Quote:
Originally Posted by pkzadb View Post
It seems like the simple way to handle this is to make the starting Elo 0 and allow negative numbers. Is there a reason this wasn't done? Perhaps a number gets divided by Elo at some point?
Negative numbers are depressing to look at


23 months ago

Quote:
Originally Posted by mojodarkale View Post
So Lyte, I know you've talked about it (or some other red), but do you guys have any plans to deal with the plays-1900-Elo-Vayne-but-1300-Alistar-guy? I can't think of any other reason why there would be huge mismatches in skill at every Elo other than that certain players only have one, two, maybe three champions they play at that level, and play them not to fall back down, but truly suck at other roles/champions. I don't know if there could ever be a solution to that.

I know I'm not good at every role and, even in the roles I think I'm pretty good at, my champ roster is kind of low. I'm exactly the kind of player that probably plays 1800-1900 level on certain champions and 1300-1400 level on other champions =/.
Yes, we're thinking about ways to combat this. This indeed is a reason why we see such varying skill levels when players are supposedly a certain Elo.


23 months ago

Quote:
Originally Posted by RyuuMasken View Post
one could reason that currently, sub-1k elo has a similar effect. loss of a digit can be pretty impactful, even if it's from 1000 to 999
This is a fair point, but it's the lesser of two evils I believe. Alternatively we could just shift everyone to be 1000-3400 I suppose...


23 months ago

Quote:
Originally Posted by Sylexus View Post
Also, what was the change you guys had in mind for queue dodging in ranked?

|s it going to be impossible for me to climb back up to 1200 now?
How come queue dodge penalties would make it impossible to climb back up to 1200?

We were considering removing the -Elo on the penalty, but doing something else instead.


23 months ago

Quote:
Originally Posted by Sylexus View Post
Sorry, I should have been more clear.

I wasn't referring to the dodge penalties about making it harder for me to climb back up, I was referring to the duo queue changes (should have been explicit). If I queue solo right now it literally takes 2 hours for a queue to pop.
I'm not at the office right now, so do you mind providing more details? Are you really low Elo right now, and therefore the queues are extremely long? Did you queue dodge and drop -Elo down on purpose, or did you play normally and end up with a negative Elo?


23 months ago

Quote:
Originally Posted by Devare View Post
Well instead of a prisoner island idea where people just get queued with other toxic people or 'prisoners'. Why not just make a lock down mode? It works similarly to a suspension, but instead of suspending for 3-7 days you 'lock down' their queue that they were being toxic in for 3-12 hours depending on what they did. I know this might cause them to go on a smurf and possibly create havoc for lower levels, but it could be worth a shot.
This is basically what the Tribunal does--after building a case on the toxic player and determining the player as guilty, the Tribunal then hands out punishments ranging from warnings to permabans. Unfortunately, it takes a bit of time before punishments are handed out, and we are working to speed up this process.


23 months ago

Quote:
Originally Posted by Bakashinobi View Post
Could an alternative be a chat lock? Yes, that diminished their ability to coordinate, but it may make them consider how the coordinate rather than raging. They'ed probably need to be banned from Ranked for the duration of the chat lock (or not, they might play better if they can't spend time raging).
We are considering ideas like this, but a concern might be that if you get a player who doesn't talk, players might assume that 1) he was a banned player and thus toxic, or 2) he is refusing to communicate. This idea might potentially outcast these players even more, and the net might capture some innocent players who are just quiet or more passive.


23 months ago

Quote:
Originally Posted by Sylexus View Post
Sorry once again for my lack of information.


I played normally from 1158 to 1350ish Elo. I then proceeded to lose about 16 games in a row (putting me back at around 950ish) then decided to give up and dodge my way to 0 so I could climb myself back up.
I finally hit the 0 mark (-5 to be exact to my knowledge, the problem being that there is NO number for my solo queue Elo even when I look at it in my account) but the problem now being that if I want to solo queue my way up without duoing, then it takes about 2 hours to finally get in a queue (and usually someone dodges when I do causing it to take even more time). I was curious if the duo queue changes you guys were considering were going to affect my ability to partner up to dig myself out of this pit.
No, the initial duo queue changes won't affect your ability to partner up and get some quicker matches; however, you shouldn't intentionally queue-dodge down to 0 like that


23 months ago

Quote:
Originally Posted by 0xFADE View Post
Have you considered lowering the entry elo from 1200? People do generally gravitate to their natural elo. People who play around the 1200 range are constantly assaulted by unskilled people. If the entry elo was say 600 or something then perhaps people in that range are more likely to play with some random person who is much better then the other way around.

Now that I mention that alternate I can see where there could be a problem. Though generally being dominated by one person is not as bad as having "that one guy" on your team.
Check a couple posts back, shifting the starting Elo actually doesn't fix any problems!


23 months ago

Quote:
Originally Posted by ColossusCrusher View Post
It's just a thought, but instead of having Newbie Island in Ranked, why not just have new players into Ranked come in at their hidden Normal elo, which then stops determining what their Ranked elo would be? That way the people with +100 wins come in at ~1900+ and the people with -100 wins come in at ~500 or so without needing to have placement matches to "determine" their elo.
We have talked about this a lot, but the main concern is that for most of our players, Normal Mode should not affect their Ranked experience. Ranked Players, or aspiring Ranked Players tend to use Normal to practice new champions, new builds, or new strategies--we should not penalize players for using the mode to 'practice' for Ranked.

Normal is less serious, and is the core game. Making it affect Ranked in any fashion potentially makes it more serious than it should be.


23 months ago

Quote:
Originally Posted by Eldaris Shadusk View Post
To me it says, go by all the 450 IP champs to fill a roster, then still play who you want.
We will know if this happens and adjust accordingly.


23 months ago

Updated main thread.

--- Completed ---

1) Free to Play Champions in Ranked Mode [5/22/2012]

Quote:
Originally Posted by Yegg View Post
Summoners!

Ranked Games are about trying hard and competing to the best of your ability. Meanwhile, Free-to-Play champions let you try out a new champion and see if you like it. These are really conflicting goals, so we wanted to investigate disabling the Free-to-Play rotation for Ranked Solo/Duo Queue.

We found some interesting statistics:

  • If a player on your team is fielding a champion he doesn’t own, you’re 8% more likely to lose the game.
  • A teammate fielding a champion he doesn’t own is 62% more likely to leave the game.
  • If a player in your game is fielding a champion he doesn’t own, the game is 34% more likely to result in a player being reported.

That’s a big deal in such a competitive environment, so we’re making a change to keep the Ranked Solo/Duo Queue as competitive as possible. We’re removing the option to select champions you haven’t unlocked in the Ranked Solo/Duo Queue as of May 22nd. If you have a champion unlocked and it’s on the Free-to-Play rotation, you can still play that champion.

We’re leaving the Free-to-Play rotation intact for Normal Games because they fit with the spirit of Free-to-Play champions. In Ranked Teams, your teammates should already be aware of your intentions to play a champion you don’t own, so we want to leave you that option.

2) Pick Order Bug in Normal Draft Mode [5/22/2012]
This bug will be fixed in the May 22nd patch.


23 months ago

Quote:
Originally Posted by Eiales View Post
Have you considered something like introducing players at 900 (arbitrary) Elo and awarding 10 "free" Elo every game until they have 30 ranked games played - with a 50% win-rate they'll still be roughly around 1200 Elo at that point, but will not have played their first ranked games with players who have 500 ranked wins and losses.
Actually, we played with this type of mechanic in the past


23 months ago

Quote:
Originally Posted by Phaceroll11 View Post
Super low priority thing that is one of the smallest problems with ranked gets implemented... In fact its not even really a problem. If you suck at a champion you shouldn't pick it, whether you own it or not. Its basically the same concept as counter picking yourself, if you lose because of doing something as stupid as this then you deserve to lose elo. If your "true elo" should be more than your current elo you should be at least able to do "ok" on a champion your are unfamiliar with.

People leaving games and games starting as 4 vs 5 are the biggest problem. All you would need to do is like give the team of 4 (assuming they lose) a loss forgiven or atleast a less severe elo penalty, for example re-calculate their odds of winning 4v5 and take/give them elo based on the % chance to win. Which is still effectively the same system as the current one except it could be recalculated in the post game lobby. Even if it took an extra few seconds to load the lobby, I'm sure every would be fine with it.
We have development teams that consist of many members, and they are working on all the things in the list. Just because one item is finished first and goes Live, does not necessarily mean the others are not in progress...

Take a look at Yegg's statistics on the issue as well. F2P Champions in Ranked have a significantly negative experience on outcomes.


23 months ago

Quote:
Originally Posted by Clerigon View Post
i never trust in those closed statistics, where you guys collect that info?
I'm not quite sure where you are coming from? We collect all this data from the game directly and it helps inform our decisions.


23 months ago

Quote:
Originally Posted by Clerigon View Post
but how, do you pick a lot of player and ask them?

I dont have a high amount of champions and i use the f2p champions and i only have a few leaves for dc or windows blue screen. i dont think people use the excuse of "omg i suck with this champion so im going to leave"
I think you might have a misunderstanding somewhere about our data collection or how research works. We collect data directly from the game--this means we can look at every single game played on League of Legends where a F2P Champion was used and every game where a F2P Champion was not used and measure things like the win percentages when a F2P Champion was used or when there was a leaver. We can look at any data we possibly want because we record all the data live directly from the game servers.

You personally might not leave a game because you play a F2P Champion, but you are 1 person in a game with millions and millions of other players. When we look at every single game in League of Legends and look at every single player, we see the statistics Yegg posted.


23 months ago

Quote:
Originally Posted by DemolitionDean View Post
In addition to Free-to-Play champions in Ranked I don't think the brand new champion should be available in ranked until 2 weeks after the initial release.
1. because the whole point of taking free-to-play champions out of ranked is to make sure people have ACTUALLY PLAYED THE CHAMPION BEFORE and know what they are doing, the same issue happens with new champions, there's no way you should be comfortable enough with a brand new champion to play ranked the day he/she comes out.
2. because the bugs and balance hasn't been totally worked out yet. As much as we'd like to think riot releases perfectly balanced champions, it simply isn't true, every champion needs a little bit of tweeks and many of them are a bit too strong upon release which could create a real issue of imbalance in ranked games. (especially since people don't know what to expect out of a new champion yet.

Hope you see this Lyte and take it into consideration!
A lot of players think this is a good idea, but unfortunately the data doesn't seem to show that it makes a big difference.

For example, a reason we don't see a difference might be because the average player tends to play a newly released Champion in Normal Mode first before trying him out in Ranked Mode.


23 months ago

Quote:
Originally Posted by RexSaur View Post
If it was that simple they would have already done it.

Lyte,so a annoucement about the no free week champ in ranked just happened...Any news about the normal draft picking order fix?Is it comming with the next patch?
Check the patch notes next week.


23 months ago

Quote:
Originally Posted by Isphus View Post
Why not just make everyone start at the top?

I know that dropping 400 doesnt feel as good as climbing 1600, but it would solve a lot of issues.
This way to get to your deserved station all you have to do is not to lose (as soon as you lave the "stomp zone" where the top players crush you for a couple games).
Unfortunately, starting at the top doesn't do anything either. Basically, starting position doesn't actually solve any problems, it just shifts the Elo distribution.

Quote:
Maybe a simple way to do this would be to give the player the ability to turn his normal-ranked interaction on and off.
Lets say i just turned 30 and want to play ranked, i'd start ranked at my current normal rank and go from there. And if i am at a high ELO and wanted to test something on normals i turn it off and go do my crazy stuff.
I know this might be hard to implement, but i just cant see the downside (although im sure you will find a very obvious one i failed to see).
There's several downsides, some of which you mention. One, we can add toggles and switches for everything but how do you educate the player on what the toggle does? This actually is not a trivial problem for such an arbitrary toggle. What happens to players who toggle it one direction, then forget about it then complain that they screwed up their chances in Ranked seeding?

How do you separate the players who will abuse the toggle? For example, some Level 30s will intentionally premade 5s, use competitive strategies, and just stomp their way through Normal games until their Normal Elo is high, then switch to Ranked. They faced completely different competition in the process and actually relied on 4 other people to up their Normal Elo when this would be impossible in Ranked Mode. In fact, our data suggests that 1800 Elo in Normals is actually far off from 1800 Elo in Ranked.

Then you have a subset of players who don't really understand the system and play Normals for several hundred games. They don't realize there is a toggle, but when they finally start playing Ranked they are confused why they are seeded lower than everyone else. Basically, toggles for Normal Mode just means you might as well have two modes because the intention of the players are completely different.


Quote:
What if you made it so everyone can select a champion the moment they enter a lobby, but cant lock in until its their turn? That way everyone can see each others intentions and start discussing from the first moments, saving their valuable time and keeping the allies from banning your main.
We are talking about something like this for Champ Select Lobby. Basically, a way for players to show their desired Champion choice before it is their turn.

Quote:
(this was about "silencing" players as a form of punishment while not influencing their ability to contribute to the team with their skill)
If a player is just quiet, the moment someone mentions "he didnt say anything, he must be silenced" he would just answer "nope" and the problem would be over.
And if this could be a worse punishment than intended, maybe the answer is to just think twice about when to use it, instead of just not using it.
This, unfortunately, is not as simple as that! Many players tend to miss text a lot, and the lower level you go, the more common this is.

Quote:
Also, i'd like to propose and idea i've seen on another thread: what about a "never match me with this piece of **** ever again" button? It would be very satisfying after a trolled match to know you will never see that guy again, and if someone would troll too much, eventually they would run out of people to play with (because everyone would have used it on him). The only problem i see would be people using this on poor unskilled people, but even then the ranked system would never match them anyway.
This would not be feasible in Ranked at all; players perceive other players as 'below the skill level' or 'unskilled' all the time when it is not necessarily accurate. What if players use this on GOOD players, to guarantee that they will not play against them?

In Normal Modes, playing the same player twice is actually very rare unless you immediately re-queue after every game. However, we are considering some ways to avoid players you perceive as toxic from re-appearing in your game in Normal Modes.


23 months ago

Quote:
Originally Posted by Mast3rRitual View Post
Lyte- first off that you for all the hard work you and your team has done to make ranked a more competetive scene.

What is your opinion on removing the option to duo in ranked solo que? I really think it provides an unfair advantage and its removal would help players reach their "true" elo.

(I know this only got 3 percent in the poll but I think it is a HUGE issue)
More players duo queue in Ranked Ladder than solo queue, so we have to take that into account. We're tweaking the numbers related to duo queues in Ranked, so we'll see if that helps. Otherwise, we might start with some light restrictions on duo queues in Ranked.


23 months ago

Quote:
Originally Posted by Triscadeca View Post
Lyte, while you are here. Just want to point out something in matchmaking. My last match, I played Leona(2/7/9), with Jax, Riv, Ez, and Renek. Our Jax had 1000+ wins, I was next with 241, no one else had over 200. No one on the enemy team had less than 525 wins. Now, I understand that the numbers average out, but one ZOMGWTFawesome player can't reasonably carry 4 moderately experienced players against 5 people with at least triple their experience in any one-to-one comparison. Thanks for your time

edit: could matchmaking be tweaked where one player can't skew the numbers like this?
We are looking at data at the impact of # of wins, but so far, we haven't seen that # of wins matters significantly more than just Elo.


23 months ago

Quote:
Originally Posted by Triscadeca View Post
I forgot to clarify, this was a normal match
Normal Matches have hidden Elos we use for matchmaking


23 months ago

Quote:
Originally Posted by Dex7ter View Post
here u go
This is known as confirmation bias. You took screenshots of the times you lost matches and checked # of Wins and noticed huge discrepancies... if you took screenshots of every single game you have ever played and looked at the # of Wins, you might see the same trends we are seeing when looking at millions of games.


23 months ago

Quote:
Originally Posted by CrazyGoNuts View Post
Lyte,
What percent of games being toxic do you consider acceptable? I have been tracking all of my games for almost a month now and its currently around 25% - that is, 1 in 4 games does not have 10 players that are all playing to win. Mostly it is disconnects / AFKs / ragequits - I'm only counting intentionally trolling from people who basically admit they are doing it (only about 4 or 5 games).
It's hard to say a number. We will never be at 0%, so I have no false hopes of hitting that number. However, there is always more we could do to lower the number towards 0%. If the number was 25%, that's too high in my opinion.


23 months ago

Quote:
Originally Posted by Cryxx View Post
Lyte,

Have you thought about not adjusting ranked Elo at all for the first X amount of games. Say 20.

So all players that are within their first 20 games keep getting ranked against one another in a random seed from that pool. Yes it could create an issue if it's an off hours time, but you could warn people.

It does keep track of their adjustment, it just isn't used for match making.

Then after the 20 (or whatever value you choose for X is) it actually moves that player to that ranking. Then it moves as normal.

Eliminate the 50-40 whatever boost you give at the start of ranking. If someone is a 1900 Elo player they will get there either way, so it takes a few more games.
This doesn't quite pan out. If a player is legitimately 1900, then playing 20 games against other provisional players means that he will most likely dominate over 60% of his matches which is a very poor experience for the other players.

Let's say a player reaches 1900 Elo in 100 games in the current system. With your idea, instead of going from 1200 Elo to 1900 Elo in a relatively linear fashion (i.e = he slowly climbs up towards 1900 Elo), he will now dominate 1200 Elo players for 20 games before he leaps to 1400-1500, then slowly climb from 1500 to 1900 over X number of games.


23 months ago

Quote:
Originally Posted by Silent Reaper View Post
@Lyte
Dear lyte, in the past few weeks I have been consistently being paired up with players WAY BEYOND my skill level, I used to be 1900 but over the course of the last 2-3 weeks I've been paired up with people WAY BELOW my skill level who are 100-200 Elo below me, this is really frustrating and is making me lose alot of games.

Can you please let the Match Making system NOT pair me up with people who are more than 100 Elo difference from me?
The main problem here is there are simply very very few players who are in the 1900+ range, so unfortunately players at the very extreme highs or lows will get larger ranges in Elos in their matches.


23 months ago

Quote:
Originally Posted by solidfake View Post
Lyte, I have a request.
Once you are done with the ranked improvements, could you pretty pretty please also take a loot at the normal draft matchmaking?

I usually play normal draft alone because I want to use it for train (which is atm not possible anyway), so usually every single game is me getting matched with a three man or 4 man premade.
The problem is, that this means that I'm also matched against another 4 man premade with a solo player. The issue is that sometimes the discrepancies between the two premades are just soooo high it is absolutely no fun to play for me and just a pure massacre.

Like last game, I jungled, and every single lane of my team lost before the 5 minute mark. Top lost by simply getting counterpicked, bot lost by just playing horribad and mid lost by getting 5 times ganked in a row and not changing his playstyle.
And then I'm there alone, trying desperatly to carry three lanes at once, while the other premades just shows superior teamplay while mine shows superior failure.
How is that supposed to be even close to any matchmaking?
Seriously, the premades add so much noise to it, that you inevitably have a cometely rape game every 5th game.

I don't say that you should remove premades from normal draft completely, but for gods sake, please find a solution to these discrepancies of the premades. This is no longer matchmaking, this is pure dice-rolling, and as a frequent normal-draft player, I would like to have my games actually balanced and not a pure rapefest like it would be 4 level 30 summoners against 3 level 2 summoners.
Many of these matchmaking changes actually affects all game modes, not just Ranked


23 months ago

Quote:
Originally Posted by enfabled mage View Post
when are these matchmaking changes coming because i would much rather a balanced game and wait a few extra minutes or so in que than have to wait for a limited amounted of time for an in balanced game.
2 are going in next patch, I will update when I have more information on the rest.


23 months ago

Quote:
Originally Posted by ZallKlos View Post
Anyways to fix Ego Ignaxio and Sauron? They are the top NA normal Elo for Summoner's Rift and Dominion. They are kinda punished for being good.

Ego can easily spend an hour in queue while queuing by himself.
Unfortunately the best players are in the game are going to have longer queues; this is simply due to the matchmaking pool being smaller for the top players in the game.

