Antichess?
Very interesting! GJ @ubdip
A very interesting investigation, thanks a lot! I am a bit surprised that the value of a move should only be 0.01 pawns in standard chess. But I guess that it can just depend a lot on the phase of the game.
Larry Kaufman made a similar investigation for standard chess (though with a different method, www.danheisman.com/evaluation-of-material-imbalances.html) where he also looked at values of combinations of pieces. For example, he attributed the higher value of the bishop compared to knights to the existence of a bishop pair. In other words, a single bishop and a single knight are of closer value than the average, the bishop just gains a lot in value if only one player has the bishop pair. I wonder if similar correlations could be found for the variants that @ubdip investigated.
Larry Kaufman made a similar investigation for standard chess (though with a different method, www.danheisman.com/evaluation-of-material-imbalances.html) where he also looked at values of combinations of pieces. For example, he attributed the higher value of the bishop compared to knights to the existence of a bishop pair. In other words, a single bishop and a single knight are of closer value than the average, the bishop just gains a lot in value if only one player has the bishop pair. I wonder if similar correlations could be found for the variants that @ubdip investigated.
@FInLIp said in #2:
> Antichess?
Antichess piece values aren't very stable over the course of a game, so it is harder to interpret the values, which is why I left it out of the post. You can however find the results for more games, including antichess, in the wiki of the code repository, see github.com/ianfab/chess-variant-stats/wiki/Piece-values#antichess. IMO the (not normalized) values are pretty reasonable because rooks are very useful in the endgame, while bishops often can be more of a liability especially in the opening. However, as always static antichess piece values are rather hard to interpret
r 0.19
n -0.08
k -0.12
q -0.17
p -0.22
b -0.34
move -0.40
> Antichess?
Antichess piece values aren't very stable over the course of a game, so it is harder to interpret the values, which is why I left it out of the post. You can however find the results for more games, including antichess, in the wiki of the code repository, see github.com/ianfab/chess-variant-stats/wiki/Piece-values#antichess. IMO the (not normalized) values are pretty reasonable because rooks are very useful in the endgame, while bishops often can be more of a liability especially in the opening. However, as always static antichess piece values are rather hard to interpret
r 0.19
n -0.08
k -0.12
q -0.17
p -0.22
b -0.34
move -0.40
Very good blog, @ubdip !
what about
q=4.89
r=5.72
b=6.39
n=3.64
p=2.56
k=0.00
move=99.99
q=4.89
r=5.72
b=6.39
n=3.64
p=2.56
k=0.00
move=99.99
@ubdip Dein Blog ist sehr gut entworfen. Allerdings sollte man im normalen Schach dem König unbedingt auch einen Wert zuschreiben! Und dieser Wert beträgt meiner Meinung nach 4 Bauerneinheiten. Es sei denn, der König muss alleine gegen zwei oder mehrere gegnerische Freibauern kämpfen. Dann sinkt sein Wert auf zwei Bauerneinheiten.
@philodendron68 said in #8:
> @ubdip Dein Blog ist sehr gut entworfen. Allerdings sollte man im normalen Schach dem König unbedingt auch einen Wert zuschreiben! Und dieser Wert beträgt meiner Meinung nach 4 Bauerneinheiten. Es sei denn, der König muss alleine gegen zwei oder mehrere gegnerische Freibauern kämpfen. Dann sinkt sein Wert auf zwei Bauerneinheiten.
Das finde ich gar net. Der König ist viel wichtiger als alle Figuren zusammen. Lieber hätte ich keine schwerfiguren als keine könig
> @ubdip Dein Blog ist sehr gut entworfen. Allerdings sollte man im normalen Schach dem König unbedingt auch einen Wert zuschreiben! Und dieser Wert beträgt meiner Meinung nach 4 Bauerneinheiten. Es sei denn, der König muss alleine gegen zwei oder mehrere gegnerische Freibauern kämpfen. Dann sinkt sein Wert auf zwei Bauerneinheiten.
Das finde ich gar net. Der König ist viel wichtiger als alle Figuren zusammen. Lieber hätte ich keine schwerfiguren als keine könig
@philodendron68 This is a different kind of piece value then. What my definition of piece value measures is the effect of the presence of a piece on the game result, i.e., essentially its correlation. If a piece is always present, like the king, you can't observe any correlation whatsoever, and you can't compare positions with and without king. So you first need to consistently define the your concept of piece values before you can meaningfully assign values.