If I have an ELO of 300 and my partner has an ELO of 100, he should receive significantly more ELO points than me if we happen to beat, say, a team of two players whose ratings were 200 and 200.
ELO calculation for joint placements.
Forum rules
Warning: challenging a moderation in Forum = 10 days ban
More info & details about how to challenge a moderation: viewtopic.php?p=119756
Warning: challenging a moderation in Forum = 10 days ban
More info & details about how to challenge a moderation: viewtopic.php?p=119756
- vigorousRoll
- Posts: 95
- Joined: 17 February 2018, 09:46
Re: ELO calculation for joint placements.
Re: ELO calculation for joint placements.
In that case, your partner will receive additional ELO points for tying with you, while you will lose points for tying with a lower-ranked player, which is what we're trying to get addressed next herevigorousRoll wrote: ↑15 June 2019, 18:05 If I have an ELO of 300 and my partner has an ELO of 100, he should receive significantly more ELO points than me if we happen to beat, say, a team of two players whose ratings were 200 and 200.
Receiving different ELO points for defeating the "same" common opponents is another matter, at least as far as I'm concerned. But ELO computation between winning parties for anything considered a tie (e.g. team games, quitting tables) should not take place in an ideal ranking system.
Re: ELO calculation for joint placements.
Why should he get more points? Did he contribute more to the win than you did?vigorousRoll wrote: ↑15 June 2019, 18:05 If I have an ELO of 300 and my partner has an ELO of 100, he should receive significantly more ELO points than me if we happen to beat, say, a team of two players whose ratings were 200 and 200.
That's just... wrong.
Check out the Allround-League: https://boardgamearena.com/group?id=7870115 --> a league where you have matches of random games vs. other players in your group. Season 9 registration oct. '25.
Re: ELO calculation for joint placements.
In the Elo system, the higher-ranked players have higher expected win probabilities against lower-ranked players. So no, they did not contribute more to the win, but they weren't expected to win at all as much as the higher-ranked player was. This results in a greater ELO point increase. For team-vs-team games on BGA, the same applies.
You're free to lobby for the change you're envisioning, but please open a new thread dedicated to that proposal, since right here we'd like to get placement ties properly implemented w/r/t ranking outcome, not re-compute team-vs-team ELO. Anything on top of that leads to confusion when folks read through the replies
Re: ELO calculation for joint placements.
F and N are in team and win a doubles tennis match.
How about someone claiming that F and N were tied?
Do you think it would be legitimate to better reward the less classified of F and N?
How about someone claiming that F and N were tied?
Do you think it would be legitimate to better reward the less classified of F and N?
Re: ELO calculation for joint placements.
Again, those two statements are separate issues. The first statement, claiming that there's a tie when really there is not and cannot be, is a real problem and you've correctly identified why it should not work like that. Right now, it does however on BGA!
The second statement, implying that somebody gets rewarded, is exactly correct: Over time, to find your true Elo rank, you must benefit more from winning against higher-ranked opponents. This works both ways and, while not obvious at first, actually is the fairer system than handing out an "average" to all players in the same team.
With that system, it would then become beneficial to grab yourself a few rookies, coach+carry them to a team victory, and reap a windfall of ELO points if you succeed – since they'll significantly boost your team's average Elo gain – while not losing that much of you don't. But yeah, let's continue that part of the conversation here: https://boardgamearena.com/forum/viewto ... =3&t=13760
The second statement, implying that somebody gets rewarded, is exactly correct: Over time, to find your true Elo rank, you must benefit more from winning against higher-ranked opponents. This works both ways and, while not obvious at first, actually is the fairer system than handing out an "average" to all players in the same team.
With that system, it would then become beneficial to grab yourself a few rookies, coach+carry them to a team victory, and reap a windfall of ELO points if you succeed – since they'll significantly boost your team's average Elo gain – while not losing that much of you don't. But yeah, let's continue that part of the conversation here: https://boardgamearena.com/forum/viewto ... =3&t=13760
Re: ELO calculation for joint placements.
Well, the thread-opener specifically referred to the case where joint placements arise because the affected players are in a team (referring to NotAlone and Werewolves). But ok I'll reply in the new thread.eoc wrote: ↑15 June 2019, 22:02 You're free to lobby for the change you're envisioning, but please open a new thread dedicated to that proposal, since right here we'd like to get placement ties properly implemented w/r/t ranking outcome, not re-compute team-vs-team ELO. Anything on top of that leads to confusion when folks read through the repliesThanks!
Check out the Allround-League: https://boardgamearena.com/group?id=7870115 --> a league where you have matches of random games vs. other players in your group. Season 9 registration oct. '25.
Re: ELO calculation for joint placements.
Exactly. Again, since I'm not sure I can be any more clear, sorry if it sounds condescending at some point: Exactly. The issue being that ties cause ELO movement within the team. Going from the higher-ranked players in the team, to the lower-ranked players in the same team that all ended in the same place. And played on the same team. This is not about any other team's involvement, not even about winning or losing. Not about how much your rank should change because of your win, or your defeat. Just that it shouldn't simply because you played in the same team as good/bad/new players.