I would recommend them find a team and try some Ranked Team matches!


23 months ago

Quote:
Originally Posted by Noxialis Rex View Post
I have read almost all of these, and this is the first post you've made that I don't feel is entirely fair. Sauron plays dominion as it is his preference, and Ranked obviously doesn't yet exist for it :/. I sincerely doubt every "high-elo" dominion player has an average of 30-40 minute queue times.
Almost every high Elo player faces similar queue time issues.

Quote:
That said, I am sure the matchmaking algorithm could be tweaked ever so slightly for the top .5% of normal queue players to allow the people at the very top to maybe half their queue times. This would need to be done carefully, and would take man hours to satisfy what is literally only a hanful of the entire LoL population. However, they are no doubt valuable players who contribute positively to the entire community, and perhaps deserve a little love.
Let me flip it back to you then. How do you fix this without the top players in the game playing 1700 level players and crushing them repeatedly with 80%+ win odds?


23 months ago

Quote:
Originally Posted by Sam802 View Post
GOOD START, after many of us (including me) asked for "no free champs in rank", u guys finally did it, but why take that long to actually do it?

Another question, why u guys don't see what need to be done first?
An Ashe art upgrade??? Come on man, it is good but the old Ashe isn't killing this game.
Matchmaking is killing this game, why u guys spent time on upgrading Ashe?
I understand u guys has different departments, but Matchmaking, trolls, afker, etc are the priorities, and need to act asap.

Can't u guys see those problems are getting out of control? People are leaving......

Thx for the Ashe art thing, but u guys got way better things to do......
Different departments have different expertise in a gaming company. The artists, animators and modelers that worked on Ashe actually would not be able to work on matchmaking or player behavior issues... just like I would not be able to update the Ashe model


23 months ago

Quote:
Originally Posted by Opreich View Post
But still, go play ranked, doesn't sound very helpful.

They obviously don't play it for a reason.
I agree it's not an ideal solution.. but this is an unfortunate truth of competition. Whether it's Chess, Basketball, Soccer or Football.. if you are among the world's best, it is difficult to find you decent competition. It is also very, very difficult for me to approve giving mid- to high- tier players impossible-to-win matches just so the top players have super quick queues.


23 months ago

Quote:
Originally Posted by Drewsifer View Post
So then if youre last pick and your team needs a support and you pick a mid and go duo mid, you shouldnt be banned because (according to lyte) riot doesnt want people to play a particular meta. That argument is hypocritical to its fullest.
If a team wants to try 2 Mid, I am not sure why they should be banned? No one said Riot does not want people to play a particular meta. We do not want to define a meta--it's a subtle difference that means players can play however they want.


23 months ago

Quote:
Originally Posted by Drewsifer View Post
Someone (a red) said that youre free to try anything you want in normals but going 2 mid, double ad cary bot, or basically anything against the meta in ranked is bannable.
I am a game designer on the Tribunal, and this is not true.


23 months ago

Quote:
Originally Posted by KultoDeSkaro View Post
So, to clarify, Riot is perfectly fine with this scenario? IMO, the quoted poster was describing a solo queue player who screwed his allies over because "f**k the team I'll do what I want." Your point about "If a TEAM wants to try two mid, I'm not sure why they should be banned" is not applicable because it implies an agreed-upon strategy rather than randoms ruining the experience for everyone because they refuse to compromise or work with their team.
I agree, but the conversation was about whether Riot wants to define a meta, and we don't! We want players to explore, try out new strategies, and allow the meta to evolve.


23 months ago

Quote:
Originally Posted by Whoami44 View Post
You have just opened the floodgates. Thousands of trolls will now play double mid, double jungle, double top, triple ad, anythingbutsupport, isweareveisviable, and they will quote this thread as justification.

Ranked is for winning games. Going "against the meta" with 4 strangers is not for winning games, and in most cases is just something people use as an excuse to troll. If I'm playing a ranked game and somebody decides to double mid, I will report them.
Our players are fairly accurate in detecting trolls That's why the Tribunal is as effective as it is. In the near future, I may do a new thread just talking about general player behavior issues, including discussions related to the Tribunal.


23 months ago

Quote:
Originally Posted by Whoami44 View Post
However, there are many other cases where things aren't so clear-cut. What if somebody is last pick and his team wants him to play support, is he allowed to pick a 2nd mid instead? According to you he is, and that is going to ruin a game. What if somebody wants to play eve top lane? They can genuinely want to try at it and so you would be wrong to report them, but there's still a 99% chance that they'll feed and ruin the game.
I agree, some cases are not clear-cut; however, when I did some analyses on the subtle / on the edge cases, our players were quite accurate compared to our Player Support team.

Quote:
It's a complicated issue for sure, but I think there absolutely needs to be some requirements to adhere to the meta in ranked. I think we can all agree that something like AP Yi is acceptable (borderline maybe), but playing triple jungle isn't. I wish Riot would more clearly lay out the rules.
If we defined some clear rules or guidelines to playing the game, how would the meta evolve? We might still be in an era where Ranged AD was playing mid!

How *would* we agree that AP Yi mid is acceptable? Would the rules be different for different Elo brackets?


23 months ago

Quote:
Originally Posted by Yummines View Post
i would say the main issue is AFKs in champion select

if you get kicked out of a game because someone dodges, then you should not be put back into queue automatically

im pretty sure people lock in and then go up to maybe take a piss or grab a drink and then come back and theyre stuck with a random character because someone dodged the other game.
We are working on a few solutions for this. rjcombo might have information on this in the future.


23 months ago

Quote:
Originally Posted by MrAngels View Post
Lyte, I'm wondering what part of the game you consider to be susceptible to the most toxicity. Can you even get statistics for this, and if so can you share them?
There's a lot of interesting research we are doing on this front. It turns out that source of toxicity depends a lot on the individual and different 'triggers' will cause toxic outbursts in different individuals. In our current research we are trying to isolate which triggers affects the most people, and how we can either remove or reduce them in the game.

Some examples of triggers are obvious:
- misalignment of intention in Champ Select Lobby
- the moment when your team gives up first blood
- the moment when your team loses a Baron
- end of game chat
- etc


23 months ago

Quote:
Originally Posted by kabaab View Post
For single / duo queue can we simply not have a player vote system?

All players can vote against a player on their team as being troll / during champ selection if that player gets 4 votes against him he is booted and banned from ranked for 15 - 30min and everyone else is requed with no penalty...

That way Riot can easily see if some players are getting abnormally high amounts of vote bans and address them...
Unfortunately, player vote systems in Ranked are very dangerous. For example, let's say 4 players in a lobby agree to play Ranged AD, Support, Jungler, Top. However, the 5th player says "Hey guys, my best role is Ranged AD, could we work something out?" Instead of even communicating, the 4 players usually decide to just vote-kick the 5th player.


23 months ago

Quote:
Originally Posted by code81 View Post
And then he gets placed back into a sub 2 minute que time, providing he's not 2k+ elo(which at that range WOULD NOT EVEN MATTER).

I don't see the issue.

I'd be completely fine with being kicked backed to re queue if I didn't get to play one of my 3 roles, instead of being forced into a role I am mediocre at and would not be satisfied if my team lost because of my mediocre play.

Why don't you guys actually draw some data for this, or have you guys already data mined before assuming?
We have a lot of data on this particular issue since we were considering how to resolve a lot of the toxicity that appears in Champion Select Lobby. Basically, you'd see the number of queue dodges increase almost 300% as players vote-kick if they do not get the roles they want. Some players might ask to be vote-kicked like you are implying because they prefer not to play whatever role is remaining. Some players might be vote-kicked because they refuse to work with the team. Some players might be vote-kicked because the team agrees they just don't want the matchup the opposing team selected.

Then we would have to develop a solution or workaround to identify or combat the players who are vote-kicked the most, which then has to be able to identify which players are toxic (and thus being vote-kicked) or are innocent and being trolled by toxic players. It gets messy

Quote:
What about this instead? Human verification after queue pops, so they won't be placed in champ select until they enter verification.
If verification is not entered, queue is dropped automatically.

This
1) Does not count as dodging as they have not entered the game lobby yet, so no elo is lost.
2) Does check for AFK
3) Does not get rid of negative triggers in champion select- same roles, troll picks, troll summoners etc.

However, I do believe my suggestion is a viable bandaid.. for now.
rjcombo is working on similar ideas.

Quote:
Also, Ranked is privilege not a right.
Repeat offenders (ragers etc in tribunal) with multiple suspensions before should also be suspended and or banned from ranked.

ie; 1 week irl suspension from league, when they come back, ranked will be disabled for them for another week. They will still be able to play norm games, but not ranked.

Champ select logs should also be shown in tribunal. A lot of foul-mouthing goes on there at lower elos.
We considered Ranked-only timebans, and they are still on the table.

Re: Champion Select Logs in Tribunal, we are still doing more research on this. Usually players who are toxic in Champion Select are also toxic throughout other areas of the game which are observable in the Tribunal already. The value of Champion Select Logs is if we can now catch toxic players we could not before... but I will keep investigating to see how large that value is.


23 months ago

Quote:
Originally Posted by Imclueless View Post
A good number of players (that I've run into) restrict their negative behavior to Champion Select or Post-Game chat because they know it's immune to the Tribunal. One gem that springs to mind is "I'm feeding, and you can't ban me for it" in Champ Select. Then, in-game, they pretend to just be terrible, including apologizing in chat.

To appearances in the Tribunal, this looks like someone got picked on by their team for doing poorly. This case would look like a 4-man premade decided to make up stuff about somebody having a bad game.

Now, the irony is that, by adding Champion Select chat into the Tribunal, we'd see similar behavior, just with no admission of guilt in Champion Select. So, ultimately, it would only serve to catch people who failed to realize that it was changed.

One thing that it could do, however, in addition to finding things like this: find individuals who are uncooperative with their team in Champion Select. Things like "I'm not playing Support, screw you" *lock Tryndamere*. Said player may not be a toxic force in chat in-game, but they are certainly breaking the Summoner's Code by refusing to cooperate with their team.

Edit: Just to note, if Champion Select ever DID make it into the Tribunal, it should include things like pick and ban messages. Chat doesn't tell the whole story, in Champion Select.
Yeah I completely agree with you that some players are intentionally being toxic in Champ Select Lobby then putting on a 'show' of innocence during the game. The fortunate thing is, many toxic players who try to do this actually cannot hide themselves as well as they believe in the game so we end up punishing them anyways.

The question then becomes: of the players who are extremely good at being toxic in Champ Select, but amazingly good at faking it during game chat... how many players is this and does it justify the cost of this feature, or should we work on something else completely that will try to fix this issue from a different angle?


23 months ago

Quote:
Originally Posted by Arisonius View Post
I empathize with the cost vs reward, Lyte, but trolling is obviously a huge concern for the gamer base. Now, whether or not it is cost effect may not actually be the problem here. A more accurate picture may be looking at the cost vs. the perceived impact to players. There's no doubt in my mind that you guys are doing what you can to weed out the negative influences in LoL, I see the attempts being made. But I would like to point out that changes like post-game/pre-game chat, champion selection, and banning, are things that would not only help to identify and remove toxic players from the game, but it would also be a quality of life change for Tribunal members. There you have a two fold reason. Also, there is the positive perception that Riot would gain by being able to say "Yes, we heard you, and we acted." regardless of whether or not the option may be cost effective, this is a major concern among the community. We, to be blunt, don't care if it is cost effective. To us, the perceived gain far outweighs the perceived costs.
I don't disagree with you, but with a background in psychological perception I definitely take player perception into account when making decisions.

It's not a coincidence that on Team Player Behavior, all of our features exhibit two sides of the same coin. There's always the pragmatic, practical value of a new feature and on the flip side there's also the perceptual value of a new feature.

I know the players don't care if something is cost-effective, but if we choose not to work on a feature, we have made the assumption that another feature will be a combined higher practical and perceived value. Sometimes we are wrong though! It's an imperfect science.


23 months ago

Quote:
Originally Posted by CupcakeTrap View Post
Just played this game.

Check out the Ranked Elos of the players. They had a 15XX and a 16XX, backed up by a 12XX and a 13XX. ... we did not have comparable ability levels. The Kass in particular just wrecked us.

It was still fun, I guess? But not a great MatchMaking success.

LoLMatches entry: http://lolmatches.com/match/na402782548
(I believe this is using "Normal Elo", though.)

Huh. So based on LoLMatches ... maybe it's just that this is a case of our normal Elos not matching our ranked Elos?
Yep, we'll be doing some work to reduce this in the future. Right now, very skilled Ranked players who don't play Normal much will be in these situations quite often.


23 months ago

Quote:
Originally Posted by Stexe View Post
I'm curious to know why you only add "RL" friends to your "RiotLyte" account when you could easily create an account without the "Riot" tag or without a known name and still have all your friends on it. Seems counter intuitive to say the least.
It's just a personal preference; I have quite a few accounts with just RL friends on them. I just like some "me" time when I am logged into League. I am super reachable through several other channels online.


23 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
I wasn't really saying so much the matchmaking would match based on this (sorry if it came off that way, sometimes I say things derpy :P) so much as DURING champ select people can see the kinda role or the specific heroes they play more often and figure out how to build their team with the strengths of the people in the queue with them.

But ya, what do you think, Lyte? Do you think that having something in queue you can use to see the person's preferred heroes of play (i.e. ones they've played the most and/or had the most success with given a number of games or something) would help? I just feel like there needs to be even more in champ select to help people see the kinda people they are matched up with (as in the heroes/archetypes they enjoy) and perhaps some more time (30 seconds before even banning begins or however long) to discuss their ideal team build and bans. I don't know if 30 seconds is enough and I do understand that it could make champ select time a long time, but I figure it'd be a fair trade off for a good game. I mean, hell, if you have a lot of time to talk before even banning, then the banning process could be turned in time (instead of 30 seconds per ban, 15 seconds) and the same with the actual champ select process (30 down to 15). It would give at least give 3 minutes before banning at all to discuss what to do (unless everyone hits ready or timer gets to 0). If 3 minutes has passed and no one hits ready, everyone is sent back to queue (means there was an AFK or someone really didn't like their team's discussion or whatnot).

I feel like this solution could solve multiple problems. First, it gives your team time to really talk about your strategy and team comp (3 minutes should be enough time) and it doesn't even add to the total time now it would take to do the pre-game stuff. Second, it would solve the afk problem in queue without having it be too abusable for "man I hate this match up, I'm gonna just not hit ready" because they don't even KNOW what the match up is. They won't even see the other team's comp and know how to ban against them.

So ya, thoughts?

Created a thread:

http://na.leagueoflegends.com/board/...php?p=24724682
There are some merits to the idea; however, showing players more information in this case might be surprisingly negative. For example, let's say you are 80% wins with Irelia, but 55% wins with Graves. As a player, you want to work on your AD Role and learn more AD Carries but you have no other games with AD Carries. Do you think your teammates will actually let you try an AD Carry or will they pressure you to play Irelia, and play top lane? In Ranked Games, it would be near impossible for players to play anything but their best 1-2 roles/champs. You could argue that these players should then play Normal Games and level up their 'stats' with these roles or champions but then Normal statistics have an effect on Ranked Games...


23 months ago

Quote:
Originally Posted by Dex7ter View Post
another duo who play only in 2 alone, they dont need team
taric 26 win corki 48,

can u just remove this duo, its just makes more problem then help.....
Dex7ter, please stop spamming this thread. I took a look at your account details and you are typically a very toxic presence in your games.

You have also reported over 2300 players when many of them did not deserve it. Please use this opportunity to curb your behavior and try to be a bit more positive in your games.


23 months ago

Quote:
Originally Posted by Burnie View Post
Can't you just ban him from the forums? Or speak to a red that can?
I'm hoping he takes the hint and tries to be more positive.


23 months ago

Quote:
Originally Posted by Guttler View Post
I don't even know is Lyte still reads this thread, but here is some food for though.

I used to play professional level pokemon and I would use a program called Shoddy Battle to test out different pokes, team comps, abilities, etc. It used the Glicko2 rating system: http://shoddybattle.com/ladder

Basically, rather then giving players a solid 'number' that they could see moving up and down as each match progressed, it would give players a range of values in which they were most likely to be rated. The range was quite large too so players could never be sure of the exact number attached to their skill. It also fluctuated quite slowly so winning or losing 1 match would not show any changes.

I think one way to adjust player behavior may be to change the elo system. Right now you've got your number and the goal of ranked is to get as large a number as possible. Players don't like to lose ranked games because of forces outside of there control, like having a leaver, just like pokemon players don't like to lose games because there move with 80% accuracy misses five times in a row. (true story)

The Shoddy Battle rating system also required players to be fairly active to progress in ranks. Right now in LoL I have several friends who have like 1295 ratings with 18 wins and 17 losses and they just do not play ranked. They like that they have an elo number in champion select and don't want to risk losing it due to a fe bad games. As season ends, you'll surely get lots of complaints about how people are sitting on there rating for skins/badges. Pokemon looked for players who played well consistently and often since the outcome of only a few games couldn't possibly be a good way to evaluate player skill with something as important as a number.
I agree, the Glicko system has some advantages over the Elo system. I know many players (and Rioters) have said that we use the Elo system but realistically, there's been so many modifications and changes to it that it's not really like an Elo system anymore.

Yegg and I are brainstorming and thinking of ways we can incorporate some of the ideas Glicko uses into our system, or even creating something new. However, before we do something extreme like this, we have a nice list of items that we want to implement that will upgrade the current matchmaking system.


23 months ago

Quote:
Originally Posted by litk View Post
Lyte,

could you please explain why we played against the same enemy team again after we have lost the match with them?

They are pre-made as well.

They are far higher levels than us...They are having two 900 wins and one 100 wins in their team while we only have 200 wins.

What's the point of playing unbalanced game again with the same enemy team after losing the match?

Did you really do anything with the match making?

Or you want to tell me that my team is really strong to play against them?

Totally sick of unbalanced games.

Please find the attachment for match screen shot.
Was this game really that unbalanced? The gold was fairly close, the kills/deaths/assists were close, and the game could have been won by your team.


23 months ago

Quote:
Originally Posted by Eph289 View Post
Lyte, are you guys still thinking about some kind of wizardry to help reflect the fact that not all players play their entire lineup with equal skill? Seems like an insurmountable task, but I feel it'd help the accuracy of ratings. Or maybe the system is designed to rate your skill holistically rather than just your specialty?

For example, Reginald is best-known for mid-lane casters. Put him as AD carry and he'd still be better than a lot of people, but not at his true skill.

In my case, I've beaten people at a 1500-1600 level in mid with some of my strongest champions. Put me on Vayne or anything that'd typically go toplane . . . and I'll struggle to beat someone rated at 1200.

If you can speak to it, what kind of thoughts are flowing about whether this is an issue and how much wizardry would be needed to fix it?
Elo systems tend to be holistic. Whether it's a FPS, RTS, or MOBA, very few systems try to take into account your specific skill with classes (Sniper, Soldier, etc in FPS) or class-equivalents. In your example, consider that your Elo range is probably not just 1400. Your Elo range is probably 1300-1500, with your best characters being represented at around ~1500, and your worst around ~1300. So what is theoretically happening is sometimes you play your best character and face off against someone who has 1600 rating but is playing their worst character (so it's about ~1500 skill)--so you win this match-up.

Ratings are never just one specific number. Behind the scenes, it's always a range, and there's always associated 'confidence levels' with where we think your range is.


23 months ago

Quote:
Originally Posted by Eph289 View Post
Can't speak for FPS or MOBA too much, but I'm not sure I follow this logic from a traditional RTS standpoint. You pick your faction in any competitive RTS and very few people will offrace consistently except on smurfs. I guess there's some variance for matchup ala TvZ vs TvP or GDI vs Nod vs GDI vs Scrin, but that doesn't seem nearly as different as role differences in League (e.g. mage vs jungler, support vs ad). Now, my assumption is that faction : champion class. You guys might see it differently?

