No ELO?

[Stroom]

If the point is to increase community & cooperation, then 5p games should get the highest return on ELO, then 4p and so on. 2p should be the lowest. It's far easier to start a 2p game, it plays faster, and it's so much easier to find 1 other player at the same convention level as you than 4 others.

But I would be happier with a separation between 2p and Multi-player instead. I dislike playing with Masters who climbed the ranks only playing 2p -- not because I'm a 5p elitist, but because the conventions are vastly different, and they take a lot of time to learn. If I'm playing a Master-restricted 5p game, I expect Masters of 5p games, otherwise we're simply not speaking the same language.

All that being said, it still makes no sense to have invisible ELO. There are far better solutions for negativity at Hanabi tables, and it starts with helping players make their table's conventions more clearly known.

Why is elo supposed to "increase community & cooperation"? This does not sound right. Especially the conclusion that 5p is somehow better than 2p based on that. And that you get more return for larger player counts. Elo will even out at your eventual skill level so reducing the gains actually does nothing in the big picture.

I play 2p especially because it is faster. But that should not indicate that my elo would be higher than someone's who only plays 4p. It's something else. More likely because there is less of a chance that of a bad player joining the game. If I can carry one player by myself, it would be hard to carry 2. The play style is different, yes, but there are quite many nuances that 3+p players do not grasp, so they always make stupid mistakes. Sure, I would probably not play as well in 3p as my elo would indicate. I don't think there would really be a way to calculate the same elo score for both player bases anyway.

If someone has never played a 3+p game, you can not really judge their skill for that based on only 2p games. So any kind of common-calculation would become meaningless for other game variations. So, yes. Separating elo calculations based on specific variants makes a lot of sense here. Nothing else would.

[Travis Hall]

This does not seem to work. It assumes that all players play to the fullest of their skill. That there are no intentional quitters.

It makes no more or less such assumptions than any rating system. If there are factors that help a player get better results, that player’s rating moves higher. If there are factors that cause a player to get worse results (including quitters and bad partners) that player’s rating moves lower. A player who plays at less than their true skill (including quitting early) will be more affected by their own detrimental actions in the long term.

And seems like it averages all the players. Overall, it still seems like a subjective way to do it.

I have no idea what you are talking about here. Everybody always enters a rating system at the same point. Everyone always has their rating change in response to a shared outcome in a team game.

All I’ve really suggested is that you can use the ELO algorithm to actually measure how difficult different forms of the game are, and use that information to adjust how player ratings get adjusted.

Let me rephrase it - there is no practical way to do this. It needs far more effort than anyone willing to put into it, and even then, you might not see the benefits of it.

As far as I know, BGA does not support multiple ratings for the variations of any game on the site. Frankly, the admins are unlikely to make the infrastructure changes required to do that. From a system design point of view, I believe that what I outlined could be implemented without any impact on other games, so it may well be more practical to implement just due to that.

But I don’t believe you will ever get either. There’s not enough value in it for the site owners.

(You have seen me here wearing, in turn, my mathematician hat, my software engineer hat, and my solution architect hat.)

[Ant Stewart]

All I’ve really suggested is that you can use the ELO algorithm to actually measure how difficult different forms of the game are, and use that information to adjust how player ratings get adjusted.

Let me rephrase it - there is no practical way to do this. It needs far more effort than anyone willing to put into it, and even then, you might not see the benefits of it.

As far as I know, BGA does not support multiple ratings for the variations of any game on the site. Frankly, the admins are unlikely to make the infrastructure changes required to do that. From a system design point of view, I believe that what I outlined could be implemented without any impact on other games, so it may well be more practical to implement just due to that.

But I don’t believe you will ever get either. There’s not enough value in it for the site owners.

(You have seen me here wearing, in turn, my mathematician hat, my software engineer hat, and my solution architect hat.)

The ELOs as far as I understand (it seems to disapear into a black box so not fully sure on the implementation) are configured with a pseduo-oponent that is beaten each game (assuming you are in a win state). The ranking of the oponent is based on the number of players, the deck (normal, multi 10 cards, multi 5, avalence) and the end score.

It looks as though the ranking could be based on any parameters, but the number of configuration needing defining would scale exponentially. Eg, to include black powder and 5 flamboyants, would quadrupal the number of configured ELO scores.

It also appears as though, Difficult 6th colour (5 cards) has no ELOs specified. I assumed it inherits from Tricky 6th colour but it is not fully clear how that works as the ELOs are just missing.

There is some documentation on this here:
https://studio.boardgamearena.com/doc/G ... os.inc.php

I have definitely misunderstood something though, since it says if you beat a value consitently, you should converge on that value. But the ELO for scoring 18/25 on the base game with 3 players is 1400, but if you got 18 in every game, you would definitely not arrive at that score. And if you are ranked above 400 ish, you'd be loosing ELO each turn on that.

[Jellby]

The ELO for 18/25 in the base 3-player game seems to be 100, not 1400: viewtopic.php?t=4814

The maximum "bot" ELO is 1040. The maximum theoretical ELO is 5200: the average of 5-player group consistently getting 30 points with avalanche should converge to 5200, if you always do that with 4 other 0-ELO players, you'd get to 5200.

[Travis Hall]

The ELOs as far as I understand (it seems to disapear into a black box so not fully sure on the implementation) are configured with a pseduo-oponent that is beaten each game (assuming you are in a win state).

…a pseudo-opponent commonly called a bot. Yes, I’m aware of how the current implementation works.

It looks as though the ranking could be based on any parameters, but the number of configuration needing defining would scale exponentially.

… up to 1362 configurations, taking into account all variations currently available on the site. I provided the figure up-thread already.

It also appears as though, Difficult 6th colour (5 cards) has no ELOs specified. I assumed it inherits from Tricky 6th colour but it is not fully clear how that works as the ELOs are just missing.

No ELO change is calculated for Difficult, as it was judged to be too greatly affected by the luck of the draw to factor in alongside the other variations.