Point Value,
a piece relative value concerning its relative strength in potential exchanges based on human experience and learning. Common rule of thumb are the {1, 3, 3, 5, 9} point values for pawn (1), knight, bishop, rook and queen, also proposed by Claude Shannon in his 1949 paper Programming a Computer for Playing Chess[1] . In the evaluation of a chess program, the balance of material, the aggregated point values for both sides, is usually the most influential term.
Jack Good in his Five-Year Plan for Automatic Chess, Appendix F[3]:
A theoretical attempt to evaluate the pieces was made by H. M. Taylor in 1876[4], reported by Coxeter (1940, pp. 162-165[5]). The value of a piece is taken as proportional to the average number of squares controlled, averaged over all 64 positions of the piece on the board. This argument leads to the relative values of N, B, R and Q: 3, 5, 8 and 13[6]. Coxeter (or Taylor) goes on to modify the argument by asking instead for the probability of 'safely' giving check, that is, without being en prise to the king, if the piece and king are both placed on the board at random. This gives the ratios 12, 14, 22 and 40. These values are good, but this modification
of the argument is artificial.
Basic values
Measured in units of a fraction of a pawn, for instance the common centipawn scale, allows positional features of the position, worth less than a single pawn, to be evaluated without requiring fractions but a fixed point score. Sample point values from various sources over the time, most were used in concrete chess programs.
The king value is often assigned a large constant such as 10000 centipaws, which is important to avoid king captures in certain implementations of SEE.
Reciprocal piece values
Concerning a piece controlling a square, its value of attack might be considered as inverse proportional to its point value, which is an issue in aggregating of mobility or square control terms of different pieces.
Jens Christensen, Richard Korf (1986). A Unified Theory of Heuristic Evaluation functions and Its Applications to Learning. Proceedings of the AAAI-86, pp. 148-152, pdf
a piece relative value concerning its relative strength in potential exchanges based on human experience and learning. Common rule of thumb are the {1, 3, 3, 5, 9} point values for pawn (1), knight, bishop, rook and queen, also proposed by Claude Shannon in his 1949 paper Programming a Computer for Playing Chess [1] . In the evaluation of a chess program, the balance of material, the aggregated point values for both sides, is usually the most influential term.
Table of Contents
Theoretical Attempt
Jack Good in his Five-Year Plan for Automatic Chess, Appendix F [3]:of the argument is artificial.
Basic values
Measured in units of a fraction of a pawn, for instance the common centipawn scale, allows positional features of the position, worth less than a single pawn, to be evaluated without requiring fractions but a fixed point score. Sample point values from various sources over the time, most were used in concrete chess programs.The king value is often assigned a large constant such as 10000 centipaws, which is important to avoid king captures in certain implementations of SEE.
Reciprocal piece values
Concerning a piece controlling a square, its value of attack might be considered as inverse proportional to its point value, which is an issue in aggregating of mobility or square control terms of different pieces.Samples
See also
Publications
Forum Posts
1993 ...
2000 ...
2005 ...
2010 ...
2015 ...
Re: Pawn value estimation by Larry Kaufman, CCC, May 09, 2015
External Links
References
What links here?
Up one Level