Unless the moral of the story is that a more accurate rating will be your performance in premade 5's where you're consistently playing your strongest champs rather than solo queue where you get whatever the team dumps you into.
Keeping the discussion to RTS games, the key there is that very few people off-race on their main accounts. If they did, that would be similar to choosing different champions/roles in League of Legends. If they chose a different race in a RTS, do you think the matchmaker would adjust appropriately and give them fair matches right away? The player will most likely lose several dozens of games until his "Elo" drops to where his off-race should be. So if you play League of Legends and never change your champion/role (i.e = "race") the matchmaker actually is super accurate and your Elo range would probably be much tighter--1200-1250 instead of 1100-1300.


23 months ago

Quote:
Originally Posted by CupcakeTrap View Post
I feel like I should briefly bring up a suggestion I mentioned earlier: Elo profiles.

Goal: let players indicate how they're playing, without needing a new smurf account.
Concern: ensure that players can't abuse this to somehow roflstomp lower-Elo players.

Your default is "all champs available".

You can also create a new profile, which starts at your "all champs" rating. When you make the profile, you can mark certain champs as "off-limits". (You must still have at least 17 champs available.) You can later modify the profile to mark additional champs off-limits, but you can't ever mark an off-limits champ "available" again.

So at any given moment, I might have:

All-Champs profile Elo: 800.
Anyone-but-Caitlyn Elo: 700.

If I created a new, "anyone-but-Caitlyn-or-Veigar Elo" profile, it would begin at 800.


My related idea, which is more modest but I think would also be very useful, is to track Normal Blind and Normal Draft Elos separately.
Isn't the profile suggestion similar to us just giving players Level 30 accounts? More realistically, the suggestion requires a LOT of client work, UI work and back-end work and I am not sure it solves the problem any better than offering Level 30 players new accounts that start at Level 30.

PS - Offering Level 30 players free Level 30 accounts is not something we want to do!


22 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
Lyte, this isn't in relation to matchmaking, but it is user based:

When are the forums finally going to remove or at least change this vote system? This has to be the only major forum (maybe only one) I've ever seen where the votes can actually out right close a thread. That's HORRIBLE for getting any sort of discussion going on.
Not sure, this is a feature I don't have too much visibility on. I generally haven't seen many threads that are closed that are genuinely positive or valuable discussions but I'm sure there are some examples.


22 months ago

Quote:
Originally Posted by CupcakeTrap View Post
What percent of attempted games are cancelled because of a queue-dodge in Normal?

I just had 4 in a row.
Was it the same player multiple times in a row? Sometimes a rare issue happens where the same AFK player hits multiple queues in a row and keeps aborting the game.

Overall, the queue-dodge number shouldn't be high enough to see 4 queue dodges in a row that often.


22 months ago

Quote:
Originally Posted by Turbobuns View Post
Bump! What the hell is going on here? This a important thread. Why is it going stale so quick
We've been quiet developing some of the upgrades to matchmaking... I'll have more to update when they start trickling into future patches.


22 months ago

Quote:
Originally Posted by Clerigon View Post
Lyte, any news about the normal draft mode and premades? before the darius patch a red said that fix is not implemented yet because it was bugged...
The next patch has this fix. Sorry it took so long!


22 months ago

I am writing a new FAQ for matchmaking, since I often see the same questions raised again and again. Could you guys help me with which questions you want answered in the new FAQ?



22 months ago

Quote:
Originally Posted by Turbobuns View Post
Should there be an emphasis on educating the importance of respecting the pick order, and building a well rounded team?
We haven't explicitly said anything about pick order before. Honestly, in the average game, being first pick or last pick isn't a significant Elo difference in the grand scheme of things so first pick shouldn't believe they are the 'best' player on the team and therefore can pick whatever role he wants.


22 months ago

Quote:
Originally Posted by Sotere View Post
I asked this in a separate thread but very shortly: do duo queuers over 1900 elo have a higher measure of success than in other levels of elo?
What do you mean by higher measure of success? We do find that currently the best metric for success is the Elo difference between Duo-Queuers. If the difference is small, they tend to win more games. If the difference is large, they lose more games. The Elo difference is more important than the actual range of the Duo-Queuers.


22 months ago

Quote:
Originally Posted by Madman Reborn View Post
Could you elaborate on what you find large and small differences. I don't often play with Z00mm3rr so our elo tends to drift at times. So we've had times that our elo was about exactly the same and we had time we had about 150 elo difference. Is that still a small amount of elo range?
I can give exact numbers at this time, but yes, you're still in the range where you see some small benefits to Duo Queuing.


22 months ago

Quote:
Originally Posted by GodEmperorLetoII View Post
Does this mean you'll possibly instigate a rule you can only duo queue with someone within a moderate range of yours?
Possibly. I'm working on some tweaks to how Duo-Queues are matched, and we'll see what happens with those changes first before considering Duo-Queue Elo range restrictions.


22 months ago

Quote:
Originally Posted by BelligerentGnu View Post
Any news on when some new stuff might be implemented?
Unfortunately, not yet. However, I just finished some side work on a matchmaking FAQ. I'll be talking more about 5-6 matchmaking changes in the near future.


22 months ago

I have added 5 new items to the "In Progress" tab, and 1 additional item into the "Completed" tab in the original post. Thanks for your continued feedback!



22 months ago

Quote:
Originally Posted by Kenjuku View Post
Hello professor lyte, I have an inquiry.

Is there anything we can do about trolls in ranked lobby?

Honestly I am at my wits end with these people who find joy by ruining a team based game with their antics. I am a season 1 player who is currently raising his elo. I started out down at 828 and I am now currently 1118 because of the last troll I had. I encounter one of these people on an average of 1/3 games.

Honestly is there anything you guys are cooking up to let us kick someone from rank queue? Or just dodge queue without losing elo? Especially when we can obviously see that he/she is a troll? Using my last game as an example it was An Eve ista-lock with teleport and surge. Never spoke in the lobby and once he/she went 0/7 this person began taunting our mid player (brand) who decided to not deal with it anymore and then went afk. Following this my lulu started feeding followed by me since every single person on their team had a lot of gold advantage.
We are removing queue dodge -Elo penalties in Ranked soon, so you can at least have that option to avoid extremely toxic players; however, we will probably not do a kick option in Champ Select lobby for numerous reasons that have been discussed in this thread


22 months ago

Quote:
Originally Posted by Asko View Post
Hi Lyte,

I'm really worried about this tbh. I really don't have time to go through this whole thread to find the discussed precautions so could you tell me real quick what are the measures to deal with incoming mass dodging based on team comps?
Basically, I am shifting my teams to take more scientific approaches. We are willing to test (as long as we understand the potential risks!) and then collect data to see if a feature works and make rapid iterations as necessary if it doesn't.

With regards to the Ranked Queue Dodge change, we realize that queue dodges may increase; however, if it increases beyond our comfort levels we will immediately go back to the drawing board with a new solution.

Right now, the changes are simply increasingly higher time bans that are more severe than Normal Mode queue dodging; however, if queue dodges increase dramatically, our immediate backup plan is to add LeaverBuster to the system such that if you continue to queue dodge way more than you actually stay in the lobby, we'll give you suspensions from the game up to 14 days. If that still increases queue dodging, we'll need to go back to the drawing board for back-up plans C and D.

Quote:
On another matter, how high of a priority is normal game balance for you? I play almost exclusively normals nowadays and 3 games out of four are utterly onesided because of huge ranked elo differences. I've had games in which I was in a team with three 2.1k players against team of 4x unranked and 1x 1400 elo and other way round, my latest game from today was like this, my team vs enemy team:
TOP: unranked vs 1650
MID: 1658 vs unranked (im sure this is a smurf based on player name, skill and games played [only 160 normals])
JUNGLE: 1855 vs 1863
AD: unranked vs 2008
SUPP: 1645 vs 1536

so basically both players have 2 ~1600 players, a ~1850 players, 1 unranked but the great difference here it that my team had another unranked and they had 2.1k player.
Is something going to be done about this? My suggestion would be that in normal matchmaking the highest elo is used, if someone is ranked 1.4k in normals but 2.2k in ranked this player would be placed on 2.2k normal level.
We are already working on something like this.


22 months ago

Quote:
Originally Posted by ncrwhale View Post
Hi Lyte,

I thought I read that groups would not receive priority for pick order. Was it random that my 2 friends and I were first pick for 6 out of 6 games yesterday?
The fix for this still has not gone live--we had some hiccups, but it should go out live next patch. Sorry for the wait on this one.


22 months ago

Quote:
Originally Posted by Princess Garnet View Post
OK this is exactly what I am worried about.

The goal in ranked games is to gain Elo. This goal, in general, has to do with winning more than losing; it does not have much to do with having time to play a lot. Therefore, a time-suspension does not actually create sufficient amount of incentive to not dodge queries as time suspension does not really conflict with any players' goal in playing ranked games.

What this is going to do is that it's going to severely nerf counterpicking; a lot of more people will dodge due to a bad matchup because it's better off dodge and get suspended for a few hours than risking and playing a bad matchup and LOSE ELO.

I think a better model is just to have two sets of Elo; one which is displayed and lowered by dodging, and another one, hidden view, which is actually used for matchmaking and is not lowered by dodging.
Here's a counter question:

What does queue dodging have to do with you being a worst player? Does the punishment fit the crime?


22 months ago

Quote:
Originally Posted by Elysion IX View Post
I think one could argue that elo is a indicator of not only a player's "skill" at the actual game, but also a reflection of how they perform in a team(and thus why things like individual k/d/a ratios don't factor into your elo). With that in mind, I think queue dodging is the epitome of "not a team player", so I could see how losing elo for a dodge makes sense.
Sure, Elo is a representation of several variables and your "team play" skills is one of them. However, how many people queue dodge because of a toxic player in the lobby? How many players queue dodge because of team composition? How many players queue dodge because they did not get the lane they wanted?

For the players who queue dodge because of an extremely toxic player... what does that have to do with being not a team player? Should they lose Elo for this decision or should they force themselves to play 35 minutes with a toxic player?

-Elo will deter some players from queue dodging for some of the above reasons. However, if we keep increasing time bans, I guarantee you at some magical threshold, players will also stop queue dodging to avoid the time bans.


22 months ago

Quote:
Originally Posted by Princess Garnet View Post
I am not 100% sure if I understand your point, but I think it does. Assuming one values Elo (which is the whole point of ranked games), and he is put into a situation which:

1. Either play the game in the unfavourable matchup (e.g. got counterpicked), and have a high chance of losing and lose Elo;
2. Dodge, receive no Elo penalty, and gets a time penalty instead.

I think the choice is fairly obvious, because Elo is gained through winning than losing, not purely through time. Therefore any action that minimizes the possibility of losing is favourable and encouraged.
Human behavior works in predictable ways though, right? For example, if the time penalty is 5 minutes then the choice is obvious. However, at some magical threshold, players will no longer want to queue dodge. For example, what if you could not play Ranked Games for 2 weeks after excessive queue dodging? Would you still think the choice is so obvious?


22 months ago

Quote:
Originally Posted by Black Sky QQ View Post
Who came up with the idea that it is suddenly A-Ok to ban people for 14 days from the game for simply queue dodging ranked games?
Because frankly that is plain stupid. I would understand a 14 days ban from playing ranked games if you queue dodge too much but from the game?
Come on what are you people thinking.
Also what exactly makes queue dodging ranked games a ban worthy offense?

/very disappointed
I haven't revealed any details about how many queue dodges you would need to do to get the punishment, or what the punishment levels look like--so let's not jump to crazy conclusions that players will be banned for 14 days for queue dodging 1 Ranked Game.

This is an experiment that we are going to keep iterating on and figuring out what the right values are. The goal is to have a Ranked atmosphere that has very few queue dodges and an environment players can not intentionally abuse to lower their Elo to 300 to stomp lower tier players.


22 months ago

Quote:
Originally Posted by Madman Reborn View Post
I just hope you for a steep curve on the ban times. And I mean very hard. You don't Q-dodge twice in a row because a friend came on while in Q. Also, read my previous post?
We will play with the curves to see what works. If time suspension curves don't work, we'll try something else. Basically, we are trying to do more experiments and then just closely monitoring the data to inform future decisions and iterations.


22 months ago

Quote:
Originally Posted by Banuvan View Post
So with this in mind what is to prevent them from not dodging but instead afking in the game or trolling the game to cause a loss. I know they can be reported but this still doesn't take care of the issue that they just caused 4 other people to lose ranking.

There is no forgiveness for a troll or afker causing a loss. People still lose ranking. Atleast if they.dodge the que then they dont cause the other players on their team to lose ranking.

I would rather have to wait extra time for a game with people who want to play than have to deal with the above situation i described.
If players choose to play a game and instead AFK or troll in the game, they are highly likely to be caught by reports and the Tribunal system and permanently removed from the game.


22 months ago

Quote:
Originally Posted by Princess Garnet View Post
Do you see the problem? Assuming the change is allow players to dodge other toxic players, then nothing actually changed; the only thing that was changed was to replace one penalty (Elo-loss) to another (time-loss). But time-loss actually gives more incentives for ladder-climbers to dodge queues because of how the ladder works, so the time-penalty has to be large. In that case you are punishing honest and reasonable players who really got matched with toxic players. This seems counterproductive.

I think this change makes dodge-smurfing impossible (bomb Elo by dodging, then proceed to own noobs), which is nice. But if the changes are geared toward preventing dodge-smurfing, then this is the wrong way to do it. A much easier way to do this (as I have mentioned) is to simply have two sets of Elos, one visible and affected by dodging, and one invisible and unaffected by dodging. In fact this is super-simple to do, just use:

Visible Elo = Invisible Elo - 20 * Number of Dodges

And only use Invisible Elo for MM purposes.
There is no perfect solution, and we realize that. The main goal of this specific change was to stop abusers from intentionally queue dodging down to low Elo and ruining games because they are actually a higher Elo player.

With regards to queue dodging due to toxic players, we just want that to remain about the same for this particular solution. We'll be trying to tackle this particular problem separately.

Invisible Elos and Visible Elos being separate systems behind the same Ranked mode is generally a poor idea. In your case, Ranked Players who queue dodge a lot might see their visible Elo as 1700 but constantly be playing against 1300 players. That doesn't make sense to the player and is difficult to educate.


22 months ago

Quote:
Originally Posted by ExpertSockThrow View Post
Sorry, I'm kind of in a bad mood today. But it's still true: knowing that the bad guys might get banned some day is small comfort when anyone can make as many accounts as they want, and more people than you can imagine will try to troll and ruin matches. It's what happens when you have an extremely popular f2p game. Our question is, why is it taking Riot so long to address this situation in a decisive manner? The Tribunal and Leaverbuster are steps forward, but they don't produce fast results that players can see. For the most part, we have to take Riot's word for it that the systems are effective. Games are still ruined, ELO is still lost, and player perception is that these systems are weak and ineffectual.
I agree with you, it is frustrating when a troll or AFK is in your game and you suffer through with just a hope that one day the troll will be banned by our systems.

We are trying to develop more features and complex systems that can prevent trolls from being in your games but this takes some time. This year Riot openly said that player behavior is a serious problem we are going to tackle and created Team Player Behavior & Justice. I recently started in February to help guide this team and we're going as fast as we can.

However, we had to launch several research initiatives to explore what player behavior even looks like in online games and what causes toxic player behavior. Then, we had to design several complex, often 'scientific-like' research approaches to feature-design that was not how traditional game development is... so these features are taking some time. We can only ask that our players be patient, and we'll talk about some of these features when we're ready.


22 months ago

Quote:
Originally Posted by Banuvan View Post
We understand that this stuff takes time lyte. We really do, it's just very very frustrating to those of us who just want to play.

On that note, is it possible to set something up that makes it so we don't get paired with people on our ignore list? This would eliminate a ton of work for you guys because we could choose who is toxic to our game for whatever reason and put them on our list which would eliminate us from seeing them again. Obviously it wouldn't catch everybody because every game we could be paired with a different toxic player but it would help not seeing the same people again and be a beginning to eliminating the problem.
Actually, you should rarely, rarely meet the same player twice unless you are in a weird Elo bracket, or playing at very off-peak hours. Are you seeing the same toxic player over and over again in your games? If so, let's dig into this deeper and figure out what's going on. Generally, if you wait 5-10 minutes after a game you will almost never see the same toxic player 2 games in a row.

The reason we don't explicitly make your Ignore List a "don't play with this player again" button is because it's very complicated in theory and implementation. The average player will use this button to ignore players they feel are 'unskilled' which undermines the actual systems trying to calculate player skill. This also makes it extremely crazy for the matchmaker who now has to find 10 players who are similar in skill, but also check to see that none of the 10 players ignore lists includes any of the other 9 players in the list.

Another complex issue is, let's say you are a support player in the 1200 Elo range. Your average stats are 0-5-12, which are decent for a support. Players might /ignore you constantly because you are an unskilled players in their eyes and eventually you are stuck in this weird pocket in the matchmaking system where tons of players your Elo range has /ignored you--not because they hate you or you are toxic, but because they just don't want to play with someone who averages 0-5-12.

Quote:
Also you could set something up so that a program could scan people's ignore lists and if somebody shows up on a large number ( no clue what a large number is for you guys ) of ignore lists then it could trigger something internally for the Riot team to look into why those people are on so many ignore lists. Maybe look at their last 10 games or something. If that is just too much work then maybe set it up so if they show up on a large number of ignore lists their last 10 games get put into the tribunal for review by their fellow players. This would eliminate work for you guys as well and put the onus back on the players to make the right choices.

I am not a programmer of any type so have no clue if this would be possible or if it would just be too much work on top of your guys other projects. Just throwing ideas out there.
We definitely are looking at which players are 'pockets' or 'nodes' of toxicity--one example of a node of toxicity is if we see a player who has been ignored by at least 1 person in every game he plays...


22 months ago

Quote:
Originally Posted by BelligerentGnu View Post
Lyte, have you considered that greater queue times due to dodging might be an acceptable cost for the ability to avoid toxic games?

I never dodge currently, and my reaction to this is "Yes! I can dodge trolls now!" Even if I spend much longer waiting for a game, time I spend in queue is time I can spend reading, or playing a different game, or what have you. It's time I don't have to spend trying to salvage a win in the face of one or more trolls who make playing the game a chore.
Yeah, we realize that for some players they might use this new queue dodge penalty opportunity to simply dodge toxic games; however, if our players are happier and ultimately play more League of Legends because of this... I no complain

If players use this to queue dodge and don't even play League of Legends anymore.. then I am a sad panda


22 months ago

Quote:
Originally Posted by Diali View Post
Would it be possible for a person to be auto-muted every game upon a lot of people ignoring them? Only issue I can think of is how important communication is but toxic players cause games to be lost anyway. Also, the player could be still be un-ignored as per usual.
Certainly possible and would not be difficult to implement; however, like you mentioned League of Legends is a very social game. If we auto-muted players not only would we be forcibly cutting some communication, we would also be labeling players as toxic to everyone else in the game.

Can you imagine the consequences of a new world where auto-mute was in play? Every game, players would be like "Everyone say hello!" and if you don't, players assume you were toxic and auto-muted.


22 months ago

Quote:
Originally Posted by Banuvan View Post
Well if they refuse to even say hello then they are refusing to communicate which is a reportable offense. How is that a bad thing?

If anything it would force people to communicate with each other at least a little bit rather than having somebody sit in champ select and say nothing at all. If they say hello then you know they are there at the least and maybe they will actually talk to the team at the best and work together during champ select unlike so many people do.
Encouraging players to communicate is a good thing. Forcing players to do so, with the threat of a Tribunal ban or other punishment is a bad thing.

I'm actually removing a lot of the Report options in the near future.


