Page 1 of 2
ELO should work differently
Posted: 21 February 2025, 20:54
by Lucan 2
I believe ELO should work differently, I feel I usually lose more ELO than gaining.
Re: ELO should work differently
Posted: 21 February 2025, 20:55
by Jellby
If that were the case, you would be stuck at 100.
Re: ELO should work differently
Posted: 21 February 2025, 21:05
by Ceaseless
Lucan 2 wrote: ↑21 February 2025, 20:54
I believe ELO should work differently, I feel I usually lose more ELO than gaining.
The elo system is currently rigged in favor of players with 100 or less rating, preventing them from losing rating. Are you saying you no longer wish for this to be the case?
Re: ELO should work differently
Posted: 21 February 2025, 22:03
by Bray
Lucan 2 wrote: ↑21 February 2025, 20:54
I believe ELO should work differently, I feel I usually lose more ELO than gaining.
* Elo isn't an acronym. It's a person's name.
* Elo isn't a trophy achievement system. It's a mathematical model for predicting the probability of a win among a group of opponents.
* By playing enough games, and assuming that you're above 100, you'll approach a rank where you and players of your same rank will have an equal chance of winning.
Re: ELO should work differently
Posted: 21 February 2025, 23:56
by Meeplelowda
Lucan 2 wrote: ↑21 February 2025, 20:54
I believe ELO should work differently, I feel I usually lose more ELO than gaining.
You only gain and lose the same amount for a win or loss against someone with the exact same rating as you. If you win against someone rated higher than you, you will (and should) gain more rating points than you would lose if you lost to that same person. If you lose against someone rated lower than you, you will (and should) lose more rating points than you would gain if you won against that same person.
Re: ELO should work differently
Posted: 22 February 2025, 20:48
by pjt33
Bray wrote: ↑21 February 2025, 22:03
It's a mathematical model for predicting the probability of a win among a group of opponents.
No, it's a model for estimating the probability of a win in a two-player game with no luck. When it comes to a group of opponents it is sometimes abused (BGA isn't the only offender), but it isn't really applicable.
Re: ELO should work differently
Posted: 23 February 2025, 19:30
by buu441
Jellby wrote: ↑21 February 2025, 20:55
If that were the case, you would be stuck at 100.
Uhhh...if this was the a definitive indicator of elo not working properly, I don't blame OP for thinking so.
Re: ELO should work differently
Posted: 23 February 2025, 22:02
by Ceaseless
buu441 wrote: ↑23 February 2025, 19:30
Jellby wrote: ↑21 February 2025, 20:55
If that were the case, you would be stuck at 100.
Uhhh...if this was the a definitive indicator of elo not working properly, I don't blame OP for thinking so.
I believe it was meant to be the indicator of them losing more rating than they were gaining, causing them to hit the floor. Basically, if a player performed so poorly BGA doesn't actually have a rating low enough to reflect it and no longer had anything to take away.
Re: ELO should work differently
Posted: 27 February 2025, 18:16
by Lucan 2
Meeplelowda wrote: ↑21 February 2025, 23:56
Lucan 2 wrote: ↑21 February 2025, 20:54
I believe ELO should work differently, I feel I usually lose more ELO than gaining.
You only gain and lose the same amount for a win or loss against someone with the exact same rating as you. If you win against someone rated higher than you, you will (and should) gain more rating points than you would lose if you lost to that same person. If you lose against someone rated lower than you, you will (and should) lose more rating points than you would gain if you won against that same person.
No, but if you lose, it should be like 4 or 5 ELO. Never more than 10. If you win also 4 or 5.
Re: ELO should work differently
Posted: 27 February 2025, 19:46
by Meeplelowda
Lucan 2 wrote: ↑27 February 2025, 18:16
Meeplelowda wrote: ↑21 February 2025, 23:56
Lucan 2 wrote: ↑21 February 2025, 20:54
I believe ELO should work differently, I feel I usually lose more ELO than gaining.
You only gain and lose the same amount for a win or loss against someone with the exact same rating as you. If you win against someone rated higher than you, you will (and should) gain more rating points than you would lose if you lost to that same person. If you lose against someone rated lower than you, you will (and should) lose more rating points than you would gain if you won against that same person.
No, but if you lose, it should be like 4 or 5 ELO. Never more than 10. If you win also 4 or 5.
What do you mean "should be"? Do you mean you think the
formula in the FAQ is being applied incorrectly? If so, give an example of a game where it was incorrectly calculated.
If that's not what you mean, then what
mathematical model did you use to arrive at 4 or 5?