Visualizing and learning from games on chess.com

Better Your Chess

What for?

This project aims to help you learn from all the chess games you've played online, starting with a simple way to download an organized collection of your games!

On Chess.com, you can view your winrate by color, your best win, and analyze a game per day for free. Building on top of the Stockfish engine and python-chess, this package aims to give the user: winrates by opening, automatic tactic generation from your library, and common deviations from book moves.

Current features:

winrate by opening
rating distribution

In progress:

automatic tactics generation from your library

Future features:

show opening positions where you often deviate from book moves
loss frequencies (checkmate, resignation, timed out)
suggest grandmasters with similar repertoires

Components

notebooks/intro_to_python-chess.ipynb - intro to working with PGNs via python-chess and polyglot opening books.
scripts/pgn_downloader.py - download Chess.com games for a user, organized in year/month directories.
```
python scripts/pgn_downloader.py username
```
chess_analytics/game_reader.py - implements a GameReader, class to work with PGNs.
chess_analytics/game_library.py - build a library of games from a directory, represented as a dataframe with 1 row per game.
```
from chess_analytics.game_library import GameLibrary
library = GameLibrary("data/user_games/user/")
library.df.head()
```

df_sampled

scripts/plotting.py - visualize a library: opponent rating distribution, winrate by opening, and more.

Rating_distribution

Winrate_by_opening_white

scripts/generate_tactics.py - Automatically generate tactics from a library. This currently supports mate-in-Ns, producing a JSON and PNGs of the positions. 10 examples are in data/output/tactics/.

KnightofCamelot_vs._Luc777_w_to_move