22 months ago

Quote:
Originally Posted by zarbiz View Post
I'm normally with you 100%, Professor Doctor Lyte, but I disagree that forcing communication is a bad thing. There have been numerous times in champ select (ranked) where I'll call out what I can play and what I'm in the mood to play, and maybe two or three other players start calling roles and we decide that the silent person will then fill the final slot. However, that never works out. They grab whatever they want and turn our plan into so much wasted energy.
I think the key point there is, "with the threat of a ban..." We want to avoid forcing players to do a certain action out of fear of being banned. I've been thinking a lot about the Champion Select process and how to improve player behaviors there, but there's nothing I want to talk about just yet.

I definitely agree that a lot of toxicity begins in Champion Select Lobby and carries into the game, which leads to quick finger pointing and such when a player makes a mistake or gives first blood for example, but resolving the toxicity in Champion Select Lobby is quite a complex problem with no perfect solution.


22 months ago

Quote:
Originally Posted by Nores View Post
But say If I want to dodge(say I got counter-picked , or I don't like certain champions on my team ) with no elo penalty in place ? - then I can simply lock in eve or some such and make sure the game is super toxic from champ select. Chances are other people will dodge and incur the penalty on themselves

You can try this yourself - go into normal draft mode and try make other people dodge on your team as much as possible. I can do that 80-90% of the time if I want
Even before the patch changes, what stopped you from picking Eve and trying to force players to queue dodge and intentionally being toxic?

The purpose of this change isn't to curb all toxicity in Champ Select Lobby--we have other things in the works for that. The main purpose of this change is to stop players abusing the -Elo penalty to drop their Elo on purpose to stomp on lower skilled players.


22 months ago

Quote:
Originally Posted by Khammurapi View Post
Lyte, there needs to be a penalty for dodging dominion queues.
I'll look into this on Monday.


22 months ago

Quote:
Originally Posted by Princess Garnet View Post
I don't agree with your 1700 to 1300 example for a few reasons. First of all, assuming this is implemented, no one is going to dodge 40 games (10 Elo per dodge), at least not in a short period of time, so no one is going to sudden experience huge Elo gaps in the MM system.
There are already players who dodge 100-125 times in a row in a few hours on purpose, dropping their Elo to 300 to stomp newer players. This is why we are even testing alternate queue dodge penalties.

Quote:
Also, it's EXTREMELY easy to state this change on the patch preview/notes. It's easy to understand (Visible = Invisible - 10 * Dodges; I frankly think any 10 years old can understand the formula itself), so it's pretty easy to keep everyone informed. After a few months after implementation, we may perhaps see an 1700 player meeting 1300 player; at that stage, probably everyone has read the patch notes, so everyone knows that if he is playing against someone with significantly lower Elo, it's because that player dodged a lot of games.
Actually, very few players read the patch notes, which is why we often have to describe important changes directly in-game.

At the very core, by definition, your visible Elo should represent your skill level. Adding leaver scores or queue dodge frequencies to your visible Elo simply means it is no longer displaying a score relevant to your skill and that's why it's a less than ideal solution.


22 months ago

Quote:
Originally Posted by CupcakeTrap View Post
I actually like this idea.

"Assisting enemy team" is a better catch-all label, IMO.

Too many people just don't seem to understand the meaning of "intentional". They treat it as though it were merely an intensifier.

Perhaps something like "betraying the team" would be even clearer?
Yes, I hate the Intentional Feeding report. I'll talk more about the new reporting systems in the future.


22 months ago

Quote:
Originally Posted by Ardnalis View Post
Won't be much to look into Lyte. Go into game -> Dodge -> Shock and Awe, no dodge penalty.
I meant I'll look into what happened to cause this and how easy it would be to throw it in


22 months ago

Quote:
Originally Posted by MouthØfMadness View Post
Can you elaborate?
As an example, rjcombo's team is working on some solutions to AFK players that get into Champ Select Lobby. Other features we are not ready to talk about yet.


22 months ago

Quote:
Originally Posted by Hazdoc View Post
what is being done about the queue times for players who are highly ranked? I know a few guys who have to wait around 30 minutes to get a match, and I hear Ego's waited an hour before.
Unfortunately this is a tough problem to solve. Top players in any game are going to have difficulty finding fair matches and we generally do not want to force top players into games with lower skill players. For example, we could make queues 5 minutes and put 2500 players with 1800 players... but that would not be a great experience for the 1800s in the game.

It's a problem we have to look at, but there are no obvious solutions at this time.


22 months ago

Quote:
Originally Posted by Gaeta View Post
Mister Doctor Professor Lyte,

I just played a game where I queued together with 4 of my friends (in Normal). We are of wildly divergent skill levels, but we wanted to play, and since we were all right there (we work together), we sat down and went for it.

Since we were all in the same room, we were able to shout at each other and plan together and work quite well as a cohesive team, and as a result, we barely won the game. I suspect that, had we not been in direct communication, we would have gotten absolutely crushed - which matches my usual experience with premade groups that include divergent skill levels. In fact, I rarely duo anymore with people who I know are much better or worse than me, just because I feel the matchmaking system often gives us a rather unfair opponent.

Does matchmaking assume that premade groups will always have external communication? Would it not be a good idea to give that information to the players, so that we don't queue together, without any voicechat or anything, into a predictable slaughter?
The matchmaker does not assume you are in direct communication; however, it does adjust your Elos to compensate for some of the advantage that being in a premade team provides.


22 months ago

Quote:
Originally Posted by Traestus View Post
Hey Lyte,

I'm just curious perhaps you can answer this for me. Why does MM sometimes put people sub 30 into my games. I generally solo queue, mostly drafts and every once and a while there will be a single sub 30 in my games. That 30 ends up getting dominated in lane, and which ever team had the sub 30 loses. I'm sure there are plenty of cases where the sub 30 does fine, but I'm more curious as to how this occurs. Is it because they duo queued with someone or is it just there to fill the occasional group if queue is taking too long.

I only ask because I know as a teammate it's not fun getting matched with someone on your team who will almost without a doubt lose their lane due to both a mathematical advantage and a lack of experience. And I can't imagine it's very fun for the sub 30 because they're placed in a situation with very low chances of success where it can be difficult to learn depending on just how badly they're outmatched in lane.
It is typically from a Duo Queue, but on average the matchmaker still tries to find the fairest match possible given a sub-30 is in the game.


22 months ago

Quote:
Originally Posted by PRPatxi View Post
@Lyte don't know if this is helpful info but something I have seen in Ranked(way too many times) is that people with 5 wins are getting queued with people with 50 wins against people with 100+ wins, and the player that doesn't have a lot of exp in Ranked games will just get caught in the jungle and other places which results in his team loosing.

Another observation is that a lot of players start to go ranked as soon as they reach 30, and even worst some players get to 30 just by playing Dominion and coop vs AI then without any sort of experience drop to play ranked .

I think summoners must get certain amounts of Games played or Wins/Creep Score/Kills before starting to rank. This will not solve all problems with queue but will prevent at least a few unnecessary loses for some teams.

Thanks for listening, hope info is helpful.
I've discussed this before the thread, but raising arbitrary barriers to Ranked actually does not solve this problem. If I set the requirements to play Ranked at 300 Wins, you will still see players that start playing Ranked and "shouldn't belong there." This is because we are simply shifting the entire Ranked population a certain direction versus actually solving the problem which is placing players in the appropriate tier as soon as possible.


22 months ago

Quote:
Originally Posted by Fatpapapanda View Post
I don't have any player data as far as pool dept goes, but wouldn't a two dimensional system work for matchmaking purposes? Elo rating and games played both being taken into account. Typically, the best games I've had were against and with experienced players on ranked, at roughly my level.

Increasing the amount of placement matches could also be a good way to properly asses a player's Elo value.
I've been looking into the data to see if number of Ranked wins has a significant impact on game outcomes, but so far I am not seeing anything. I think in many cases, players are just remembering matches where there was a player who was Unranked or playing their first 10 or 20 matches (half of those being Provisional)... and thinking that number of Ranked wins made all the difference. Actually, in those situations, it's just because the system has not acquired enough information about the new Ranked players to place them appropriately.


22 months ago

Quote:
Originally Posted by ongelvin View Post
Bungie Studios released the source code to the matchmaking system for Halo Reach in an open challenge to beta players to break the system. The source code stood up to the test.

Has RIOT looked at this source code yet?
We are constantly exploring other rating systems and trying to upgrade our own system. Yegg and I have many a long afternoon discussion on the pros and cons of each system and how to improve ours...


22 months ago

rivanor,

I have thought a lot about how individual metrics can affect Elo but there are too many abusable cases. If you check back in this thread, we have a lot of discussion about why individual metrics are not a good way to adjust Elos.

For example, in you final bullet point you claim that if a lane is feeding, players would not care as much because they know they can still win their lane and gain Elo. What's the point of ever winning a game then? If a player simply ensures 100% of the time he wins his lane and does not care at all about his teammates and never helps out his other lanes, he theoretically will be the #1 League of Legends player in the game. Is that an OK system? I don't believe so.



22 months ago

Quote:
Originally Posted by yoshiwaan View Post
I think you make it sound so black and white Lyte.

You could have a 50% team ELO and 50% personal ELO. So if you lose, but play perfectly, you don't go down ELO (-6/+6), but comparatively if you win but play horribly you don't gain any (+6/-6). This mean the average amount you climb or drop would be less (as you're unlikely to play perfectly or completely uselessly every game) so it'll be (+/-6)+(+/-6) every game.
The maximum potential gain or loss is the same (so you could adjust the total if you wanted the average to be similar).

It wouldn't just have to be about laning. You obviously already collect a bunch of interesting metrics for your Business Intelligence and post game stats, if you collected a few more you could probably extrapolate out how well someone played. Off the top of my head some interesting markers would be:

-Getting First blood
-K+A/D ratio (combining kills and assists then makes it fair for supports)
-CS
-Placing X wards in key map areas (designed so you can't spam 5 wards in -1 useful place)
-Last hitting baron/dragon (for smite steals)
-Number of blue or red buff/sgame (for good counter jungling)
-Highest damage done to champions (with tweaking around aoe)
-Some sort of damage taken from champions/deaths metric for displaying your survival rate in team fights
-Double/Triple/Quadra/Penta Kills
-Killing champions higher level than you
-First to hit level 18 in game
-Enemy jungle CS (once again for counter jungling)
-Turret hits taken/deaths for lower levels for pro tower dives.

It's hard to tell without know what you track, but you might already have a lot of this!
Apologies, the intent was not to make it black and white. We've just discussed individual metrics numerous times in the other threads and the black and white example was the easiest to point out to the new visitors to the thread.

A lot of the metrics you point out are team-oriented metrics and are things we are doing analysis on to determine if they can be usable for adjusting Elo. For example, there might be ways we can incorporate number of towers destroyed as a team to determine if your entire team deserves less or more Elo.

However, we have done some research on the individual metrics you listed and very few of them are great predictors of success or skill in League. Most of the individual metrics tend to promote gameplay that is not productive to winning the game. For example, getting to Level 18 first in the game does not necessarily predict a top level player--it does encourage some players to sit in their lane and attempt to farm to Level 18 as fast as possible.


22 months ago

Quote:
Originally Posted by VangelisGr View Post
Doctor Proffesor Lyte I have a question for you. Since many teams travelled to Korea for the ogn tournament and are playing on the Koreans server's solo que I was surprised to find out that over they do indeed use individual metrics effect how much elo you win or lose rather than how it works on europe and na servers. I actually agree that it is a bad idea to have those individual metrics so I really dont understand why they would be impletented on another server? Is it that riot has no control whatsoever on what happens over there? Is it that you actually think its good for the korean playerbase but not for the european/na playerbase?

Now I dont know exactly how it works but here are some quotes from scarra's twitter about this :

"Also @dlocust lost 110 elo in one solo q game"
"I just lost 91 elo lol."
http://i.imgur.com/j9oJY.png
"If you ever go 0-3 or more you are losing 70+ elo apparently"
"just lost 93 elo lol... This system is pretty ridiculous"
"my day consists of losing 1 game for 90+ elo then spending the rest of the time when we aren't scrimming gaining it back."
It's not implemented on another server. This is a bug that we fixed in the latest patch actually.


22 months ago

Quote:
Originally Posted by Kazhang View Post
Lyte, question... do you do ranked matchmaking too? ;o
Yes I do, I work closely with Yegg on Ranked aspects of matchmaking. We recently removed F2P Champions from Ranked (unless you own them) and changed the Queue Dodge penalties (to be time suspensions instead of -Elo). We are currently monitoring the Queue Dodge change closely to see what we need to adjust because it is not a perfect solution; however, we are working on the next change for Ranked matchmaking as well.


22 months ago

Quote:
Originally Posted by CrazyGoNuts View Post
Lyte,

Why is it that wins are always worth less ELO than losses?
Wins should not always be worth less Elo than losses. If you have played a lot of games, you will typically get +12-13 for a win and -12 for a loss.


22 months ago

Quote:
Originally Posted by Imclueless View Post
Hey, Lyte. I have another matchmaking question for you. I posted a thread, but it never got a response from the forums, and I'd rather not bump my own thread. So, I'll just transcribe it here:

Hey, I'm hoping a Rioter can answer my question. How does matchmaking work in bot games?

Serious question. Generally speaking, when I queue up for a bot game, it's all level 30's and we do fun things like steamroll with AP Garen. It's quite relaxing.

Today has been different. VERY different. I LOST a bot game today. My team was level 7, 11, and 24. None of them had any idea how to play. The only one who had a reasonable build was level 11, and we (me and my duo) told her what to buy.

Like I said, I love things like AP Garen in bot games. It's a blast, because you can still win if you're not AFK. I did so the other day. But, this wasn't a stupid build for fun. These guys were legitimately feeding the bots, because they were so unskilled. Unfortunately, we couldn't win 2v5. I probably should've just built Orianna into an AD Carry and done it that way, but hindsight is 20/20.

The point is: How do these games even happen? I mean, is there a matchmaking rating for bot games? Because nothing is worse than having to tryhard in a bot game because I got a level 7 who keeps running down mid over and over. When I get a bot game, I want to have fun, or play someone I never have before. That's what they are there for, once you hit level 30.

If matchmaking is totally random in bot games, would it be possible to change that? Match me with other people who can steamroll with AP Garen. Please. This was a very depressing game to play.
There is looser matchmaking for Co-op vs AI games; however, we have recently been working on a few matchmaking changes that will change how the matchmaker uses 'Summoner Levels.' Basically, we've given Summoner Level a lot more weight in the matchmaker, so low levels should see Level 30s much less frequently. No ETA yet on when this matchmaking change goes live.


22 months ago

Quote:
Originally Posted by Arcolyte View Post
Might there be a way to be flagged as a mentor for CoOp vs AI? I usually play 3+ games a week, to help new people, and for the ability to be totally trolltastic vs the bots (ap trundle, crittlesticks and what not.) I understand this would be an unlikely candidate for even something low priority, but figured I would ask since you seem cool with questions.

Thanks for your time.
We currently don't have plans for players to flag themselves as mentors for Co-op vs AI, but mentoring in general is something we have talked about and will probably re-visit in the future.


22 months ago

Quote:
Originally Posted by DiscworldDeath View Post
1. I assume this change will affect all matchmaking, not just for AI games.
2. Does this mean players who smurf, and thus have a really high normal Elo on sub-30 accounts, will have to expect even longer queue times?
Yes, smurfs that have really high Elo sub-Level 30 will see some increases in queue times. I generally prefer players to not smurf and to stick to their main accounts; however, I do realize that some players do it to play with their friends who might be new to LOL. We're going to address the 2nd part of that though and figure out ways veteran players can play with their newbie friends.


22 months ago

Quote:
Originally Posted by Dex7ter View Post
i just lost gold thx to fantastic trolls i just love improve my rating and play with unskiled people i dont know if u have more rating then u have more trolls or what, but this system works like that.

League of Trolls ftw.
Dex7ter, please stay more constructive in this thread.


21 months ago

Quote:
Originally Posted by Dex7ter View Post
the problem is that i have to carry all the time people and i tired to do so, every time u have to carry it dosent matter if u are 1200 or 1500 or even 1700 if u want to be higher in this leader u have carry people who dont know how this game works, and this realy problem to me when i play so many games and at 1500 have person with 20 wins who was carried by elo bust or better players, and now i have free lost or amumu who afk at blue in the jungle and died like boss, after next death he go afk after 3 min of gameplay at 1500 elo.

Bust can help only 5% players to find they true elo, when they they are very good at this game rest just have this luck to be carried, and i dont have the luck to play with normal people lests say that 80% people dont use brain. Mosto of them just random pick champions on go in.

Maybe i should duo or cheat with this system to get my true elo, but what is the point to do so when this system is full of bugs and most of them use duo like 1800 and very low person to get better rating and low skiled players. There no restricion at ranked, and this realy said part, when there are no rules, when u guys are lazy to do few simple steps, and fix duo quo or bust system or mm to have better place to play.
I have been doing a lot of research on how long it takes players to reach their true Elo in the current system and the majority reach a good plateau in about 200 games. For some, it takes up to 300 games. After 300 games, you really need to significantly improve before you can lift your Elo higher.

Remember, your "Top Elo" is not your skill level. It doesn't matter if you hit Gold once, or hit Platinum once. That's your upper bound. In a matchmaking system, you have your lower bound, upper bound and average and the average is actually your "true Elo." The bounds represent a system's confidence level and unfortunately, your true Elo is probably not Gold or higher.

In saying this, I think 200 and 300 games is too long for players to reach their Elos and Yegg and I are going to continue upgrading this system until those numbers go down significantly.


21 months ago

Quote:
Originally Posted by Baccani View Post
4. Refund RP to permabanned players. Show them that you aren't out to take advantage of them. Toxic players aren't your playerbase and you don't want their money or time.

This one is extra tricky. You know full well that some jerks will play and spend money on the game, but I think that if you can't afford to refund their money, then at some level you DO want toxic players in your game. That is a discouraging thought.

That's all for now. I know I'm pretty much a nobody and have no credentials, so I want to thank you a lot for taking the time to read this.
The trickiness with the last one is... if players know they are going to quit League of Legends, do you want these players to intentionally be toxic over and over until they get permabanned so they can get a refund?


21 months ago

Quote:
Originally Posted by Aparkhurst View Post
What does this mean to a smurf that gets matched EXCLUSIVELY with/against level 30 players? Will it just be a queue time multiplier for the same quality of games, or will this put you with low level players again?
You'll just wait a bit longer to see if we can find lower level, high-skilled players similar to yourself, or you will break out and still match against Level 30s after a few minutes in the queue.


21 months ago

Quote:
Originally Posted by WaveofShadow View Post
I'm watching a game right now that Salce is playing with Crs Elementz, KingWills84, GotSwag and AdzyBoltz. (I think the first 3 trio queued....I think Gotswag is a crs cop smurf)

Their opponents are all sub level 30 after having looked them up. How is this possible?
We also have to keep in mind that at some point in the matchmaking queue, the matchmaker decides to keep expanding the search parameters until a match is found.

So, let's say that there was only 10 players in the entire world playing League of Legends and one team was 1200 Elo and one team was 2100 Elo. For the first few minutes, the matchmaker would try to find a fair opponent for these teams but eventually it will shift its priorities and just try to find a match at all.

So when players say, "How is this match possible?!?!?! Is the matchmaker broken?!" In many cases, players might not have all the information. For example, how long were these teams in the queue? What time of day was it? Were these teams both premades of 5? These all can have an impact on why a particular match was created.

The existence of these 'rare' mismatches does not mean the matchmaker is broken... they are rare afterall.


21 months ago

Quote:
Originally Posted by Ledtim View Post
Would it be possible to display summoner levels during champion selection screen just for Co-op mode?

Unlike PVP where the difficulty is based on your previous performance and you can expect a certain level of challenge no matter what, the difficulty in co-op mode varies heavily on your teammates.

And sometimes, you get a team of entirely lvl 30s, and sometimes you are the only lvl 30 in a team full of people who are new.

