It is an idea that has been proposed many times before; however, consider the concerns surrounding duo-queue. In duo-queue, when two friends queue together who are greatly different in Elo, lopsided games tend to happen.

What happens when you have a duo-queue scenario where now you have widely different Win-Loss records, and greatly different Elos? How many people only play with friends who have similar Win-Loss totals? How many players have friends who all have greatly different Win-Loss totals and Elos?

What if we improved the Elo metric to be more representative of player skill such that a 1500 player who is 20-15 actually is pretty equivalent to a 1500 player that is 100-90? Then would we even need a Win-Loss band parameter in the system? It is an interesting idea, but there is a lot to consider.

I dont see how you could change the metric, since every game elo gain and loss is based off a flat elo gain for win or loss, making more elo for shorter or longer games negates half the game, making it based on score wouldn't be effective either.

I dare you, make a test, send 5 people all with 5 wins 5 losses at one elo and then place them in a game with guys 200 wins and 200 losses same elo, the ones with more games will win almost everytime, thats a serious problem if you're a guy with 100+ games played and your allies have under 5 or whatever, and yes ... many dont have full rune pages or morer than 14 champs -.-.

Personal performance is the only metric that would be legitimate for changing the win/lose gains but Riot doesn't like that idea and doesn't believe it's possible.

I got an idea... compare player score to that of the community on that champion... if they beat the average they gain 1 extra elo, if they break even they gain normal elo, if they do worse they gain 1 less elo for wins, the opposite effect for losses and then average in team score at that elo, beat the averae as a team? Everybody gets 1 extra elo, break even? Same elo.... do worse than average? Gain 1 less elo.

Possible to win and gain 2 less effective elo for everygame giving a 20% change in elo gains further seperating baddies and goodies.

Edit@ Also, why not let people in solo queue chose to be placed with duos or not, giving further power to us to decide we prefer proper matchmaking, if we want to duo queue we can more likely b placed with other duos. Duos ARE a problem though yiou are right, it skews the game AGAINST the team with a duo, especially if they are far away in range, AND that duo will report anybody that suggests they are anything but top tier players -.-

It is possible. Let me give you some examples:

1. Players with few CS who play in bot tend to have more winning games when they place wards in this set of places at these times
2. Players who kill jungle creeps before 5 minutes tend to have more winning games when they kill or fight near the dragon before 9 minutes
3. Players who play in the top lane tend to have more winning games when they kill creeps in the enemy jungle
4. Players who play AD carry champions tend to win more games when they deal more damage to turrets
5. Players who play Teemo tend to win more games when they place more mushrooms
6. Players who play Shaco tend to win more games when they kill both blue and red buffs before 2 minutes, 15 seconds

I made that list up based on some random poorly thought out stuff that may or may not be correct. You'd have to examine some data from a large set of games even to make the list. It's not discoverable unless you have access to the database (because of sample size issues and obscurity - we don't know what the criteria are and we can't play enough games to correlate anyway), so it's not abusable as long as the list is secret. If the list is compromised change the criteria.

easy peasy, just takes a little while

Hey Lyte, would you guys ever consider opening up a small anonymized database of just who in a game was grouped, their normal elos at the time, and the results of the games? Then some of us with math backgrounds could play around with it for you guys. Though selfishly I'd just like to test my theory a couple pages back (last post on 75). I'll try and clean it up, wrote it on a touch-screen keyboard originally

Hey I'm not arguing against it, I like the idea of personal performance affecting elo gains but Riot has said time and time again "This is a team game, you win or lose as one."

1) Grant the Ant
2) I am talking about the Ranked game mode
3) Summoner’s Rift
4) No, I am solo queuing
5) I am level 30, my ranked elo is low around 700

6) Hello, I'm not sure if this is the right place to make this suggestion, please move it if required.

I am playing in low elo ranked games (around 700), and right now there is a large amount of queue dodging going on.

