ethanjtang/PAWN-piece-value-datasets

PAWN: Piece Value Analysis with Neural Networks

arXiv LINK COMING SOON

We define piece value as the difference in Stockfish evaluation between the original position and the position with that piece removed.
Both datasets used Stockfish 17 at depth=20 with a timeout=300s per evaluation for generating piece values.

MC-Large — Derived from 6,925 Magnus Carlsen games: 11,673,269 piece value entries across 549,410 unique positions.
TF — Derived from 7,656 GM-level Classical games (both players ≥2500 FIDE Classical Elo, played in 2023): 12,263,049 piece value entries across 533,540 unique positions.

Dataset Structure

Column	Description
game_id	Internal ID identifying the worker/game
fen	Original position as a FEN string
move_number	Move number
side_to_move	Side to move (w or b)
eco_code / opening	ECO code or opening name
white_material	Material string for White (e.g. pNBRQK)
black_material	Material string for Black (e.g. pppNNK)
piece_type	Piece character (uppercase = White, lowercase = Black)
rank	Rank index 0–7 (ranks 1–8)
file	File index 0–7 (files a–h)
original_eval	Stockfish eval of the original position (centipawns)
eval_without_piece	Stockfish eval with the piece removed (centipawns)
piece_value	original_eval - eval_without_piece

Dataset Sample

Three piece value entries from the starting position after White's first move 1. d4, evaluating White's pieces on the first rank:

Piece	piece_type	rank	file	original_eval	eval_without_piece	piece_value
Knight on b1	N	0	1	29	-500	529
Bishop on c1	B	0	2	29	-561	590
Queen on d1	Q	0	3	29	-747	776

All three piece value entries share game_id=w0_g0, move_number=1, side_to_move=b, eco_code=A40, and opening=Queen's Pawn Game.

Citation

CITATION COMING SOON