How many games do you expect to play before you reach your true Elo?

< 10 Games 180 1.63%
11-40 Games 819 7.41%
41-80 Games 1,162 10.52%
81-120 Games 1,738 15.73%
121-160 Games 665 6.02%
161-200 Games 929 8.41%
200-399 Games 1,703 15.41%
400-599 Games 1,087 9.84%
600-799 Games 457 4.14%
800-999 Games 170 1.54%
1000+ Games 2,140 19.37%
Voters: 11050. You may not vote on this poll

Help Riot improve matchmaking! Looking for examples of bad matchmaking

First Riot Post
Comment below rating threshold, click here to show it.

rekenner

This user has referred a friend to League of Legends, click for more information

Senior Member

02-15-2012

The best suggestion I've heard is a "This guy didn't deserve to lose Elo for this loss" mechanic. I think it would be the most fair way to tweak Elo loss/gain (as compared to MVP (which is harder to judge) or a reduction in Elo loss due to a leaver). Remove the ability for duo queue partners to vote for each other (and maybe even people that have friended each other) and I think it might work.


Comment below rating threshold, click here to show it.

Lyte

This user has referred a friend to League of Legends, click for more information

Lead Social Systems Designer

Follow RiotLyte on Twitter

02-15-2012
14 of 173 Riot Posts

Quote:
Originally Posted by The Furies View Post
So, don't steal somebody's work without notifying him. He has actually come up with a faster algorithm for matchmaking. If you can increase the speed at with you can matchmake. Then you could implement more routines for finding better quality matches in the same amount of time.


Here is the paper http://www.cs.unm.edu/~hayes/papers/HatsProblem.pdf

Your welcome
Thanks, this will be a good read over the weekend. You just crushed my dreams of not reading anymore academic papers after my PhD.


Comment below rating threshold, click here to show it.

Lyte

This user has referred a friend to League of Legends, click for more information

Lead Social Systems Designer

Follow RiotLyte on Twitter

02-15-2012
15 of 173 Riot Posts

Quote:
Originally Posted by PurePowah View Post
I sent you an email of an 1800 game that I feel we had not even 1% (not exaggerating) chance of winning.
Thank you!


Comment below rating threshold, click here to show it.

C00LST0RYBR0

Senior Member

02-15-2012

Quote:
Originally Posted by rawpower405 View Post
I'm very confident that TEAM kda demonstrates the superior victor. I wrote a lengthy post about this about a year and a half ago showing it.

Of the top of my head, simple differences in ELO predicted a 55% win rate, while the true win rate was 69%. Using the TEAM'S kda predicted a 68% win rate. Considerably better.

KDA is very much correlated with the winning team. IE the team with the better K+A/D is generally the team that wins. The greater the disparity, the greater the stomp.
never said it wasn't. What I'm saying is how would you rate performance. Just giving the players with more kda more elo? Supports and tanks might have contributed just as much to a win, or even more, but this can't be seen in any statistic.


Comment below rating threshold, click here to show it.

LOLOL BYE

This user has referred a friend to League of Legends, click for more information

Senior Member

02-15-2012

Lyte - Can you comment on this?
http://na.leagueoflegends.com/board/....php?t=1828013


Comment below rating threshold, click here to show it.

Rammblin13

Member

02-15-2012

Quote:
Originally Posted by C00LST0RYBR0 View Post
never said it wasn't. What I'm saying is how would you rate performance. Just giving the players with more kda more elo? Supports and tanks might have contributed just as much to a win, or even more, but this can't be seen in any statistic.
Supports often end up with the best kda actually. Tanks can rack up a ton of assists too.


Comment below rating threshold, click here to show it.

Dragonk

This user has referred a friend to League of Legends, click for more information

Senior Member

02-15-2012

Also please remove the placement system for the first 10 games. Getting -80 elo for having an afk or something along those lines is incredibly frustrating and quite devastating as it may end up taking 6 wins just to make up for one lose. This is especially true if you get really unlucky like me and get 4 afks in my first 10 matches....


Comment below rating threshold, click here to show it.

Lutcikaur

This user has referred a friend to League of Legends, click for more information

Senior Member

02-15-2012

I have to post this. I just sent an email of what i have experienced, and UNDER A MINUTE LATER, YOU GUYS RESPONDED! you are my heroes. I cant explain how much more (the already large amount) of faith i have in you guys increased. Awesome work. Really. A+++

also, if you guys ever are up for a game, i will play O-o


Comment below rating threshold, click here to show it.

rawpower405

Senior Member

02-15-2012

Quote:
Originally Posted by C00LST0RYBR0 View Post
never said it wasn't. What I'm saying is how would you rate performance. Just giving the players with more kda more elo? Supports and tanks might have contributed just as much to a win, or even more, but this can't be seen in any statistic.
No, give the TEAM more elo based upon the total TEAM's KDA versus the other team.

If Team 1 has a K+A/D of 10/9, and Team 2 has a K+A/D of 11/9, that's a close game. and elo gain's losses for the two teams should be similar +5/-5 or something.

If Team 1 has a K+A/D of 93/23 and Team 2 has a K+A/D of 50/70 then it was a stomp, and the elo gains/losses for the two teams should be more divergent. The winning team should win more elo and the losing team should lose more elo.

This would NOT evaluate a players contribution to the KDA, but rather would inform the elo gains and losses.

Elo is 1-0 system. which can also be read as a margin of victory(Hence why it works so well in soccer, since most matches are won by an average of 1 point). But in higher scoring games, basketball, football, etc. elo stops working so well and needs to use margin of victory.


Comment below rating threshold, click here to show it.

Reiver

This user has referred a friend to League of Legends, click for more information

Senior Member

02-15-2012

So Lyte, purely anecdotal (since it was a couple months ago now), but what I've seen of newbie island matches vs actual normal ranked matches is that newbie island ranked matches at 1400 rating are SIGNIFICANTLY easier to win than normal ranked matches at 1100 elo.

This is a SEVERE problem because it means the average skill of a newbie island player is well below the ranked average, which means newbie island players are typically inflated to well above their appropriate elo rating.

Unfortunately I don't have the appropriate account resources to get you some more hard evidence so this is all I can bring to the table right now =/