So knowing the summoner level of your teammates during champ selection would allow you to know how difficult the match would be, and whether you should play something like AP pantheon or something viable.
Yegg and I were talking about this fairly recently. We probably won't reveal Summoner Levels because some players will likely be abusive/rude towards players they don't feel like they should be playing with; however, we are considering some level-based (but not Elo-based) matchmaking for Co-op vs AI. We're currently collecting data and doing some analyses first.


21 months ago

Quote:
Originally Posted by isobold View Post
Don't you do that already? I had the impression you would. When I was playing Coops on my Level 13 Smurf, I would get matched with Level 1-10 fairly frequently and hardly ever met a Level 30. When I do the same on my main, I get 80%+ Level 30th and rarely ever a real Newbie.
Basically most game modes have the standard Elo-based matchmaking with a slight adjustment for Summoner Levels. We want to do more matchmaking upgrades that are specifically tailored to the different game modes; as an example, we are working with Brackhar and Nome to specifically improve Twisted Treeline matchmaking. The matchmaking changes applied to that game mode will not apply to Summoner's Rift matchmaking.

Yegg and I were discussing what else we want to do with Co-op vs Bot matchmaking and we're focused on how we can improve level-based matchmaking there.


21 months ago

Quote:
Originally Posted by Guttler View Post
Dr.Lyte,

I've heard that the 'elo' system league of legends uses is so modified from the original elo system that they are basically not the same system anymore. So why are we still calling it elo? Can we call it something new?
We probably should, Elo is just a convenient way to say "matchmaker rating."


21 months ago

Quote:
Originally Posted by Ledtim View Post
Wouldn't separating lvl 30s from lower level players simply make co-op too easy for lvl 30s and a bit harder for the lower level players?

When I play co-op most of the time, I get matched up with lvl 30s most of the time, and i don't really get enjoyment or challenge out of it unless I go something completely stupid like ap pantheon.

I don't mind getting teamed up with some newer players in co-op since I think it honestly makes the mode a bit more fun for me, but it'd be nice to know ahead of time if I might have to actually try a bit by knowing other players' levels, or if that's too revealing, maybe the average level of the summoners on the team.
As with most matchmaking solutions, there's rarely a perfect solution. We simply have to try to do upgrades that target as much of the playerbase as possible.

With Co-op vs Bots for example, the key is intention. Some Level 30s don't want to go into Co-op vs Bots and carry new players; instead, some Level 30s want to go into Co-op vs Bots to try wacky builds or try out new champions (and of course, still win in the process). However, for most new players in Co-op vs Bots, they actually are still learning the game and most of the items, champions and abilities are brand new.

When you merge populations with different intentions into the same game you often get a negative experience. The new players are pressured by the veterans to 'know' more about the game than humanly possible and the veterans don't understand that these players are actually brand new and just want to sit at the store and read item descriptions.

So when it comes down to it, we'll be trying to figure out which upgrades are best suited to matchmake players with the same intentions into the same game.


21 months ago

Quote:
Originally Posted by BelligerentGnu View Post
Lyte, how goes the new dodge penalty experiment?

And any updates on your work for us.
The first 2-3 weeks after a patch are really noisy in terms of AFK/leavers/queue dodges/reports. We often see spikes in activity in all of these after a patch, so we don't have stable data about the queue dodge changes just yet.

I'll probably start taking a look in ~1-2 weeks and as soon as we get the data we'll be able to quickly make some iterations if necessary. Yegg is keeping a closer eye on it then I am though, as I am currently polishing off 4 matchmaking upgrades.


21 months ago

Quote:
Originally Posted by Archparagon View Post
I know this is a little off-topic but why isn't there a co-op vs Bots mode for Twisted Treeline? The reason I ask is my wife started playing recently but my brother and I are both lvl 30. What we wanted to do was do a Co-op vs Bots match where I could sit on Bot lane with Alistar and heal my wife as she learned to play and my bro could just do whatever in Top Lane. The idea was for us to be able to give her tips and let the game go on as long and my brother and I wanted so she could get a good experience playing lanes with out hindering/taking time from other players.
In Summoners Rift I can just sit there and stall because it would be unfair to the other players who wanna win and get XP/IP. At the same time I cant help my wife learn game stages, laning and last hitting as easily. Its her first MOBA and she doesn't learn video games very fast.
It seemed like learning was part of the thread so I went for this.
rjcombo and I are currently doing research on non-Summoner's Rift game modes to see what kinds of players are playing and whether we are satisfying all the different needs of these players.

Unfortunately, I don't have anything to add about why there are no bots on Twisted Treeline but we'll take a look at the demand for such things.


21 months ago

Quote:
Originally Posted by Zielmann View Post
What's this mean if you patch every 2-3 weeks
Haha well, without going too science-y, measuring the effects of changes in a patch is a tricky endeavor. What we need to do is measure the 'impulse' or impact of the patch itself and try to separate the impact of the patch from the impact of the features. On Day 1 after a patch, the variance or "noise" is so overwhelming that it's super hard to find the actual impact of the features (the "signal").

However, if we aggregate the data from Day 1 after a patch to Day 16 after a patch (but before the next patch is deployed), we have a lot more data to try to pinpoint the signal before the next patch comes out and adds noise into the system again.


21 months ago

Quote:
Originally Posted by Zielmann View Post
What's this mean if you patch every 2-3 weeks
As an addition, interestingly there are features like the [All] Chat Change that had dramatic positive impact that even immediately after the patch, the signal greatly outweighed the noise making the analysis super obvious.


21 months ago

Added 6 items to the "In Progress" Section.



21 months ago

Quote:
Originally Posted by rivanor View Post
Dear Lyte,

Thanks for your attention to this thread.

As you have pointed out in the past, it takes 200-300 games for a person's ELO to stablize. With that in mind, I am not sure how this game I just played could be considered a fair matchup for the losing (enemy) team.

They have 3 players who are clearly on their way down the ELO ladder and have not bottomed out yet. In effect, this means that their ELO is unpredictable/unknown. Yet 3 of them are all put on the same team, leading to disastrous results. This doesn't seem like something that would be too costly to fix for your developers (I am a professional dev which is why I say this)

thanks for your consideration.
Actually, there is very little evidence that those 3 players are on their way down. Any single data point isn't necessarily predictive of their eventual skill ratings. For example, why is a 5/5/3 Yi on his way down? Why would a 4/5/1 Tristana be on his way down?

Even if you are looking at the 0/5/4 Twisted Fate, have you never had a similar game statistically? If we see games like that are your match history, are you on your way down?


21 months ago

Quote:
Originally Posted by rivanor View Post
I say they are on their way down because they started at 1200 ELO and are being matched against me (1050) so the only way I can see that they would be matched with me is if they have lost most of their games to this point.

I still have them in my "recently played" list. The 0/5/4 twisted fate's ranked history is:
3 ranked losses
2 normal wins
1 leave
4 normal losses
0 ranked wins

The 5/5/3 yi's match history is:
3 normal losses
1 ranked loss
1 leave
5 normal wins
0 ranked wins

The 4/5/1 tristana match history is:
3 normal wins
2 ranked wins
4 ranked losses
1 normal loss


I can use this data to either say there is not enough data to make a conclusion (which is what I gather you are leaning toward) or to say that the current trend may suggest evidence that these players may have more descent ahead of them.

At any rate, since I think we can agree that these three players' target ELO is unpredictable at this point, why does it make sense to put three unpredictables all on the same team? Would that not increase the chance of a highly random experience for both teams? Isn't the goal of matchmaking to create an environment where a player can have a reasonable expectation of the skill level of his teammates and his opponents?

I guess what I am saying is since I am 1050 ELO, I fear that I am going to be routinely matched up and against players who are "on the way down." I've tried to play a lot of games to get a more consistent experience, but it is frustrating to think that I am "doomed" to be subjected to this scenario since I am "too close" to the starting 1200 ELO. Then again, dexter's rants make me think that maybe the problems I am seeing are present everywhere... something to ponder.

anyway thanks for your time.
One of the main problems with suggesting that those players are "on their way" down is based on statistical probabilities.

Let's say for every player, I just rolled a 6-sided dice 10 times (3 or lower is a loss, 4 or higher is a win) and that's their 10 game match history. Theoretically, that's a 50/50 win chance and means the player has stabilized.

If I rolled this dice 1 million times to represent 1 million players, tons of players would show the exact same ratios you have shown for those 3 players. Does that mean players are on their way down?

However, it is more accurate to say that the system is less confident in the ratings of those players than other players with 1500 games. But, it gets more complicated than that because when a brand new player joins Ranked Mode the best assumption is in fact that he is 1200.

Let me put it this way: you won't be stuck in an Elo range because of these "new" Ranked players. I've researched that bit time and time again and it has never been true.


21 months ago

Quote:
Originally Posted by Dex7ter View Post
again lost

2new players duo op
In that same game, a player with 136 wins is 9/6/7 and a player with 230 wins is 0/6/11. On the other team, there's a 29 win player that is 4/3/22 and a 156 win player with 16/3/7.

Your reports are now worth nothing to the Tribunal. Please stop falsely reporting in every single game and being a negative force in your games.


21 months ago

Quote:
Originally Posted by ElementzPRO View Post
The TF is 1/6 in ranked with an average KDA (not KD, KDA) below 1

I'm not gonna bother checking the others but you can't seriously say he isn't on his way down.

(his name was domhan unless I'm mistaken here)

BTW I honestly don't think I've EVER seen someone with an average kda below 1 before in my hundreds of ranked games. I would be mad too if I was matched with him.

His kill to death ratio is 1 to 10.5 which is on the level of intentionally feeding
My point wasn't that those specific players are or are not going down. My point is, screenshots don't paint the full picture.


21 months ago

Quote:
Originally Posted by rivanor View Post
If what you say here is true, then that resolves my concern. It means that if I play enough games, I become the master of my own fate. I think I am probably not alone in this concern.
Yes, you are definitely the master of your own fate Usually by about 200 games, players will have settled into the appropriate Elo tiers. For a very small percent of players, it takes about 300 games.

One thing players have to remember is, your top rating is not an indicator of your skill. In fact, for the average player, their true Elo is typically their topRating - 150 or so.


21 months ago

Quote:
Originally Posted by Xeanar View Post
I was thinking about this one for a while... I have no idea if it could be done or something but..
The ELO you gain or lose could be determined by how many towers and big objectives each team had taken.

For example: A match where the amount of objectives taken (Baron, Dragon, Turrets) are close to equal should have a lesser ELO loss for the losing team and a little higher gain of ELO for the winning team.
While a one sided game should perhabs keep the ELO system as it is at the moment.

This is just a little idea out of my head, hope it helps in anyway
We've talked about objective-based Elo adjustments but we have to be careful about encouraging or discouraging certain meta compositions. For example, if the Elo adjustments are too high for objectives, it greatly encourages players to do push compositions where they push objectives at all costs. In the extreme silly case, what if killing Baron gave you +10 Elo regardless of win or loss? This would force players to heavily base their play around Baron, possibly doing multiple Smite compositions or even make Nunu far more valuable to teams.


21 months ago

Quote:
Originally Posted by Aliothale View Post
@ Lyte, please explain to me how this wouldn't work and solve a lot of problems?

If players are required to have 16 champions in order to play ranked. Why not just make it even more fair by requiring at least "XX" amount of WINS with EACH of those champions in SR PVP Normal queue.. Draft or Blind.

Players should have incentive into owning 1 account that they want to play ranked with. The best way to curb trolling is to make it a complete waste of time to do it. So...

25 Normal Queue wins for 16 champions would even out to 400 Normal queue wins. Is that not a fair number of required wins that players must achieve before throwing themselves into a competitive environment?

I know for a fact my normal queue ELO is a pretty good number. I refuse to play ranked games in a system that is unfair and punishable because matchmaking couldn't do it's job correctly. I'd rather grind out in blind/draft pick than play ranked games. I hardly even run into too many trolls in blind pick with my 900+ normal wins nowadays. I shouldn't have to deal with that twice, normal queue is for practice.. ranked is for serious players. I've just spent so much time practicing.. that when I get to ranked games it's a freaking joke. I blew through my first 15 ranked games with only 2 losses. Every single game after that I experienced a leaver, a feeder, an afker, or a mismatched newbie. There aren't very many hyper carry's that can turn the tide in ranked, and there is no guarantee I would even get to pick a hyper carry to get myself up in the brackets.

You need to distinguish your players commitment to the game before allowing them to partake in the competitive environment with other serious players.
I've said this before, but changing the pre-requisites to playing Ranked doesn't fix anything. You'd just come back and post about how people need XX + 100 Wins before playing Ranked after the change...

The point of Ranked is not to restrict players from playing with arbitrary numbers of wins. Everyone should be able to play Ranked--we just need to place players in their appropriate tiers faster so that a 500 Elo player actually gets slotted quickly into the 500 Elo bracket in < 10 games.

Also, Normal Elos tend to be inflated and about 150-200 Elo higher than what your Ranked Elo should be. You should be more constructive in your threads by the way.


21 months ago

Quote:
Originally Posted by UpDownLeftRightAB View Post
Ok, honest question about how the matchmaking system works. Does the system try to match it so that your win/loss has the best chance of being 50/50 or so that both your opponent and allies are of the same skill level. Since I find that whenever I go on a winning streak, my teammates seem to get worse while I would judge that my opponents are at my skill level as myself.
No. The matchmaker does not try to make people have a 50/50 win percentage. The goal of the matchmaker is to be as accurate as possible in predicting any given match.


21 months ago

Quote:
Originally Posted by Just me Soraka View Post
So... you're saying the matchmaker DOES know if I'm likely to win or lose a game?
Yes and it uses the information to adjust Elo gains/losses which is why sometimes you might see -10 and sometimes -13.


21 months ago

Quote:
Originally Posted by BelligerentGnu View Post
Out of curiousity Lyte, how many people are on your team? Have we met any of them on the forums?
About a dozen. In general, Team PB&J introduces developers when they worked on a feature that is being released so hopefully players will meet some more soon Devs that have worked on the new Tribunal features have talked a bit on the forums over there already.


21 months ago

Quote:
Originally Posted by Ymris View Post
Lyte, out of curiosity, which of these do you feel contributes the most/least to toxic behavior in games?

-People who genuinely fail to understand that their behavior is inappropriate. (i.e. "this is how a competitive gaming community behaves, man up or get out")

-People who have come to assume that because "everyone else" does it it must be OK (i.e. "there is always a person raging and doing x/y/z in my games, they don't seem to go away, so that must be fine to do)

-People who know that their behavior is out of line, but figure that they can get away with it/want to see how far they can push the envelope.

-Some other group/s of toxic players.
I find that a major source of the problem is that players think it is OK to behave a certain way because they have done so in other games in the past; to these players it seems ridiculous that we do not tolerate such negative behaviors in League of Legends.


21 months ago

Quote:
Originally Posted by Letric View Post
You said it takes around 300 matches to get to your true elo. Are these 300 matches the normal "getting to my elo" time considering a constant value for Skill in playing Lol?

I mean, if my Skill to play this game is always 1, then it makes sense that after 300 games I'm at my elo. But since human beings usually learns from their mistakes, guides, streams and in depth analysis of champs, how can those 300 games be accurate.

I might believe that considering you can always learn, there will be a % of players who will keep increasing in their skill, would that cause to reset those 300 matches?
How true can be your elo when you're actually improving on every match? Is there any analysis being done on the rate of how players improve?

True elo is a 100 gap? 200 gap? 500 gap?
The 200-300 games metrics takes into account the natural improvement a player gets playing those games.


21 months ago

Quote:
Originally Posted by YoQuieroPancake View Post
IDK it sounds good and all... but this doesn't take into account allt eh times that you get a dc'er a mismatched duo partners on your team, trollers, leavers. i have lost countless games do to these events. IMO, Elo should be left for 1v1 play NO ONE man can win you a game. if one fails hard ussually the rest fall of and lose as well. Leave Elo for 1v1 games. Like Mortal Kombat on-line. Perfect place for Elo based rankings.

Edit: Bring on the down votes
The 200-300 games does take into account the random DCs/AFKs. That's why the number is so high... which is something we are trying to fix.


21 months ago

Quote:
Originally Posted by Goumindong View Post
This doesn't make any sense. Its the same player base and the same matchmaking system, just one has a number next to it right?

Shouldn't normal ELO's be similar to ranked ELO's ?

Only two ways I can see that not happening would be if dodging really did remove a bunch of ELO from the system or if a significant proportion of the less strong players in normal ELO are not playing in ranked games. But I am not sure that either of that is true [and anecdotally from looking at lol matches data from when normal ELO was available from players I know their normal and ranked ELO's tend to match pretty well]
But it's not the same playerbase--in fact, a vast majority of LOL players never play Ranked.


21 months ago

Quote:
Originally Posted by YoQuieroPancake View Post
... then i guess players like me who cant seem to get out of the bad elo brackets aka Elo Hell should NOT be playing Ranked games huh cause it's not meant for us? Oh the PAINFULL thruth is PAINFULL!!! *Goes to cry in a corner* :'(
Haha, of course not! Ranked is for everyone. I was just mentioning that many players don't play Ranked because they don't want to.

I believe that everyone should be able to play Ranked if they want to, we just need to place them in the appropriate Elo tiers faster.


21 months ago

Quote:
Originally Posted by Ewendor View Post
As much as I like LoL, I do beleive that the WoW arena rating system is significantly faster at getting players into the correct brackets, and further, it makes the people in the lower brackets feel better because they climbed to their rank instead of falling to it.
I agree that the feeling of 'climbing up' is better than falling down from 1200 to your true Elo and this is something we are trying to address.

However, the reason other rating systems are 'faster' at getting players to the correct bracket isn't because of the system. It's mainly because the game mode is different--you can't queue up with random players, you have to always be a cohesive set of players with a group of friends so there's less inherent noise compared to playing with random players every match.


21 months ago

Quote:
Originally Posted by Brodhi View Post
Although it is true that the people who get Gladiator in WoW are probably a tight-knit group of friends, I can confidently say that most of the people who are 2.2k rated or less in WoW were random people who picked each other up in trade chat (or paid someone else to carry them to 2.2k).

It would take, on average, about 30-50 wins to go from 0 ELO to 1200 in my model, then an additional 25-40 to go from 1200 to 1500. This is substantially faster, I feel, than the hundreds upon hundreds of games that you currently need to play now in order to slowly climb to the 1500 bracket.

The biggest problem right now is that people who start rank today are at a disadvantage compared to people who played rank on the first day of the season--all the "good" players are already in their correct ELOs by now, so the only people left to play with at the entry-level are the trolls and people who's true ELO is 900-1200. It is incredibly hard to progress when your system has no way to separate progression from the trolls--my suggested system does, because there's no way for you to "fall" into the troll bracket, you can only progress out of it.
The key wasn't that people in the WoW system are friends--the point is the same 'pairing' plays together every match which means there is inherently less noise trying to predict their skill rating. Imagine if we had to predict the skill rating of a player when every single game, he was paired with a random player.

Secondly, it only takes about 12 games to hit 1500 Elo since there is a seeding system in play just like in other games.

Third, it's actually much harder to climb out of ~1200 Elo at the start of a season... this is because everyone is clumped at 1200 at the beginning of a season so the vast majority of players are in the wrong bracket and every game is extremely noisy and hard to predict.


21 months ago

Quote:
Originally Posted by Brodhi View Post
I see where you are coming from now. But at 2.2k+ ELO, you tend to see the same "pairings" regardless. Hell the top 20 players in the world right now almost exclusively play Purple side because you guys place Higher ELO on that side. Then there is duo-queueing in Ranked, which of course right now is ****-near the only way to get out of 900-1100 without winning hundreds of games and trying to dodge trolls.

Why does Riot have to "predict" the outcome of games? No other eSport does this in the world. Why do we -have- to have a Matchmaker system determine how much ELO we get because it "think" a game will be a stomp or a close game, or whatever.

Why can't there be a set amount of ELO you get for a win and a loss, and just queue people of similar ELO in together.

