Provide a new statistical tool for evaluting chess postions (in addition to standard score). This tool is stability score (0-100). And describes how easy it is to break position / how stable the position is.
- Download stockfish
- Do:
****
pip install jupter
pip install python-chess
git clone git@github.com:eyalk11/chess_stability.git
- Update
STOCKFISHPATH
in the notebook.
cd chess_stability
jupyter lab -> select notebook
You can view your own games by adding a cell display_game(pgn)
(This shit is interactive!)
or evaluate a position by display_board(fen,iswhite)
.
By default, takes iswhite
from fen
.
See examples.
We define reasonable move (or not blunder) , as a move that does less damage than 200 centipawns.
We define a function
The method is the following:
- Evaluate all reasonable moves of certain depth (for both players)
$R$ - Evalute current position
$\mu$ - If the position is worse than 200 centipawns (to each direction):
- Otherwise:
- Denote stability by
$\frac{V}{|R|}$
This is a twist on gaussian distrubtion, taking current position as
It is licensed under by-nc-sa
Contact me for questions
Eyal Karni