This is a very nice analysis! Could you enlarge it to other games of other players? (e.g. by using the evaluated database database.lichess.org). I am most interested in the large average distribution over games of many players having similar strength (rating). Cheers, Nikodem
I'm analysing more games, but the main bottleneck with that is computing time. Currently I'm analysing the games from this Candidates tournament.
I'm a bit worried about using the Lichess evaluation database, since it uses evaluations from different versions of Stockfish and I'm not sure how this impacts the centipawn loss.
Sure, if you want to do a scientific project, you should be careful with various engine versions. But for the first observation how the distribution looks like I would take it. They are grouped by dates anyway, so once can restrict to a time period where they should be all analyzed by the same engine. Best
This is a very nice analysis! Could you enlarge it to other games of other players? (e.g. by using the evaluated database database.lichess.org). I am most interested in the large average distribution over games of many players having similar strength (rating). Cheers, Nikodem
I'm analysing more games, but the main bottleneck with that is computing time. Currently I'm analysing the games from this Candidates tournament.
I'm a bit worried about using the Lichess evaluation database, since it uses evaluations from different versions of Stockfish and I'm not sure how this impacts the centipawn loss.
Sure, if you want to do a scientific project, you should be careful with various engine versions. But for the first observation how the distribution looks like I would take it. They are grouped by dates anyway, so once can restrict to a time period where they should be all analyzed by the same engine. Best