Opening Inaccuracies Tool

@Toscani said in #17:

I should of used 10 games just to see the results before increasing it to 100. The wait time to get the results takes time.

It is mainly because it is a browser based stockfish so slightly slowly and it does take time. Games x ply gives you the total number of positions that stockfish will churn through. So that is 1000 positions at depth 24.

I am thinking of ways to optimize this, but that is pretty decent in terms of performance.

@asdf32608 said in #18:

Thanks for sharing. I am a programmer who dabbles with chess. Is the site running stockfish on the browser, or do you have a server backing this all?

I really like the UI and the idea. Some of the cases where I got worse than 1 cpl were interesting but unconventional lines (e.g. playing pawn moves for most of the first 10 moves to build a big center).

I tried it on my most recent 100 games as black and white. I ended up having a lot of positions only happening once. Also I had a lot of lines which were ok. For performance 2 pv and depth 15 seemed ok.

Yes, the site is running stockfish in the browser and using 8 individual instances to speed things up. I am looking at optimising though.

@Toscani said in #19:

One hour, 15 minutes and 0.5 seconds later ...
Why is PV1 and PV2 the same?

Yes, someone on my discord reported this bug, I need to have a look into it.

@Maximu5 said in #20:

what is the github link?

Close source project. No GitHub link available.

Sorry, if I am repeating someone else. My main gripe with it is that it is not that intuitive and I would like to see something like an amalgamation of 3 moves into the opening. I can transpose into the same 3 or 4 moves with very different starts, even maybe regardless of what opponent plays. I understand that the whole concept is that we try to find the mistake according to what opponent plays, but that could also be what I am practicing is how to play the same opening into almost everything and I try to find the adaptations in my opening and understanding of when it can be used or when it is basically impossible to play it.

For the user interface, you could build a split view (PGN File Comparison) on the left side showing the user's "analysis#explorer" games and on the right side showing an engine opening book. When the user clicks on their move, the right side board highlights the divergence between the user move and the engine opening book and shows possible continuations. The engine runs when the move is clicked. So you see a list of all possible PV lines. The engine chessboard lists the possible transpositions.

You should list the users moves like:
https://lichess.org/analysis#explorer/Toscani

I like this site, because it shows my performance results and which opening works best for me.
https://www.openingtree.com/

Maybe you could compare pgn openings like text files and analyse only what is different ...
https://www.diffchecker.com/cnlI4Wxs/

Your games have to be indexed to compare your moves vs the engine moves or other databases.
https://lichess.org/analysis/standard/rnbqkbnr/pppppppp/8/8/8/N7/PPPPPPPP/R1BQKBNR%20b%20KQkq%20-%201%201

Performance ratings are much more accurate over a series of many games. The less there are in our database, the less accurate are the values. It goes the same got a move. If we only played f3 once per year, it shouldn't even be worth analyzing. A year later we would have forgotten the correct replies. In 2025 after using chess tree website, I discovered that I never player f3 but I played every other legal first move. So guess what, today I played f3 just to say I played every first ply move in 2025. Also it was to fill in the blanks in my 2025 spreadsheet. The year is not over yet. So I can still play some other moves to fill in the zeros in the best win and worst loss columns of my spreadsheet.

NOTE: We need to correct our bad opening habits. The moves that are done very often and are not at all logical or helping us win games.

looks like a bug? the engine lines dont make sense

@SendMeCapybaraVideos said in #27:

looks like a bug? the engine lines dont make sense

Yes, I need to look into this, it is a first draft so there will be a few bugs.

nice one of the BEST thanks !

One day a prodigy starting with the benefit of all technology will reach his prime and beat every engine and it will be cool to witness.