In response to your "seeding" statement, you yourself already stated the seeding process can be unfair because people who get lucky and have players who have 300+ wins can be placed TOO high, and players who get a bunch of 1-3 wins can be placed too low. If the latter happens, it is incredibly hard to get out of 900-1100 ELO. You can't deny this. There would't be talk of an "ELO Hell" if it wasn't.

At the start of a season there are far, far less trolls, because it is a fresh start for everyone. I am not saying there are no trolls, or that the troll numbers decrease by 99%, but there are substantially more trolls the longer a season progresses because at some point in someone's LoL career they say "**** it, I am stuck at 900, might as well make everyone's life miserable".

Also, you guys talk that the queue times at 600 ELO are just as high as 2.6k ELO. WoW's system removes the extremely long wait time of the lower rated brackets because you can queue 0-1199 together all in one group, since they will lose 0 rating.
I think you might have a misunderstanding of how matchmaking systems work. Almost every game with a matchmaking system tries to predict which team might be the winner statistically... in fact, I'm not even considering just eSports but many real sport ratings do the exact same thing. Not having set amounts for a Win or Loss is also one of the reasons all the matchmaker systems that do use granular Elo adjustments are more accurate too. For example, if a 0 Rating player beat a 2200 Rating player, should he still get 40? What if a 0 Rating player beats a 1100 Rating player, still gets 40? Or if you want to specifically look at games within a 'bracket,' isn't a 0 Rating player beating a 1199 Rating player more meaningful than a 0 Rating player beating a 400 Rating player?

Also, for a 1500-1800 player, it only takes a dozen games or so to climb out of 900-1100 Elo so I'm not sure where you are coming from about "Elo Hell." I also never said it takes 300 games to get to true Elo because you got players with 1-3 wins during seeding... or that players are placed too high because of players with 300 wins during seeding...

Confused.


21 months ago

Quote:
Originally Posted by Metronomotopoeia View Post
Actually, you could use the same logic to say that it's easier to climb at the start of a season, because half the population is fighting teams they'd never otherwise be against, which results in super quick stratification.

For example, a 2200 team can get matched against several sub-1600 teams, which would be practically free wins. By the same token, 900-tier players would lose almost every single match, as opposed to if they were just fighting straight 1100-1200 teams, which would have some opportunities for back-and-forth given the lower relative skill difference.

It goes back to the earlier point about it being a committed, stable team. It'd be different if the 2200 team in question were 3 players with wildly different individual/"real" ratings that just averaged out, but no one ever really made those kinds of teams. In other words, you'd never find two 2400 players who would be willing to hang with a 1800 player on their team because he was a super cool dude.

Not disagreeing with your line of thinking or philosophy about matchmaking, but I think this point in particular worked the opposite way in reality. To the system, things definitely seem noisy, because it's trying to put together even matches. But to the goal of getting people to their ratings quickly, it worked pretty well.
This wasn't an opinion :P It was just data. You have to consider that statistically the outcomes are almost the same, except the matches are more lopsided due to more noise. This is because at any point in a season about half the players starting in Ranked should be above 1200 and half should be below, so at any point in the season the vast majority of players are playing against players outside their brackets until they are placed appropriately. At the start of a season, however, almost all of the top players are bunched at 1200 and climbing out. At the middle or end of a season there's far fewer Ranked players starting who should be at the very top of the ladder (because they all jumped in early), so there's slightly less noise at the 1200 range.


21 months ago

Quote:
Originally Posted by Brodhi View Post
Clearly, you have a ton more metrics than we the Community have, so all our ideas, comments, and thoughts are based off our own individual experiences and not what the numbers say. Go check out a Streamer that is in the 600-900 ELO. Tell me how well he does in "climbing" out after a dozen games. In fact, I don't even know where you think a dozen games gets you from 1100 to 1500? The LoL Matchmaker has it so that, in general, you will more than likely accumulate a 50% W/L ratio. Using this idea, you will only gain about 36 ELO from 6 wis / 6 losses (assuming 13 ELO for a win, 10 for a loss because the Matchmaker put you up against "better" players for getting on a win streak or whatever). Even if you won all 12 games, that would still only be 156 ELO (if all games gave 13 ELO), which only puts you at 1256. Can you explain what numbers you are using to figure out how a 1100 player can get to 1500 from a dozen games?

I 100% think that a 0 rated player beating a 200 rated player or a 1199 rated player should give the exact same ELO amount, because for all intents and purposes, they are NOT RATED. You don't get rated until you break 1200 ELO, and as such it is unimpressive for a non-ranked player to beat another non-ranked player. Now, if a 0 rated team beats a 1250 rated team ( how this happened, I do not know), then sure, award the 0 rated team like 140 ELO, because a non-rated team beating a rated team is a big deal--its an upset.

The same is so true when you gamble on the spread in a Football game, for example. You don't bet money on who is going to win, you bet that a team either will win by X amount, less than X amount, more than X amount, or will flat out lose. The predictions they use come directly from HOW THOSE TEAMS HAVE PLAYED AGAINST EACH OTHER IN THE PAST. It is very rare for the same 5 players to play against the same 5 players over and over in LoL, as such it is highly inaccurate to determine if one Mid-Lane opponent is going to beat another, or even if a team is going to beat another, without any data about these exact same teams playing against each other before.

Confused.

Edit: What I was saying is that you said it takes 300 wins to get to your true ELO. Therefore, using my understanding, a player with 300+ wins at 900 ELO belongs there. When a player first plays ranked and they are on a team with a bunch of people with 300+ wins against a bunch of players with 1-3 wins, the team with 300+ wins is more than likely going to win, because that team is more proficiency in the "900 ELO mindset" as well as having a lot of experience in the game.

3) Provisional Matches in Ranked
When a player joins Ranked for the first time, the system starts them at 1200 Elo and begins their "placement matches." Unlucky strings of losses or lucky streaks of wins can propel a player into an Elo tier they do not belong in. Alternatively, players generally do not like playing with or against players in their placement matches and seeing they have only 1-9 wins.

^ That is you saying the seeding process can place someone in a bracket they do not belong in. This has a LOT to do with during seeding being placed on a team with a bunch of people who have 1-3 wins against a team with a bunch of 300+ wins. Data can show you this.
Ah, there are definite misunderstandings about how matchmaking works.

1) For the 600-900 Elo Streamer, he should only climb out if he actually is a top tier player.

2) The matchmaker does not try to put players at a 50/50 win ratio. It just tries to place players as quickly as possible into their tiers. For players who are not 1200, they most definitely will not be going 6 Wins/6 Losses at the 1200 bracket.

3) If you actually won all 12 games at the beginning of Ranked, you would actually get about ~500 Elo. This is because we have advanced seeding algorithms that will give a lot more than +14 Elo when a player is climbing out of 1200 range making it actually very easy to get within 100-200 of your true Elo in a small number of games.

4) Finally, no data so far has shown that # of Wins has a huge effect on skill. As far as statistics go, a player with 300 Wins and 1500 Elo is about the same as a player with 200 Wins and 1500 Elo. However, if you keep randomly selecting 300 Win, 1200 Elo players and put them against 0 Win (and therefore automatically 1200 Elo) players... you actually will see them split their matches 50/50.

5) Unlucky strings can put you at brackets you don't belong in but it has nothing to do with players with 1-3 Wins or 300 Wins. It has to do with things like even if the match is an expected 50/50 match, you could dice roll and lose every game. Or, it has to do with connection issues. More importantly, even if you are placed in the wrong bracket, you are placed in the right direction within 25 games because of the advanced Elo adjustments as you play the ladder that give you more than +/- 14.


21 months ago

Quote:
Originally Posted by Brodhi View Post
1) I am not trying to tell you the skill level of that particular Streamer, what I am trying to do is to get Riot, and specifically the department within Riot, to sit down and actually watch how these games unfold. Read the chat (because you can't in Spectate), learn the champions the 600-900 ELO players play, etc. I want you guys to legitemtly BECOME a 600-900 ELO player and use that to understand why people think there is an ELO Hell, and why so many people think it is incredibly hard to climb out of 600-900 once you get placed there. Riot has been denying ELO Hell saying, "if you are good enough, you will climb", but the problem there is that how long does that take? If it takes 500 games played is that fair? Shouldn't we have a system that gets them there faster?
Many Rioters and many players have actually intentionally dropped down to 300 Elo and climbed their way back up. They actually had 80% or higher win percentages and climbed back to 1800+ in less than 300 games. In fact, I saw a record of 180-20 from one of our top players...

If a player is high Elo, drops to 300 Elo and actually takes 500 games to climb out, we have a big problem; however, it's easy for them to climb out and players and Rioters have tested this repeatedly.


21 months ago

Quote:
Originally Posted by Brodhi View Post
Where's the video? Where's the data? I am not calling you a liar, but it is easy to say "many players and Rioters have done this, with this result." but I want actual raw numbers to show me how many games they played, what champions they played as, and the average ELO of their teams vs. the enemy team.
Please just search the forums, lots of players have tried this experiment. If you can read Chinese, you should check the Garena-Taiwan forums as well as many of their players tried the "Elo Hell" experiment and climbed out fairly easily.

Very, very, very few players are 'trapped' in an Elo they don't belong in--especially if they have played at least 100 games. In psychology, you learn quickly that on average, players tend to think they are above average. This personality flaw applies in all aspects of life. For example, in a study at a local business, the average employee rated their performance as above average--statistically, it's impossible for everyone to be better than average. This doesn't mean you are 'trapped' in a lower bracket.


21 months ago

Quote:
Originally Posted by GnomeDigest View Post
Sure, but what about the players whose true elo is around 1200? Why should they have to eternally suffer the worst matchmaking the game has to offer due to all players entering ranked in this range?

If your true elo is around 1200 you are essentially "trapped" in the worst matchmaking zone.
Regarding players who actually are 1200 Elo, we're working on a few things that will try to address the extra noise that bracket gets due to new Ranked players.


21 months ago

Quote:
Originally Posted by Brodhi View Post
How many is "a lot of players"? What percentage of these players succeeded compared to those who failed (got stuck below 1300)? What role did they play? What was the average ELO of their team?

If Rioters have done this experiment, it should be easy for you to get those numbers. And have you not considered that a lot of people dipping down into 600-900 only to rise back up as a reason the losing team keeps getting stuck at that ELO?
As another player said, this might be interesting for you to read:
http://en.wikipedia.org/wiki/Dunning...3Kruger_effect

As far as analyses go, almost no players are stuck in low Elo if they don't belong there. If we focus solely on players who have played 300+ games, you're looking at less than 1% of players who are stuck in a low Elo if they actually are a high Elo player.

I am sorry this is disappointing, but most players actually belong at the Elo they are currently at. Statistically, this is very true after 200-300 games.


21 months ago

Quote:
Originally Posted by Brodhi View Post
I appreciate the responses Lyte. Let's hope that you guys figure something amazing out on the drawing board. =)
We do agree that the current system has quite a few weaknesses though. In this thread, we've been addressing ~10 or so different issues with the current system. After the 10 issues are complete, we'll have at least 10 more... I don't think we can ever say we are 'completely done' with the matchmaking system.


21 months ago

Quote:
Originally Posted by BelligerentGnu View Post
Lyte, could you talk about this guy?

http://na.leagueoflegends.com/board/....php?t=2366194

I don't care about the case specifically, but he's saying he got his permaban reversed - which you're saying you no longer do.

Confused, I am.
It's just super rare. All cases that are reversed are heavily monitored by Player Support.


21 months ago

Quote:
Originally Posted by simpsonboy77 View Post
@Lyte,
I sent and email in and did not receive a response. Do you read all of them but can't reply to all of them, backlogged a lot, or something else?
I generally read all e-mails but can't reply to them all personally. Most of the time, I try to answer a thread or question that answers frequently e-mailed questions since it's the easiest way to get more exposure on the responses.


21 months ago

Quote:
Originally Posted by DiscworldDeath View Post
Hey Lyte, a couple of questions about the new patch:

1. How were these things changed?
2. The "Must confirm to go into champ select", is it going live this patch, or had this been pushed back?
The AFK feature before Champ Select will be turned on at a later date, but it won't necessarily have to wait until the patch after Zyra. We needed to do some additional testing on the feature.

For the matchmaking changes, I can't go into too much detail but basically the matchmaker now gives more slack when higher Elo players try to play with lower Elo friends. Before, the averages would pull the lower Elo player quite high resulting in pretty difficult games; however, now the matchmaker will be a bit more lenient on the lower Elo player.


21 months ago

Updated the thread, moving a couple items to the "Completed" section.

By the way, he's pretty elusive and mysterious, but all the matchmaking changes in the current patch were completed with the help of NinjaCoder. We should thank him for putting in extra time to make sure these came out quicker



21 months ago

Quote:
Originally Posted by hugocugo View Post
"4) AFKs in Champion Select Lobby [7/19/2012]
Players who queue up for a match and then AFK force players to wait through the timers until the lobby boots the AFK players out. Alternatively, some players AFK and get assigned a Random Champion, which is also a negative experience for many players. In Ranked Games, Captains tend to see what the bans/picks are like and be able to react accordingly in the next lobby as the players will generally be the same."

How is this completed? People still AFK and random, I don't understand.
This feature will be turned on at a later point, just not immediately after the patch goes live. Some of my developers will talk about this feature and what it does when it gets turned on!


21 months ago

Quote:
Originally Posted by U2Q SafetyX View Post
Something I have seen an issue with is the ranked 5's ratings. I find it ridiculous that people can make a brand new team and be in the top 1000 with just 5 games. While on the other hand, teams play upwards of 100+ games and have a high win percentage, yet are ranked below a team that has only played 5. I cut out a picture of the rating system around our team.

If you notice in the picture, our team (U2Q) is ranked 1567 ELO with 49 wins and 34 losses. Which means we win about 59% of our games. However, if you look a bit lower at "PuntingBabies" they are ranked 1567 ELO (same as us), yet have only played 9 games and won 4 of the 9. Thus, they win 44% of their games. So if we have 15 more wins than losses and they have 1 LESS win than losses, why on earth are they the same ELO as us? I find it disturbing that we could recreate our team, and within 9 games be at the same ELO that took us 83 games to do before.
Hey SafetyX,

There are systems in place to seed the initial ratings of teams based on the skill level of the team members. The reason for this is simple: we don't want new teams of extremely skilled players to be stomping teams that are legitimately near 1200. So even though the other team is 4-5, the individual players had proven through playing on other 5 man teams that they're skilled in ranked teams play. Based on that initial seeding, they've been primarily playing against 1600+ Elo teams for the duration of their short career, whereas your team might have worked its way up from 1200. Win/loss records aren't a good judge of skill since everyone will eventually end up going roughly 50/50 once they've achieved an appropriate rating for their skill level.

As for recreating your team, you would quickly end up at the same rating only if your team demonstrated that it could consistently beat 1550 rated teams. Playing lots of games helps the system hone in on your team's skill level, but doesn't mean that you'll necessarily be above teams that have played fewer games.


21 months ago

Quote:
Originally Posted by ncrwhale View Post
Hi Lyte,

Do you find ideas from the community to be generally unhelpful? It seems like we don't get to put any feedback towards possible ideas, just ones that have been implemented.

I know a lot of comments wouldn't be useful (you've had to explain over and over that the number of wins, win %, and other factors aren't strong indicators of elo), but it seems there are a number of intelligent and dedicated League of Legends gamers.

Perhaps you don't want to discuss a possibility and then not be able to deliver, but what's preventing you from talking about upcoming possible changes in MM? I don't see people getting really excited and then disappointed for possible and then unimplemented MM changes. Sure, I care about these changes more than pulsefire ez and a stealth rework because they more directly effect game play. However, there isn't that coool! feeling to MM changes (and therefore not as great a letdown.)

It seems like a common line from Rioters to say "we're working on something but not ready to discuss it."

edit: maybe you don't want to share data, and without data it is too difficult for players to offer intelligent input?

ps thanks for your hard work!
You touch on a lot of valid points. Because of the speed of development, we generally think about ideas and iterate on them for months before players actually see them "Live" in a game. So in many cases, we've already thought about and seen a lot of the ideas pitched by players. However, it's still valuable to hear feedback because sometimes players will notice something we missed (normal draft picks are an example) and we'll start brainstorming and iterating to polish the idea into something deliverable to players.

I've given criticism and explained why many ideas wouldn't work or added the concerns that players need to think through, so is there more I could be doing with 'possible' ideas?

Regarding discussing upcoming changes in the works, this is something we're constantly working on as a company. In the past, we may have talked about things too early leading to expectations that came back to bite us so we are still figuring out the best ways to communicate with our players. Unfortunately when it comes to data, it's just not something we can release to players for many reasons. As a scientist, I know it's hard to deal with a problem without seeing the data; however, many of the problems we are discussing do not necessarily need data because you can make some clever assumptions or glean data off certain websites to make some educated assumptions


21 months ago

Quote:
Originally Posted by Ginga View Post
Lyte, if you're still reading this I want to ask this one question that I have no where better to ask.

This has nothing to do with making a fairer match making. This has nothing to do with reducing player toxicity. But it DOES have everything to do with making player experiences with this game dramatically better.

Why don't you send us a notification, player name, date of report, etc. everytime someone we reported gets punished?

Everyone has bad and good games. But horribly unfair games where a single person camps in the fountain and taunts the entire party after being told to follow the team? THAT kind of games just downright overshadows any good games one might have. THAT is the kind of game that makes people reluctant to return sometimes.

This is certainly not helped by the fact that we players feel so powerless to do anything about it after it's too late and we just wasted 20+ minutes losing a game while the person just dances in the fountain proclaiming that individual offences can't get them banned in all chat, being brushed off as try-hards, etc.
We actually did an experiment where we did send players notifications when a player they reported was banned. We learned a few lessons and there's a few issues we want to iterate on and polish. However, it's a feature that we will re-visit in the future.

In fact, we're brainstorming and thinking of as many ways as possible to empower our users to feel like they can make a difference in those toxic games.


21 months ago

Quote:
Originally Posted by Ryuzaaki View Post
If there is a way to polish this feature, I hope it can be done soon. Outside of having faith in the system, there's little to no indicator that your reports have any effect. Reporting a player and then seeing him repeat the same behavior a week later really feels as if your powerless. Even though in reality the player likely faced a short suspension or warning, and is well on their way towards a permaban, it leaves the victim feeling as if they have no power and that the troll is getting away with it.

Obviously we can't go around spreading personal details of players to everyone who reported them, but some sort of notification would be a huge help. Perhaps more players would feel encouraged to report if they knew their reports would have a direct effect, rather than hoping that enough players report to have them sent to the Tribunal, and that they are voted guilty.
We're opening Tribunals in all the other international territories, so we constantly have to double check with privacy laws in different countries to figure out what we can (or should) show to reporters when we ban someone they report. I don't have much to say on timelines now, but it's definitely something on my mind for Team PB&J.


21 months ago

Quote:
Originally Posted by ncrwhale View Post
Does elo decay over time? A couple times I've gone on 2+ month breaks, then returned only to lose a bunch in a row until I was back in the hang of things.
Yes, there is a system in place to temporarily reduce Elo when someone hasn't played for awhile. After they return and start playing games, the decayed Elo is given back to them on top of whatever Elo they gained or lost normally.

This hasn't always been the case though, so your breaks might have been before that system was implemented.


21 months ago

Quote:
Originally Posted by GetMad View Post
Are you able to go into a little more depth regarding this? Perhaps an example? I'm more concerned with you saying "start playing games", does that mean 1 game, 5 games, x games?
Unfortunately, we can't go into detail on the numbers.


21 months ago

Quote:
Originally Posted by jokull View Post
wait so decay gets reversed if you play again? how come ive never noticed this, i dont play ranked on my main account very often and ive been decayed 3 or 4 times this season, and ive never gotten the elo back.
The system I described is only for normal Elo ratings, which is why we can't discuss the numbers. It's just intended to account for the fact that players who are away for awhile tend to be rusty and then gain their skills back over time.

Ranked Elo decay hasn't changed recently.


21 months ago

I can't talk about what we are doing on the 'carrot' side of things for now. We're working closely with Pendragon to setup multiple experiments that will test the effects of positive behavioral features.



