Board Game Arena

Sorry if this has been asked - turns out, there are too many results just searching "ELO," so you get nothing instead

If I win or lose a game, does it matter by how much I win or lose when my new ELO is computed? Or is it just the binary 1/win, 0/lose that matters? (or the .5 for a tie)

Thanks.

Score is irrelevant, ELO only cares about how many other players you beat or lose to, and what your respective ELOs were.

Maybe it should matter? How seriously has this been considered? Could games implement an additional K-factor to approximate "how much" a player lost by and factor that into ELO? Theoretically a player should lose less ELO if they lost by 1 point than if they lost by 10.

Fletcheese wrote: ↑27 September 2022, 19:18 Maybe it should matter? How seriously has this been considered? Could games implement an additional K-factor to approximate "how much" a player lost by and factor that into ELO? Theoretically a player should lose less ELO if they lost by 1 point than if they lost by 10.

Why?

In many cases you have win more skilfully to win by 1 point in a close game than to trounce someone by 10 points. In many many ganes the biggest wins happen when one player gets lucky and another unlucky, not because the winning player is amazing at the game.

frogstar_A wrote: ↑27 September 2022, 19:48

Fletcheese wrote: ↑27 September 2022, 19:18 Maybe it should matter? How seriously has this been considered? Could games implement an additional K-factor to approximate "how much" a player lost by and factor that into ELO? Theoretically a player should lose less ELO if they lost by 1 point than if they lost by 10.

Why?

In many cases you have win more skilfully to win by 1 point in a close game than to trounce someone by 10 points. In many many ganes the biggest wins happen when one player gets lucky and another unlucky, not because the winning player is amazing at the game.

Additionally, a 10-point win means different things in different game. A 10-point win in a game in which the average score is 15 points would not mean the same as in a game in which the average score is 50 points. So is there supposed to be a different adjustment factor scaled for every single game?

And many abstract games don't even have a score.

Meeplelowda wrote: ↑28 September 2022, 02:00

frogstar_A wrote: ↑27 September 2022, 19:48

Fletcheese wrote: ↑27 September 2022, 19:18 Maybe it should matter? How seriously has this been considered? Could games implement an additional K-factor to approximate "how much" a player lost by and factor that into ELO? Theoretically a player should lose less ELO if they lost by 1 point than if they lost by 10.

Why?

In many cases you have win more skilfully to win by 1 point in a close game than to trounce someone by 10 points. In many many ganes the biggest wins happen when one player gets lucky and another unlucky, not because the winning player is amazing at the game.

Additionally, a 10-point win means different things in different game. A 10-point win in a game in which the average score is 15 points would not mean the same as in a game in which the average score is 50 points. So is there supposed to be a different adjustment factor scaled for every single game?

And many abstract games don't even have a score.

Yeah the intention of my suggestion was for this to be implemented on a game-by-game basis at the developer's discretion. Some games don't really even have "points" per se so the "amount won by" would need to be estimated by some other metric. I'm not necessarily saying it's a good idea but I do think it's worth entertaining. It would certainly open up a metagame of playing for a close game instead of just playing to win. Taking that big risk on the last turn now comes with the drawback of potentially losing more ELO if it doesn't pay off so maybe you just take the safe play and lose less.

Thanks for the quick response. I assumed it was just "binary," as the formula in the FAQ suggests.

I think NOT having the score difference matter is absolutely the correct way to go. Especially since having it mater could add an additional level of kingmaking that no one should want. Also, in 3+ player games, when you're the player who is totally out of it - for me anyway - it frees me up to just make a quick, benign move, rather than *have* to fully analyze my "best" move because it might affect my ELO.