Hanabi bots' ELO

[dskloet]

I agree that taking the average of the ratings is bad. I've noticed that most people don't want to play with lower rated people because it almost always means you lose rating. The virtual rating of the team should be much closer to the minimum among players.

I think a ELO bonus should be given to games ending with no bombs. We cannot do perfect games in difficult decks, but no bombs means a good cooperation. So players in thes games should be rewarded.

On the contrary. Bombs are a resource to be used. If you end the game without bombs it means you didn't manage to use your resources efficiently and should have taken more risk. If there's any bombs left in the last round and there's a chance I might have a playable card, I don't care if I risk getting a bomb because the game is about playing cards.

[Jellby]

Is it so much more difficult to play multicolor with 2 players? With 5 or 6 colors, the higher the number of players the higher the rating, but with multicolor 2 players is actually higher than 3 or 4 (and only 10 points lower than 5).

[qpenguin]

@Jellby

To achieve a consistent score of 30, 2 players are actually more difficult than 3/4 or sometimes 5 players. This is because you have a very limited hand space (2p=10cards, 3p=12cards, 4p=16cards, 5p=20cards) and thus sometimes have to sacrifice cards.

Unfortunately, this high elo has been exploited by many players who abandon games when the situation is not going in their favor. The high elo, quick set up and quick play, and ease to abandon game (only needs 2 players to agree) of 2p games make it easy to boost one's elo. I've seen players with more than 40% abandon rate. That means they quit almost every other game. How ridiculous do you think this is?

[ollyfish2002]

I feel that there is a lot of 2p games for boosting ELO.
I just think it could be helpful to be select the kind of games I don't like (like 2p Hanabi). It would be nice not to see them and not be invited.
Just a question :
How do you see abandon rate ?

[qpenguin]

@ollyfish2002

Go to the player's profile and under Hanabi select XX's statistics at this game. On the top left corner of the page under the player's name, click the number of games. You can then see the number of completed games the player has. Now untick the option "Finished games only". That would be his total games started. The rest will be maths.

[ollyfish2002]

thanks

[Jellby]

To achieve a consistent score of 30, 2 players are actually more difficult than 3/4 or sometimes 5 players. This is because you have a very limited hand space (2p=10cards, 3p=12cards, 4p=16cards, 5p=20cards) and thus sometimes have to sacrifice cards.

What puzzles me is that such a difference is only seen (in the ELO points) in multicolor mode, not in 5 or 6 colors. I have not enough experience to know if this matches the actual difficulty.

[Jellby]

Have the bots' ELO been updated or is the ELO calculation still done with the "old scale" (+1300) internally? I've noticed that a failed game is now -10 instead of -20...

[demiurgsage]

There was an update of K-factor. You can find it in the recent news of BGA.

[RobertBr]

Is it possible to see a measure of player results for each of the variants and number of players, ie what percentage of games do end in a 23 for the base game? Or, potentially interesting, what is the average Elo of the last 100 games in which 4 players ended wih 23 in the base game?
The latter stat especially (and other measures like SD, max, median) might give a clue as to whether the bots are over/under rating the difficulty of particular configurations.