21 months ago

Quote:
Originally Posted by CupcakeTrap View Post
This response pleases me.

I will just add that I do urge you to also consider a system with "negative" aspects other than Tribunal.

Tribunal is great for removing those with an overwhelmingly toxic attitude, but it seems like kind of a slow, heavy, expensive hammer not well-suited to dealing with players who aren't "the absolute worst of the worst". Even if you're periodically culling the most toxic 0.1%, the most toxic 10% can still have a pretty serious impact.

I'm glad to hear that Lyte and company are on the case, though. I have high hopes!
Totally agree. We're going to play in the positive space for awhile and do some heavy exploration and iteration, but we have a few things in mind we want to experiment with on the 'punishment' side as well. The Tribunal is ideal for targeting the absolute worst players in the game, but there's a lot we could do for the somewhat frequent toxic players that are never bad enough to get sent to the Tribunal.


21 months ago

Quote:
Originally Posted by ncrwhale View Post
Hey Lyte,

I asked this a little bit ago, but I think it may have been passed over...

Is being able to swap order in draft mode an option that's on the table? Few things can mess up a game as much as player 4 saying "get me x" and then not having anyone player 1 wants. Swapping orders would fix this.
We're figuring out the flaws of the current Champion Select Lobby and I know how frustrating it is to navigate trades in the current game flow; however, just letting players trade pick orders is a pretty costly feature to implement and has quite a few concerns.

For example, do we give players X seconds at the start of a lobby to exchange pick orders first? This extends every single game by X seconds which actually adds up over time.

Do we allow players the same amount of time, but they can swap anytime they want? If the first pick swaps with another player but only 25 seconds are left... does the new first pick only get 25 seconds, or should we add more time?

There's a lot of small nuances like this that when you think it through makes pick order swapping seem like a very inefficient feature.


21 months ago

Quote:
Originally Posted by WarlordAlpha View Post
Hey Lyte, while you're posting, I have a really random question I've been dying to ask for a while:

In general, how do peoples' normal Elos compare to their ranked Elos? So, like, if I had a 1250 normal Elo, would my Ranked Elo probably be 1250 +/- 50 ish? Or is it more likely to be a lot lower?
All I can say is there is a surprisingly high correlation between the two Elos


21 months ago

Quote:
Originally Posted by Sir Snicklefritz View Post
Lyte, what psychological benefits/consequences could be gained by implementing a "punish" or "pardon" button on someone's card?

Would they actually feel bad punishing themselves thus spurring the idea to change their behavior? Would someone feel better about them self if they pardoned and continue their behavior?

Would there be any data worth tracking from such an outcome? Possibly accelerate one's case to permabanning if they continue to pardon their cases that are very toxic?
We recently discussed 'self-reflection' experiments which are similar to this and there's definite potential. Not sure what the best implementation would be at this time though and we're already moving quickly on some new stuff.


21 months ago

Quote:
Originally Posted by Guttler View Post
Dr. Lyte,

I see the fix to AFK players in champion select is coming out soon, but I have a question as to what kind of AFKers it effects. So from the description it appears that it prevents a player from entering the lobby if they are AFK after entering the queue and it also prevents players from hard randoming a champion because they are AFK, but does it address players who go AFK after making there pick?

I ask because I just had to dodge a game where the captain did bans and locked in a champion for another player but went AFK without saying what he wished to trade for and stayed AFK for the entire 30 second trade window which made me dodge the game at the last second. Will a player who AFKs after picking there champion be caught by the new feature?

Another thing I've found is that players often go AFK on the load screen, causing them to AFK in the fountain for sometime. Now this would probably be fine if it were a bot game or dominion and everyone had to be in the fountain until the match actually started, but summoner's rift games require players to quickly buy items and invade the jungle other teams jungle or rush to key areas to give protective vision to there own jungler.

I don't think the problem isn't necessarily that the game takes too long to load either, but rather that the load screen should have a chat box so that you can strategize with your team as the game loads.
The AFK feature will only catch players who queue up for a match then AFK prior to entering Champion Select Lobby. After this feature, we still need to follow-up at some point to boot players who hit "Accept" and get into Champion Select Lobby and still AFK, or purposely let the timers to run out.

Unfortunately, even after this point, these features won't be able to catch players who "Accept" into a Champion Select Lobby, select a Champion then go AFK. If this is a large issue after these features go out, we'll take a look and see what we can do. For now, we're waiting to see how much of the issue these features will resolve first.


21 months ago

Quote:
Originally Posted by DiscworldDeath View Post
Any ETA on this going in? I can't wait to know if someone really hit accept and is reading the pre-game chat or just got into the lobby without knowing it while they've went elsewhere.
No ETA yet, the feature is still going through some testing.


21 months ago

Quote:
Originally Posted by Ginga View Post
I'm actually curious, maybe it's just my region or my bad luck.

But has there been an especially large number of Reports lately this past 2 weeks?

I've played this game for quite a while, but this past two weeks I've found an especially large amount of players that needs reporting. I think I reported atleast one player each day I found time to play, some on my team, some on my enemy's team.


And these players aren't just the AFK-ing due to bad connection types either, I've been seeing a lot of players lately who have been especially juvenile even by this community's toxic players standards. Did anything peculiar happen this past 2 weeks or is it just pure coincidence? Is there some american holiday or something?
I haven't seen this globally in the data, but are you playing at a specific time of day and in a specific game mode these two weeks?


21 months ago

Quote:
Originally Posted by LydeckerCain View Post
Hello Lyte, I've come to respect and admire you for your work and your interaction with the LoL community in this thread. You do great work.

That said, something that really kills my enjoyment of LoL is a streak of bad games. This usually involves getting into queue without a full five and dealing with the venom the community is capable of producing- multiple games in a row. I know that after two games tonight with my brother, both of which garnered me and him verbal abuse, I was done playing for at least an hour.

The other side of streaks is when it seems you just can't win a game, and you're suddenly looking at 4 or more losses in a row. I don't know if the matchmaker accounts for recent games played and loss streaks, but I feel it could help mitigate one of the bad experiences I've had with League if Legends.

TL;DR: It's easier to not get extremely frustrated with Lol if negative experiences (Verbal abuse, loss streaks) are spaced out by positive experiences (Good teammates, a win every now and then).
I've been working on some ways to mitigate streaks of losses; other games tend to give players 'freebie games' where they match you up with lower-tier players until you win. I've moved away from this solution because it really isn't a solution that is a net positive for the playerbase--it basically means you break a loss streak while an opposing team suffers through a match they are expected to lose.

We're testing some possible solutions, but can't talk about anything meaningful yet.


21 months ago

Quote:
Originally Posted by Turtle Omega View Post
I wasn't able to vote for the new subject, Extra noise in the ~1200 Elo Bracket.
But this has just been a subject that I've been concerned about for a while now. As you know, Bronze medals are awarded to those who are roughly: 1249-1409 Elo. This is considered the top 25% of players. Bring it down to 1200 Elo, I would be over-exaggerating to say that's the top 40% of players. But I feel that number is just completely invalid, seeing that players start at 1200 Elo. In my honest opinion, it would make more sense lowering the beginning Elo.
Unfortunately, lowering the starting Elo doesn't actually solve anything. Mathematically, the starting Elo simply becomes the average of the population's Elo distribution.

To illustrate this:
Starting Elo: 1200 Elo
Player A: 1250 Elo

If I changed starting Elo, it might be surprising but this is what you see:
Starting Elo: 900 Elo
Player A: ~950 Elo

Changing starting Elo basically just shifts the population Elo distribution and nothing else. Changing starting Elo does become weird when you start going closer to 0 and don't allow negatives... which means you create a system where the overall population Elo slowly drifts upwards.


21 months ago

Quote:
Originally Posted by Turtle Omega View Post
Thanks for the response. I see, and to be honest, I really have no background in how Elo works and its distribution, so I'm not trying to argue against you or anything. But I'm just confused on why the average isn't lower than it already is, seeing that the Bronze medal status states it's the top 25%, when it has to be a higher percentile than that. I'm not thinking about a major change, but I feel that 1000 Elo is much more average than 1200.
Well, the fact that Bronze is the top 25% of players is actually a consequence of a lot of different things. For example, players that tend to lose games and fall below 1200 are more likely to quit Ranked Mode than players who rise up in Elo above 1200--so there's a lot of accounts below 1200 that are just players that do not play Ranked anymore when they probably could have risen up to 1300+ if they kept playing and learning.

Before, it was also possible to lose Elo from Queue Dodging so that shifted the population Elo downwards. Remember, "average Elo" just means a lot of players have that Elo. If you look across all of League of Legends, 1200 is the average Elo.


21 months ago

Quote:
Originally Posted by Semantic Memory View Post
Hey Lyte,

As a fellow cognitive neuroscientist, I sent you an e-mail a while back with a bunch of questions . I know it was a lot, but I never got a response...but I'm still curious about the gaming industry as an option for someone going through the same education as you. Would be interested in talking sometime...would be happy to talk to you in game briefly as an alternative
I replied before but you might have missed it!


21 months ago

Quote:
Originally Posted by Twitch Please View Post
Can you tell us what the approximate distribution is across all Elo's in terms of %? Lolwiki has it listed as:

Bronze: Between 1250 and 1399 (3v3: 1249-1409, pre-made 5v5: 1249-1409) (Top 25%)
Silver: Between 1400 and 1519 (3v3: 1410-1489, pre-made 5v5: 1410-1499) (Top 10%)
Gold: Between 1520 and 1899 (3v3: 1520-1699, pre-made 5v5: 1500-1749) (Top 3%)
Platinum: 1900 and above (3v3: 1700+, pre-made 5v5: 1750+) (Top 0.2%)

But I feel like this is outdated and that ratings have inflated since. Is that correct?
Yegg is probably the best one to answer this!


21 months ago

Quote:
Originally Posted by Twitch Please View Post
Can you tell us what the approximate distribution is across all Elo's in terms of %? Lolwiki has it listed as:

Bronze: Between 1250 and 1399 (3v3: 1249-1409, pre-made 5v5: 1249-1409) (Top 25%)
Silver: Between 1400 and 1519 (3v3: 1410-1489, pre-made 5v5: 1410-1499) (Top 10%)
Gold: Between 1520 and 1899 (3v3: 1520-1699, pre-made 5v5: 1500-1749) (Top 3%)
Platinum: 1900 and above (3v3: 1700+, pre-made 5v5: 1750+) (Top 0.2%)

But I feel like this is outdated and that ratings have inflated since. Is that correct?
The percentages are still somewhat close to this. That said, we've been reevaluating our philosophy on who should get rewards, and we no longer believe that only the top ranked players should be placed into a tier. We'll be making an announcement on this sometime soon™.


21 months ago

Quote:
Originally Posted by Lailu View Post
http://jkfweb.com/files/Imbalanced_L...5_July31st.png

Lyte I just played a game where we got slaughtered bad, but one thing I noticed after the game was that our team was 1289 average elo, and their team was 1328 average elo.

The weird thing is their whole team was above 1300 elo, and our whole team was below 1300 except for one person at 1302.

Does this seem right to you? I would think that it should sprinkle those two teams together so you have a mixture of higher and lower elo players.
The Elos you are looking at are post-game adjusted Elos. Before the game, both of the teams was actually ~1300 average Elo.


21 months ago

Quote:
Originally Posted by Lailu View Post
Your right after you adjust for the elo.. my team was an average of 1300, and the other team was an average of 1316.

Thanks for the prompt response, I appreciate .

How close in Elo does the match maker try to get usually?
Initially, the matchmaker tries to get as close as 5 Elo between teams. Over time, that number expands until a game is found.


21 months ago

Quote:
Originally Posted by Shibbibidibdub View Post
Question for Dr. Lyte,
First, I would like to say thank you for your continued dedication to this thread, and helpful insights on community ideas and upcoming features. Just keeping us in the know is always appreciated.

Question: In the event of a leaver before the twenty minute mark, unable to surrender, could the players left in the game surrender early or leave/get out, without penalty? Could this be a possible feature?
I've been taking a look at LeaverBuster and how leaving is in League of Legends in general. On one hand, LeaverBuster has greatly improved the rate of leaving; unfortunately on the second hand, the leaver rates are still higher than I would like.

I'm still brainstorming on some things we could do, but I am generally avoiding letting players surrender early or leave the game without penalty; in research on this type of implementation, the end results were always a net negative for the players. For example, if we made it so that players can leave without penalty (or Elo loss), than a lot of bullying occurs because players will try to force a player to leave whenever they feel the game is lost.

If we do a feature like this, leaver rates actually don't go down--the only thing that changes is bullying increases and leaving becomes something that is considered OK in the community. When we figure out what we want to do with LeaverBuster, we'll make sure to not only decrease the leaver rates but also somehow mitigate the impact to other players when a player leaves their game.


20 months ago

Quote:
Originally Posted by nemo14 View Post
how much has queue dodging in ranked changed since the penalty was changed? has a rioter posted about this elsewhere? if so, where?
Yegg and I are currently pouring through the data on queue dodging and we'll talk about some of our thoughts in the future; first, we need to figure out how we want to change the penalties again.


20 months ago

Quote:
Originally Posted by MasterMirage View Post
Hey Lyte, in regards to Toxic players and their punishment coming way too late (i.e ending up in the tribunal) I was thinking about implementing a tool thanks to spectator mode.

Basically, players in the game can select an option to "Request Support" which allows for tribunal summoners with a high percentile of cases judged to come in and spectate the game. If enough of these tribunal players deem the player being judged toxic, they can "fast forward" the summoner to the tribunal to be judged even further.
This is a neat idea that statuskwoh asked me about when his team was working on Spectator Mode. It wasn't exactly the same, but the idea was to allow spectators to report toxic players in Featured Games. One of the main concerns is privacy; most spectators would actually flock to high Elo matches creating an environment where those players were observed (and reported!) far more often than any other subset of players. To allow spectator reporting we would also have to enable team-chat for these players, which is another can of worms.


20 months ago

Quote:
Originally Posted by Sabaster View Post
Lyte, could you tell me the correlation between Season 1 elo and season 2 elo? Also, those first 10 games of ranked?

I'm sitting at around 1350 elo, but my W/L for this season is 14-3, I think
I was certain I started ranked towards the end of season 1, but it seems my ten games were factored in for season 2.

The point: When transitioning to season 3, is it a completely clean slate?.
I'm actually unsure what the plans for Season 3 Ranked are at this time. Yegg might know more, he's the design lead for Season 3--I just work with him on how matchmaking in Ranked works.

Generally, Season 1 Elo and Season 2 Elo are very similar for the average player. However, it depends on factors like how many games you played in each season.


20 months ago

Quote:
Originally Posted by Sagethyme View Post
Heya Lyte, apologies if this isn't the best place for the question.

In Co-op vs Ai, is it possible to have the AI scale against the ELO of the human team?
Right now, there's no way for us to scale the AI depending on player skill; however, we're interested in knowing what kinds of features or improvements interest Co-op vs Bot players. Happy to discuss this if there are any players who play a large number of Co-op vs Bot games here.


20 months ago

Quote:
Originally Posted by Seven Score View Post
I almost exclusively play against bots, though I've played a few PvP matches and like to spectate my friends when they play PvP. So I like to think I have an idea of how the 'actual' game works. The reason I mention this is because a lot of people (including my friends) seem incredibly dismissive of Co-op vs AI and anyone who plays them; I've been harassed quite a bit about it.

Anyway, I'm really glad that you're willing to listen to our thoughts on the state of bots. I do have a few issues I'd like to bring up myself. One of them is matchmaking-based, but the others are about the AI. I should probably bring them up elsewhere, but I figured I'd mention them anyway in case it could be redirected to an appropriate Rioter.
I'll take a look at matchmaking in Co-op vs Bots, it's probably something we need to re-visit and upgrade in the future.

Quote:
• Matchmaking by level fails sometimes?

There have been instances where I (being in the 20's in level) have been matched to level 5-7 players in Intermediate bot games. Every single one of those games resulted in a loss.

I don't blame the players for the loss, but it confuses me why they'd be playing Intermediate bots at such a low level. Trying to protect and guide new players against a unforgiving, cheating (see below) opposition is incredibly tough. Could we perhaps get a notification for new players below a certain level about how difficult Intermediate bots can be? None of them were up to it and would have had a better time against Beginner bots.
Hm, we might want to just consider unlocking Intermediate Bots at a certain level.

Quote:
• Playing against Intermediate bots can become unnaturally punishing mid-to-late game.

Due to the way bots' passive gold gain functions, they become fed no matter what you do. It feels very punishing compared to a normal game of LoL since your opponent will always be able to buy items faster than you no matter how well you outplayed them. There is little feeling of reward in that, I think.

I believe that this is encouraging the human teams to play as aggressively as possible, abusing the bots' weaknesses to beat them before they can become too powerful despite the fact that they should, in theory, be losing.
I am going to forward the rest to our bot designer and get some of his feedback.

Quote:
Now I just hope that I don't get flamed into the ground for being a noob bot player or something.
I recently did an analysis on Co-op vs Bot mode and hopefully players aren't flaming bot players. Many players play Co-op vs Bot mode and I can easily see why--it's lower stress, a place where you can try wackier builds (and still win) and generally a more peaceful game mode.


20 months ago

Quote:
Originally Posted by Apollinarius View Post
If you already want to change penalties again, that probably means that the data isn't good (too much dodging).

Honestly, that surprises me. I've had very few dodges in ranked and only slightly more in normals.
Overall it hasn't been a bad change, but there are pockets in the population where the change created some new problems.

It's definitely better than the old Elo penalty and we don't have any more queue dodge smurfs (players who queue dodge down to low Elo then stomp on newbies towards 1100 Wins / 100 Loss records)... but we just have to do some iterations to resolve some of the new problems that popped up.


20 months ago

Quote:
Originally Posted by YurdleTheTurtle View Post
Hey Lyte, I don't know if this is still relevant (since a long time ago you were interested in screenshots of matchmaking gone wrong), but here's a recent game.

Basically, apparently one of my teammates was over 1550 Elo before the win. This is very one-sided considering the rest of the players in the match were in the 1350-1400 range.

What's up with that? My guess is perhaps people duo'ing with a large difference in Elo.
Unfortunately, these are because of players Duo-Queuing.


20 months ago

Quote:
Originally Posted by Silent Warden View Post
@Lyte:

Do you know if any of the teams are working on a way to make surrendering less....annoying I guess? I mean, for me, I hate it when the surrender bar pops up right as I'm about to kill someone so I can't target him and he gets away. I've been supporting a thread about having a toggle button for surrendering, and once everyone toggled yes, the game would end. Nothing would pop up; the game would continue uninterrupted until 4/5 people toggle yes to surrender.

http://na.leagueoflegends.com/board/....php?t=2376564
I do like the idea of a toggle that you opt-in or opt-out for a game because it's far less intrusive on gameplay and morale; however, I'm not sure when it's something that would be higher value than some other things the team is working on


20 months ago

Quote:
Originally Posted by Shibbibidibdub View Post
Dr. Lyte,
Might have been asked already, but...
Could you reevaluate Bot logic after one or two inhibitors go down? In three games I played today in Co-op vs AI Intermediate, after two inhibitors were down and four of my teammates were in the Bot base not attacking Nexus towers for a couple minutes, three Bots came down lane to kill me. Even when my team started attacking the Nexus turrets, they (the Bots) continued to push out to the tri-bush area in top lane after they killed me, only turning around once one Nexus turret was down. Shouldn't protecting the Nexus/Nexus turrets be top priority after inhibitors fall? And even if Bots are out of the base can they not run back in one by one so it isn't easy to pick them off?

Thanks and keep up the good work.
(You are probably my favorite Red)
I'll forward this feedback to our Bot designers, thanks


20 months ago

Quote:
Originally Posted by Sabaster View Post
When transitioning to season 3, is it a completely clean slate?.
The problem with having a completely clean slate is that the first few weeks of the season are miserable as everyone is thrust randomly into the same games regardless of skill level. So we won't be completely resetting ratings. We might do a soft reset, and we might also reduce the system's confidence in the current ratings so that everyone has a chance to quickly earn large gains if they're currently below their true Elo.