I think there is a problem with incentives at this elo level that i would like to address. There are many good players at this level despite what u might think. However, there are others that struggle a lot with this game, and are extremely negative. Most of the time at this elo level you can tell during champion select if the team you have is going to win or lose. I'd like to set up a few examples for you.

example 1 (A winning teams champ select):
player 5: "I am best at mid, but i can try to play an role"
player 2: "I see the other team picked Karthus mid, and i can beat him bad in lane with Mordekaiser"
***player 2 picks Mordekaiser***
player 5: "Ok I'll support bot"
***player 5 picks soraka***
I estimate i'm winning 90% of these games

example 2 (A losing teams champ select):
player 5: "I am AP mid"
player 2: "I see the other team picked Karthus mid, and i can beat him bad in lane with AP Mordekaiser"
***player 2 picks Mordekaiser***
player 5: "You suck i called mid you noob. I hate you."
***player 5 picks Swain**
player 5: "I am going mid i don't care what u do!"
(Morde is forced to support bot or jungle or something weird)
i estimate i'm losing 90% of these games

The people at my elo level that are climbing out are dodging every queue where there is team discord in champ select like in example 2, and only playing games where their team is working together in champ select like in example 1. I would like to do this as well, but I think queue dodging is wrong, so I am sticking out in the games where there is a lot of discord in champion select like in example 2.

What seems to be happening right now is a get 2 or 3 champ selects where my team works well together like in example 1. These champ selects seem to end with somebody on the other team dodging. Then I get on into a champ select where there is strong team discord on my team like in example 2, and no one on the other team dodges. Since i don't like to queue dodge I play these games out. Usually they end in a loss. This causes me to get to even lower elo.

I believe a possible “Band-Aid” for this problem would be to split the lost elo from a player's queue dodge equally between the players on the other team. That way I would be getting something for these games where my team is most likely going to win, but end in the enemy team queue dodging. As it is, I am feeling more and more incentive to queue dodge in order to escape this elo, and I don't want to be that guy.

So to summarize here is the issue. In low elo ranked games players can tell most of the time in champ select if they are gonna win or lose. Once they detect they are part of a losing team their main incentive is to dodge, or better yet convince an ally to dodge. Right now players at my elo that respect the summoner’s code, and are unwilling to dodge are getting 2-3 queues part of the winning team that end in an opponent dodging. They get no elo for these queues since their opponents dodge. Then they get matched up on a losing team. Since they are unwilling to queue dodge, they play out the game, and get even lower elo.

So the basic problem i am having with the ranked match made system is this. I am getting 0 elo for games where i am matched on the winning team because my opponents are dodging. When I get matched on the losing team i am unwilling to dodge so i lose more elo.

7) I don't have a good example of a specific end game score since talking about queue dodging in champ select.

(Yes, i did just talk about incentives in league of legends, and to answer your question, yes I am an economics major. Also, I should charge you some sort of consulting fee. I believe that's what these econ firms do to make money ;-)

At this time, there are no plans to open up an anonymous database for others to tackle; however, I did enjoy following 'competitions' other companies have done for their algorithms whether it is recommendations for movies, or improving matchmaking in Chess.

The good news is, I have been looking for brilliant analytical people to join the team working on this problem and they can tackle all the data they want internally; so if you are reading this and have a strong education and experience tackling massive datasets, predictive modeling, Bayesian approaches, machine learning and more... apply to Riot!

Make an option "Don't match me with duo's".

This will make queue times longer, for sure. But I don't mind having to wait more for a better quality game, specially when we're talking about rankeds.

Electrical engineering student with experience in MATLAB, programming, PDEs, reasonably advanced lin alg and complex calc, and I'll have Maple experience next year too I believe. Graduating next year from McGill University. What position would I apply for to join your team? Sorry to do this on the forums, wasn't sure where else to ask.

Edit: Oh and even come Access experience with databases

took me 50 games to get to 1600's
200 games later still there
curse this true elo