Hires Chess Trainer 1.0 - open-source GUI that trains on your own games

[gaelazzo]

Hi all,

I've released the first version of Hires Chess Trainer, a free and open-source
desktop GUI built around one idea: train on the user's OWN mistakes instead of
generic puzzles.

How it works:
- Imports your Chess.com/lichess games (incremental, append-only, dedup by the
[Link] URL, so you keep one growing PGN per player).
- Runs Stockfish (UCI) over them and flags every position where the move played
loses more than a configurable centipawn threshold.
- Those positions become a "learning base" and are served back with spaced
repetition; each attempt is re-checked by the engine.
- Endgame mode judges moves with Syzygy tablebases (<=7 men) and falls back to
Stockfish out of range.
- Opening mode walks PGN variation trees; per-position statistics are computed
against a reference PGN (your own games DB) and cached to disk.
- Full analysis board with variations, annotations and comments.

Stack: Python 3, pygame, python-chess (UCI + Syzygy probing), Stockfish.
License: MIT.

Windows: one-click build, no Python needed (Stockfish not bundled - you drop
it into engines\, instructions included). Runs from source on Linux/macOS in
principle, but those are currently untested - treat them as experimental.

Repo: https://github.com/gaelazzo/python_chess
Download: https://github.com/gaelazzo/python_ches ... ses/latest

This is the first public release. Feedback and criticism are very welcome, and
I'll fix bugs quickly.

[Frank Quisinsky]

Hi Gaetano,

I only took a quick look at your work.
Definitely, that looks really good !!!

I am working on a chess computer tourney.
Right now, I'm embroidering various sets of figures with new felt in purple, pink …

What you're offering here seems to be another great achievement for the community.

Play against a computer is great!
Important is that users can add other engines like Stockfish, allways Stockfish is a bit boring.

Wasp is great ...
Playing strenght build with nps ...
For the level from Novag Super Conny chess computer Wasp need around 40nps

So many others engines can be interesting, of course not Wasp only.

Example:
Very popular in the chess-computer community is CT-800 for self-playing chess.
https://www.ct800.net/

With different engines you can reduce the playing strength.
So, I am thinking a chess-trainer would help a lot, perhaps also to bring together many of the good ideas from chess programmers over the long term.

For "Study Openings" you will have many possibilites to make this point great.
I don't even want to think about all the possibilities your topic opens up.

Your GUI offer many interesting points!

I will add the link in my link-selection in the next minutes.
https://www.amateurschach.de/main/_comparable.htm

Edit: I do that!

What I can do for the moment!
THANK YOU!

Best
Frank

[Frank Quisinsky]

Such a GUI can be also use for a blunder-check.

Example:
Programmer from Engine A produced 500 games to a new version.
Now he can start a blunder check from the games with your GUI.

In the past I created such things with Martin Blume (Arena GUI, blunder-check).

Of course, it would be better to have a "Special-GUI" for such things that handles this and offers many additional features.

[gaelazzo]

Thank you!
Note that when you analyze a game and you have set a reference database (a pgn with a set of games, I use the set of all my 40k+ games) you can obtain the stats of all the times you reached that position, what were the next moves and what results did you get in that position.

You can also practice endgames easily: you create a pgn with the endgame start position and then you start practicing.
You can analyze all your games and search them for major or frequent blunders, he will create a learning base that you can use to learn from your mistakes and never make them again.
And much more, really!!!
Have fun playing chess

[Frank Quisinsky]

Interesting ...
I've now added your site to my engine overview as well.

Engine-Overview:
https://www.amateurschach.de/main/_engines.htm

Best
Frank