We'll announce more about how the transition will work at a later date.


20 months ago

Quote:
Originally Posted by Dromar420 View Post
Point being this is a bad analogy what if there was 3 people in the NBA that got swaped from team to team and they are what caused that team to lose every time they were on that team would they then not have something done about them to keep them from destroying the chances of that team in or any other in any further games?
The NBA analogy was just to provide context on 1 specific point: just because some players are "bad" does not mean they are not allowed to play Ranked. Everyone of all skill levels should be allowed to play Ranked and that isn't going to change any time soon.

Quote:
Also as has been stated before the report system is very lacking with the fact there are many cases that dont git to the tribunal for months at a time as ranked there should realy be more imedeate action taken many times ive heard people talking about giting trolls 4 or 5 times in less than a week then about 3 weeks later the same person again is trolling and they run into them another 3 weeks and hes on the forums talking about his ban cant remember who.
We are always working on improving the speed of punishment.

Quote:
Also on the problem of the elo floor you stated that trolls would just troll because there no droping to a point the problem with that thought is that as of now they can troll every game and people can still manage to win games with them on the team because your 1200 and all 4 of your team mates are people who played games and are actualy ranked there and the other team has 3 people who are "new" to the ranked and are as such at that elo or god forbid 1300+ because our troll is a decent player when not trolling went 8-2 and is now trolling and tanking elo of other players in the 13-1400 area that are ranked there because it is there elo do you honestly beleave he cares ive seen posts talking about being proud of how low they got there elos so how do you justify this.
It's unclear what point you are trying to make. Setting Elo floors would allow toxic players to troll even more and harder than before, without any punishment other than waiting for the Tribunal to act on them.


20 months ago

Quote:
Originally Posted by ThirdTimeLucky
Simply put, newer players are not average. What the system is virtually guaranteed to do is to have new players sink like a rock when they start playing ranked. This "downdraft" makes it more difficult to move up. (because if you end up below the starting level, you are more likely than average to be paired with players who should be going down". Finally, the new player finds themselves stuck with players who have richly "earned" their low ELO over time due to trolling, frequent afking, or other issues.
Quote:
Originally Posted by Goumindong View Post
Actually in League "new" players to ranked are pretty average. Its one of the advantages of having to play 200+ games and own 16 champions before starting ranked. There are no "newbies".
Goumindong is correct. We ran this analysis awhile back and players who are new to ranked finish their provisional matches with an average rating that's very close to 1200 (and it included some players with sub-500 ratings because of heavy dodging). Now that players don't lose Elo from dodging I expect it would be even closer.


20 months ago

Quote:
Originally Posted by Tricolor View Post
.....what you are saying is that everyone that enter ranked mode now is a 1200 player(for real)...and thats kind of impossible....

the average elo can be around 1200 if you pick all those new players (since new players play with new players for the first matches... )....but that doesnt mean that they are 1200 players.......like...1 guy will be at 800 elo and the other one at 1600....the average ll be 1200 but in no way both players are 1200 players... pretty sure the "real"rating distribution of the "new" players is close to the real distribution considering all players...and not what you just said....
No, that's not what we're saying. For example, let's assume a normal distribution in the population--some new players entering Ranked are going to be amazing 2000+ players and some are going to be terrible 300 Elo players. However, the majority will actually be ~1200ish players.

For example, given 100 players entering Ranked for the first time... about 60 of them will end up at 1200 Elo after their provisional matches. 20 of them will end up below 1200, and 20 will end up above 1200.


20 months ago

Quote:
Originally Posted by Maverick Renegade View Post
That said, over a large enough sample size of games you'll make enough impact, good or ill, to slowly win or lose an extra game here or there and start getting out of the 1200 mire in either direction. Maybe you are that feeder/carrier.

I think you're putting the cart before the horse here. People aren't ending up averaging 1200 because they are, on average, at that skill level upon entering ranked. They are ending up there because of the nature of the system you set up. Have you considered that the average is your inadvertent creation, rather than a symbol that you're correctly gauging skill with your method?
Hm, you make some good points but I think there's a misunderstanding somewhere. I never said players end up at 1200 because they are average--we don't know what skill they represent until they play more games. Most of these players will probably still end up around 1200 after hundreds of matches.

The misconception that we're discussing spans a couple different points:

1) Changing the starting Elo for Ranked doesn't actually fix anything -- many players assume new Ranked players starting at 800 Elo 'fixes' matchmaking.

2) The average Elo of the system actually is 1200... but the fact that Bronze is the top 25% is actually a side effect of other issues.

3) After about 150-300 games, most players end up at their true Elo. We're trying to shorten the time it takes to get there because we feel ~300 games is too long.

4) The new players that enter Ranked represent about the same distribution as the whole Ranked population, which is why majority of people end up at 1200 after their placement matches.


20 months ago

Quote:
Originally Posted by albert2006xp View Post
How about you just reduce the elo won/lost by placement matches? As you say, it takes 150-300 matches to average out your true elo.
Say for argument sake you are a 1800 player. You have an incredibly huge bad luck streak on your placement matches (aka 2 afks in your team each game, just for sake of argument) You will end up somewhere under 1000 and now you will only gain <15 elo per game. So it will take you quite a long time to get to your true elo.
Rare cases,but they happen. Didn't happen to me but that terrifies me greatly for next season.
Actually, you get +/- up to 40 Elo for several matches after the placements, so there's plenty of time for people to get a boost up to their proper Elo if they got unlucky during placements.


20 months ago

Quote:
Originally Posted by Koechophe View Post
Lyte, I have to ask, how do you feel about the pre-game for draft and ranked picks?

Right now, it seems like a sesspool of "I CALLED MID, IM LOCKING IN AS ANNIE AND GOING MID EVEN THOUGH IM LAST PICK" or even "Even though I've been completely silent throughout the banning process when everyone was discussing roles, im going to auto lock as whatever because im first pick".

There's also a lot of flaming, swearing, trolling and people trying to get other people to leave the que. That's how it appears to me, anyways.
Champion Select Lobby is probably one of the greatest sources of toxicity in the game at the moment and is something we're actively investigating for potential solutions.


20 months ago

[QUOTE=isobold;28413043]@Lyte: I've discussed with some German players on the German forums, and since I'm an Emissary there, they asked me to translate a proposal for them. Their issue is that they are very high rated in normal games and thus get really long queue-times, but since the new ready-check-system is in place, aren't even allowed to leave their computers to grab some drinks while waiting. While I guess it won't be of much use to you, this is the thread we discussed the following proposals in: http://euw.leagueoflegends.com/board...d.php?t=808045
[QUOTE]

rjcombo and Brackhar are the designers of the AFK Feature and are mapping out some iterations currently.

Quote:
C) I have been told by users that your solo-queue Elo would influence the teams start-elo when queuing for the first time in a 5vs5 or 3vs3 ranked team (I'm not talking about players queuing with a new team after having player other team-matches with other teams, but about accounts that never played team matches before!). Is that true? Does the teams start-elo depend on the solo-queue-elo of each member, if no member ever played in a team before?
There certainly is some seeding that happens when you form a Ranked Team. I can't say how the seeding happens or which Elos are used though.


20 months ago

Quote:
Originally Posted by Aparkhurst View Post
This does not seem to be the case. Was this statement an oversight? Did you just assume that i was a 1700 player? My last recorded Dominion elo was 2641
(source: http://lolmatches.com/top/dominion/all )

Heres a thread I made with a screen shot of me waiting 2 hours for a dominion game on my EUW smurf http://na.leagueoflegends.com/board/....php?t=2442558
Well, there's multiple issues at play. Quite a few players are playing Proving Grounds now and many of these players might have been Dominion players because it's a similar play style--so the matchmaking pool for Dominion is smaller now.

Furthermore, your smurf is almost never going to find a match in Dominion because you are a high Elo smurf and low level at the same time. There's very few players who play Dominion at low levels and even fewer players who play Dominion at low levels and are high Elo.

Unfortunately, the matchmaking changes did not change much for Dominion. The reason you are seeing additional issues is probably a combination of Proving Grounds launching, slight changes in how Summoner Levels are used in matchmaking and you are a low level super high Elo smurf trying to play Dominion.


20 months ago

Quote:
Originally Posted by Aparkhurst View Post
Thank you for responding.

May I ask why matchmaking looks at level at all? If I am able to reach a high elo with the low level handicap, what does it matter if I am treated like a level 30?

If there is some crazy difference in power between a lvl 20 and lvl 30 account, wouldn't that just organically reduce my elo down to where the handicap didn't matter?
New players joining League come in various skill levels--some learn quickly and rocket up their Elo. These players would then face Level 30s pretty early on and perceive the matchmaker as broken even when they won their games.

In League of Legends, levels actually do correlate with more Summoner Spells, Runes, and Champions so I can understand why these players perceive the matches as 'unfair.' We do realize that these matchmaking changes would affect smurfs negatively in some ways, but in all fairness smurfs are not something I want to encourage

I realize some players smurf to play with new friends they bring into the game, but that's a legitimate edge case we're going to work on.


20 months ago

Quote:
Originally Posted by Panzerfaust View Post
and I'd hope are planning some changes after they've seen how including level as a stronger factor affects things (i.e., after 5-10 minutes it starts drastically reducing the level weighting for smurfs).
The matchmaker already stretches the level ranges over time; however, I could see an argument to make the ranges even larger after 5-10 minutes.


20 months ago

Quote:
Originally Posted by Skaarrjj View Post
Lyte. What is a player to do when theres players that are trolling or just being general ******bags, but you don't have enough people to report him because they just dont take the effort to report?
If it's excessive racism or something similar that needs to be dealt with immediately, send a screenshot in to Player Support.

If it's more run-of-the-mill toxic behavior, just report and move on It'll be taken care of over multiple games when the player gets reported by others.


20 months ago

Quote:
Originally Posted by Twitch Please View Post
I'm dismayed with how ineffective this is. A week ago I submitted a Player Support ticket about a Summoner named 'Prince Fluffi' who, after the game ended, was advertising in the player lobby saying he was actually a 2.1k Elo player who was selling Elo for money and then provided his contact information. This is someone who will obviously evade punishment by the Tribunal and has to be dealt with directly, but I check his match history and after a string of mediocre performance in normals, there he is absolutely crushing a ranked game now 200 Elo higher. This game was an hour ago. What gives?

http://www.lolking.net/summoner/na/20245760#profile (lol @ that Elo chart)
Death threats, extreme racism and other such behaviors are dealt with very quickly if you submit a Player Support ticket.

Dealing with players selling Elo services is less priority than the above so it will take awhile.


19 months ago

Quote:
Originally Posted by Nyutrig View Post
since you are a psych major and all, have you considered that your research is done with a degree of confirmation bias?? in my opinion they should not have a "psych major" determining the changes involved in the matchmaking system, but instead a MATH major.

http://jeps.efpsa.org/blog/2012/06/0...vious-results/

http://blogs.discovermagazine.com/gn...in-psychology/

http://whywereason.com/2011/09/07/ps...ted-reasoning/

or you know you could.....ignore these studies.....

you cant argue with the sheer ammount of people who have the same or similar complaints with the matchmaking system. and if you are going to argue it, you should give examples with data. is there a reason that you withhold all of the data that you describe?
I lead teams of people that include mathematicians and statisticians who are experts in machine learning and predictive modeling. Team PB&J also has a new analyst starting that got his Masters in Bioinformatics and used to do cancer research.

I'm well aware of the common cognitive biases you mentioned which is why Team PB&J approaches problems very differently from traditional development teams. We have a hypothesis (null hypothesis and all) and use data to test the hypotheses repeatedly. Many times, we've had the null hypothesis come out victorious and many times we've had results that surprise us and counter intuition.

If you wonder why many players have the same complaints that may or may not be valid problems with the system... you may want to review some other cognitive biases as well


19 months ago

Quote:
Originally Posted by Eledhan View Post
Can you say...PWNED?

Seriuosly, though, Lyte... I really think you might just be my favorite Rioter, simply because you blend human behavior, morality (the scientific version, of course), logic, statistics, and factual evidence all into one wonderful department designed to make my gameplay experience as enjoyable as possible...

Of course, the rest of your team is wonderful also...I just don't know their names!!
Keep an eye out for threads that announce new Team PB&J features. We typically have the team members that worked on the feature do Q&A and introduce themselves to the community.

In the past couple weeks we've been focused on releasing Tribunal features, so awesome members like RiotAaronMike, RiotNinjaTabby and VonBurgermeister have been presenting work like this: http://na.leagueoflegends.com/board/....php?t=2380577


19 months ago

Quote:
Originally Posted by Imclueless View Post
Completely related: How long are you intending to collect data with the reform cards before working on the analysis (and presumably modifying it or removing it)?

I'm curious, because I'm VERY eager to see what sort of effect, if any, the cards are having (though I won't have a chance to see the data myself, of course).
It's hard to know the exact time to wait since this is the first experiment of its kind. There's some weird issues like when a player is banned for 2 weeks, we need to wait 2 weeks before we can even start measuring the data for that player. When we actually start collecting data for a player, I'd want to see at least 6 weeks of data to see whether their behavior trends for the better or back towards the Tribunal.

I'm just as excited about the data as you are


19 months ago

Quote:
Originally Posted by TheDjinni View Post
Conclusion

Look, I'm not saying that all or even most complaints about elo hell are valid. All I'm saying is that elo hell could be an issue for some players. I have no idea whether it is the all-encompassing problem that people assert it is; I am only pointing out that it is entirely logical to expect it to exist at some level and affect some people negatively.
I've always said that for a small (<1%) of our playerbase, it will take thousands of games to reach their "true Elo." This is due to a number of factors such as noise, statistical probabilities not falling in their favor, etc. Does this mean those players are in Elo Hell? Possibly. For the average player though, it takes about 200-300 games. We do agree that 200-300 games is too long to reach their true Elo, so we're brainstorming ways to reduce this time-to-true-Elo dramatically.


19 months ago

Quote:
Originally Posted by DirtyNate View Post
Hey Lyte,
Just a question for you about retaliation. I just played a game as Malzahar where it was kind of close at first but we started snowballing in the last 1/4. Once the snowball was in full effect the enemy ziggs said "gg malz get carried". This kind of surprised me since I was 8/3 and had same CS as the ziggs, who was 6/6. So I replied "8/3?". To which he said "you got shat on in lane." Ziggs had killed me once in lane, but I had killed both him and hecarim in a 1v2 when they came for a gank attempt. So I said "you killed me, but I killed you."

Ziggs then proceeded to blame his death completely on hecarim, and kept talking about how bad I was. So after the game in the lobby. I said "lets look up your elo, lol 1477, too pro for me, peace loser". With the exception of the "peace loser" I don't think anything I said was too offensive. Bringing up elo was probably uncalled for.

But my question is how to deal with people that attack you completely unprovoked and continue to attack you? It only seems fair that if he can attack me personally without any provocation, that I should be able to defend myself. Is this justified or is my only option to /ignore and move on?

TL;DR Is defending yourself from attack ever justified?
I'd just ignore and move on.


19 months ago

Quote:
Originally Posted by BooleanLobster View Post
Hey Lyte,

Are there plans for tricky social engineering hacks to nudge behavior? I recall that the Economist website used to begin their comment sections with "SIR-", making the first comment appear more formal and respectful. Has something like an automated "gl hf" been considered to set the tone of the game? Does that kind of thing even work?
In research settings, contextual cues like that have noticeable effects on behavior. Will it work in the real world outside of a lab? It's hard to say, many academic fields still don't have the resources or tools to do large-scale experiments in the real world to make sure that lab results generalize to different contexts and environments.


19 months ago

Quote:
Originally Posted by Clerigon View Post
Whats if the matchmaking use the normal elo in ranked games? for those who are new to the game and jump to ranked they dont start at the same place with people who have a bunch of normal games.

The system will not use 100% the normal elo.
This would add a lot of additional stress to Normal Blind Pick and other "Normal" modes that are intended to be less hardcore/competitive and completely separate from Ranked play.


19 months ago

Quote:
Originally Posted by CupcakeTrap View Post
Hey Lyte, I'd like to report a mismatched game. Look at this Garen here. What is he doing in this game? Look at his CS. He's clearly dramatically better than everyone else.

Final score: 18-4. Not a great MatchMaker success story here.
Clearly Garen is OP. I need to talk to Geeves asap.


19 months ago

Quote:
Originally Posted by Etherdreamer View Post
Hey Lyte, how about the "unskilled" trolling issue? You can be forced to group up with incredible less skill level for a troll that report you as unskilled, even you do it good, for score CS, you can still be reported and trolled.
Unskilled Player reports do not automatically lower a player's Elo. We'd have to manual approve it in a massive player sweep for it to take effect.


19 months ago

Quote:
Originally Posted by DemmoxMKII View Post
Why is there no "Other" option when reporting someone?

I feel "Other" would help for ones that quite honestly dont know how to report them: "Must report garen for leaving mid game ' Hey lets make this fast I have to leave in 12 mins' "

EDIT: Play with me? :3
At some point I do want to re-visit the report options and consolidate some of them and just polish the system in general. Right now, too many report categories are confusing, are mis-used often (Intentional Feeding!) or even overlap with each other (Offensive Language and Verbal Abuse!)


19 months ago

Quote:
Originally Posted by KultoDeSkaro View Post
I hope you realize, though, that having this as a report option does more harm than it can possibly do good. You've clearly never seen all the ragers who feel justified in abusing and reporting players who are doing badly just because this report option exists.
This is part of the reason we've been challenging players to use reports more accurately and we recently punished players who had been extremely toxic with their usage of reports and bully of players in the game.


19 months ago

Quote:
Originally Posted by Goumindong View Post
Would it? So long as the normal ELO is only used as a seed value and discarded after a player starts playing ranked then there shouldn't be any carry over with the exception of people who are practicing with the intent to start ranked soon.

You've said before that there is a lot of correlation between the normal and ranked ELO of players. If that is the case then it should be possible to get a decent estimate for the ranked ELO of players based on their normal ELO.
In a perfect world, if players continued behaving as they currently do, then yes, Normal Elo would be a great seeding parameter for Ranked play.

However, the odds of players making new Level 30 accounts, going premade 3-5s to carry up to excessive high Normal Elo to start at a high seed in Ranked is actually pretty high.


19 months ago

Quote:
Originally Posted by EFG View Post
Hey lyte. Not sure where to put this to best get your attention. I've recently been levelling a second account while playing with a friend of mine who is new to the game. The biggest thing I've noticed in the process is the insane amount of 4v5s that take place at lower summoner levels. It seems like every second or third game someone leaves after 2 or 3 deaths, sits in the fountain at level 1 until the game boots them or never connects after champ select. What are you guys doing about this?. It seems like this would drive a lot of new players to give up on league before they can actually get into it.
We're thinking of ways to upgrade LeaverBuster to address low level leavers. We feel much of the problem is that players don't realize leaving in the MOBA genre is bad, especially if they came another genre where leaving is OK and simply ends the game.


19 months ago

Quote:
Originally Posted by Ulanopo View Post
Dear Lyte,

My friends and I have been trying to play more PvP lately and we've noticed some things about matchmaking that trouble us. Specifically, that the normals matchmaker seems to heavily prioritize speed of matching over balance. I happened to spectate a game by a friend of mine who is level 19 with no one on his team above level 24. The opposition had a Rengar player with a ranked Elo of just shy of 1400 who proceeded to take a dump over my friend's team. (I can give you details privately so that you can see the match for yourself.)

I'm kind of curious why such blatant mismatches happen. Was it the late hour? Is it the way teams are assembled?
Could be late hour, could be who his teammates were (i.e = smurfs but on Level 24 accounts). It's hard to explain any specific 1 game without a bit more details about the context and even then, a nearly perfect matchmaker will have some % of games be lopsided.