#29897: "ELO system (for Spades) unduly penalizes player for being teamed with weaker player."

• Please copy/paste the error message you see on your screen, if applicable.
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Please explain what you wanted to do, what you did and what happened
• Which browser are you using?
Google Chrome v87

• Please copy/paste the text displayed in English instead of your language. If you have a screenshot of this bug (good practice), you can use Imgur.com to upload it and copy/paste the link here.
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Is this text available in the translation system? If yes, has it been translated for more than 24 hours?
• Which browser are you using?
Google Chrome v87

• Please explain your suggestion precisely and concisely so that it's as easy as possible to understand what you mean.
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Which browser are you using?
Google Chrome v87

• What was displayed on the screen when you were blocked (Blank screen? Part of the game interface? Error message?)
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Which browser are you using?
Google Chrome v87

• Which part of the rules was not followed by the BGA adaptation?
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Is the rules violation visible on game replay? If yes, at which move number?
• Which browser are you using?
Google Chrome v87

• What was the game action you wanted to do?
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• What did you try to do to trigger this game action?
• What happened when you tried to do this (error message, game status bar message, ...)?
• Which browser are you using?
Google Chrome v87

• At which step of the game did the problem occur? What was the current game instruction?
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• What happened when you tried to do this game action (error message, game status bar message, ...)?
• Which browser are you using?
Google Chrome v87

• Please describe the display issue. If you have a screenshot of this bug (good practice), you can use Imgur.com to upload it and copy/paste the link here.
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Which browser are you using?
Google Chrome v87

• Please copy/paste the text displayed in English instead of your language. If you have a screenshot of this bug (good practice), you can use Imgur.com to upload it and copy/paste the link here.
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Is this text available in the translation system? If yes, has it been translated for more than 24 hours?
• Which browser are you using?
Google Chrome v87

• Please explain your suggestion precisely and concisely so that it's as easy as possible to understand what you mean.
The existing ELO calculation after a win (or loss) treats the result as a win (loss) against each of the opponents, **and a tie with the player's partner**. That means that if you are paired with a weak player, ot only is winning more challenging (which is fine, luck of the draw) but also your ELO is docked because of partner's weakness.
As an extreme example, say a 600 is paired with a 20 ELO, playing against a couple of 310 ELO opponents, and everybody has K=20. If the 600/20 team wins, the 600 gains about 4 points against each opponent, and ties his partner, which is about (-18+2)/2 = -9 ELO points. The sum is -1, so the expert player, having beaten a couple of strong players despite being paired with a beginner, would *lose* about half an ELO point. Since the odds of winning, based on linear approximation using the combined ELO's, were about 50%, that expert (and his partner) should have gained 10 points.
The best solution is to treat the game as a two-player game, using the pair-averaged ELO rating for each pair to determine the effective ELO. If that is awkward within the overall BGA framework, then at least the system should not be indicating a tie against your partner.
• Which browser are you using?
Google Chrome v87

What is this report about?

What happened? Please select from below

What happened? Please select from below

Please check if there is already a report on the same subject

Detailed description

Report history

Add to this report

Change report status to:

Help

Contribute

Follow us