Boards›Gameplay

Winning/losing doesn't contain enough information for efficiently determining ELO.

Great Ozymandius·5/26/2018, 4:18:44 PM·3 votes·1,312 views

If you are a skilled toplaner, and, at your current ELO, win 56% of your games, it will take a long time to gain ELO.

Yes, if you are better than the average toplaner in your elo, you will climb, but not quickly.

The problem is that winning or losing is the only way to gain elo. Bad matchmaking and simple bad luck plays a large role in your overall winrate, which will obscure the signal of your skill.

RIOT has said that they will not incorporate things like K/D/A and cs into ELO calculations, and that is a good thing. You don't want to create incentives that cause players to make decisions that improve their personal ELO while also reducing their chances of winning the game.

But there's another way around this. If there are objectives on the map that are necessary for the team to win, getting these objectives should reduce the amount of ELO you lose when you lose the game. Conversely, losing these objectives should reduce the amount of ELO you gain when you win the game.

Take inhibitors, for example. You need to take at least one inhibitor to win the game. If you take an enemy inhibitor and lose, it was a closer game. Give the team that lost but got an inhibitor more ELO. Give less elo to the team that won the game but lost an inhibitor.

This will improve the efficiency of gaining/losing ELO without distorting the incentive to win.

11 Comments

Marshbouy5/26/2018, 4:33:15 PM3 votes

There are a couple problems I could see with this.

If one lane does extremely poorly and eventually loses an inhib while the other two do extremely well, do the other teammates deserve to lose LP because of one poor teammate? Even if they manage to carry 4v6, do they still deserve to gain less LP than usual because they lost an inhibitor.
If a team gets smashed early but manages to turtle until late game, get a baron, then steamroll down mid and win the game, do they deserve less LP because they lost 3 inhibs early, but only got one inhib before winning? I don't think so. They may have done poorly in the beginning, but if they managed to stall out the game long enough or win one crucial teamfight then they shouldn't gain a restricted amount of LP.

Khristophoros5/26/2018, 4:38:51 PM3 votes

I disagree with the idea that anything you do or don't do should decrease the value of a win. You can't contest every objective. Sometimes the better move is to just farm and wait for another opportunity. When this is the correct decision you should not be punished for it.

But the idea of softening the punishment of a loss is possibly ok if you're really careful about it. I dunno it's just hard to get it right because a lot of the time you could make a play that gets your team an objective but it wasn't the right play. Like if you suicide for a drake? I mean it's not bad exactly but the risk of failure is extremely high so it's not necessarily a signal of skill that you decided to do it.

ValyrianBlade5/26/2018, 5:26:09 PM3 votes

[{quoted}](name=Great Ozymandius,realm=NA,application-id=3ErqAdtq,discussion-id=cUVEz3Ts,comment-id=,timestamp=2018-05-26T16:18:44.740+0000)

If you are a skilled toplaner, and, at your current ELO, win 56% of your games, it will take a long time to gain ELO.

Yes, if you are better than the average toplaner in your elo, you will climb, but not quickly.

The problem is that winning or losing is the only way to gain elo. Bad matchmaking and simple bad luck plays a large role in your overall winrate, which will obscure the signal of your skill.

RIOT has said that they will not incorporate things like K/D/A and cs into ELO calculations, and that is a good thing. You don't want to create incentives that cause players to make decisions that improve their personal ELO while also reducing their chances of winning the game.

But there's another way around this. If there are objectives on the map that are necessary for the team to win, getting these objectives should reduce the amount of ELO you lose when you lose the game. Conversely, losing these objectives should reduce the amount of ELO you gain when you win the game.

Take inhibitors, for example. You need to take at least one inhibitor to win the game. If you take an enemy inhibitor and lose, it was a closer game. Give the team that lost but got an inhibitor more ELO. Give less elo to the team that won the game but lost an inhibitor.

This will improve the efficiency of gaining/losing ELO without distorting the incentive to win.

I've made this suggestion myself a load of times, but I now realize a problem with it.

It unfairly rewards players who play early game champions, and punishes players who play late game champions.

If I win 50% of my games by getting a big early lead and stomping efficiently, and lose 50% by either not closing out those leads or not getting them in the first place - I'll still climb pretty steadily because I get bigger wins and closer losses.

Conversely if I win 50% of my games by playing a late game hypercarry and stalling until I can 1v5, and lose 50% by being unable to stall, I lose early and win close games. I'll lose rating like mad.

Neither player is doing better than the other.

Emerald Fang5/26/2018, 5:22:34 PM1 votes

Problem is that if it was performance based, people would try to preserve KDA instead of trying to make risky plays that might be worth it. Sometimes a play where 3 of you die but you get an ace will win the game, but performance-based means that many players will just splitpush and avoid that altogether, making the game very unfun.

PhearBunny5/26/2018, 5:51:04 PM1 votes

there is no good way to determine ELO in a team based game where you essentially are at the mercy of a dice roll. just the unfortunate reality. Play to have fun, don't bother worrying about ELO. I promise you, you will be called a worthless piece of trash at every ELO in the game. Regularly.

DW Diana5/26/2018, 5:51:39 PM1 votes

[{quoted}](name=Great Ozymandius,realm=NA,application-id=3ErqAdtq,discussion-id=cUVEz3Ts,comment-id=,timestamp=2018-05-26T16:18:44.740+0000)

If you are a skilled toplaner, and, at your current ELO, win 56% of your games, it will take a long time to gain ELO.

Yes, if you are better than the average toplaner in your elo, you will climb, but not quickly.

The problem is that winning or losing is the only way to gain elo. Bad matchmaking and simple bad luck plays a large role in your overall winrate, which will obscure the signal of your skill.

RIOT has said that they will not incorporate things like K/D/A and cs into ELO calculations, and that is a good thing. You don't want to create incentives that cause players to make decisions that improve their personal ELO while also reducing their chances of winning the game.

But there's another way around this. If there are objectives on the map that are necessary for the team to win, getting these objectives should reduce the amount of ELO you lose when you lose the game. Conversely, losing these objectives should reduce the amount of ELO you gain when you win the game.

Take inhibitors, for example. You need to take at least one inhibitor to win the game. If you take an enemy inhibitor and lose, it was a closer game. Give the team that lost but got an inhibitor more ELO. Give less elo to the team that won the game but lost an inhibitor.

This will improve the efficiency of gaining/losing ELO without distorting the incentive to win.

I agree that the current all or nothing system is naturally flawed, however I also feel that working entirely around objectives is a bad idea.

Every champion has it's advantages and disadvantages. For example in midlane a Diana will find it much easier to take towers then say a lux because of her passive/natural build.

I'd be looking more towards a system that rewarded the player based on their contribution as a whole. For example the current mastery system which works off KDA, CS, objectives. These should not of course dictate everything, the win is still important, but if you gained bonus LP based on things like this it would not only mitigate a loss where you couldn't carry but did well, while enchance the achievement of carrying your team to win.

I feel this would be a far fairer and more beneficial option for improving ranked, especially for classes and champions that aren't DESIGNED to carry such as tanks and supports.