Game opening checker

[Frank Quisinsky]

Hi Ferdinand,

your programs works also if I convert with ChessBase my main database from *.cbh to *.pgn format back. That is fantastic, so I can check all the games I have in my database too.

With other words ...
Great work by yourself and I can improved my database and opening book again.

Best
Frank

Have many thanks !!

[Ferdy]

Hi Ferdinand,

it works ...
Really great ... all what I need for checking new games, Shredder GUI saved.

My steps with new games in the past:

1. Loading the *.pgn database from Shredder in ChessBase GUI.

Now I know why there are two different comment format, if you save the game from CB GUI.

2. Looking for fast draw games, time-forfeit games, games with 300 moves (bad games).

This 300 and above moves, will you replay the games even if the result is 1-0 or 0-1? The command long-game in this tool will extract games with any results.

6. After a test-run I am looking for games same ECO code produced, double games undo move 8 or higher. Such games are replayed after a test-run too.

Point 6 I forget.

Example:
Deuterium got after the Fizbo 1.5 test-run 50 new games.
Game 24 and Game 48 is ECO Code B11. Double game undo move 8 ... I replayed one of the games (latest game, game 48).

This is too much work, so double games is determined by eco code? and not from the
the same position out of book?

Game 4 and game 9 = B38 ... double game undo move 7 ... I do nothing.

Is this double games before move 8?

[Frank Quisinsky]

Hi,

yes, game download from my Webpage is *.pgn CB GUI produced.

Each game ended with 300 moves are replayed, not important is the result after 300 moves. Not many games ... have a look in my bigger download file (around 180 Mb). All that games are included in an extra database (I believe around 20 games I have, again ... all are replayed).

Yes, it is a lot of work to check after a test-run "double ECO code games". Need time to look in CB after such games.

To your question:
Not easy to answer ... often I replayed games the first 6 moves are the same. In average around if the first 8 moves are the same ... important is the opening system.

Example:
Engine don't understand the opening ...
Why should go from 50 games two of such games in my database? So I am thinking ...

Yes, first 8 moves are the same ... so I replayed the game.

For you ...
In average ... 2050 new games vs. 41 opponents ... around 10-15 of such games I have to pick out and games are replayed. I think it's a good stat with my opening book concept.

Best
Frank

[Frank Quisinsky]

Hi Ferdinand,

I will upload the version in the late evening on my webpage with a news about it.

Tested a bit ... look good!

Best
Frank

[Ferdy]

Hi,
Yes, first 8 moves are the same ... so I replayed the game.

PGN-extract can extract games that have more than 1 game of the same position after a given move or plies.

In your case extract games when after 8 moves, if there are more than 1 game in the pgn file with the same position. If you are interested on this feature, I will let the tool extract those games.

[Frank Quisinsky]

Hi Ferdinand,

of course I have interest.

Example:
2050 games one engine played vs. 41 opponents ...

Tool have to look in 41 different matches and sorted out games with the same ECO code and same first 6 moves. Such games should be saved in another database too.

Deuterium - Fizbo ... 2x C01 and 2x C03.

So I can checked this games ... normaly ... if the result is an other I do nothing. If the result is the same I am looking in which of the game the eval is in the near of 0,00 ... after three moves after opening book. I hold the game with more balanced eval and the other game are replayed ... again if the result is the same. If I have 2s C01 and the result is 1:0 and 0:1 I do nothing ... if not to many of the first moves are the same.

In average around 15 of such games I found in the database after a test-run ended.

Important for me is only ...
That the tool saved the games in an new database, not more. I can look in the games ...

But this isn't not so important, the other things you do with Universal PGN / EPD Tool Version 2.0 are much more important.

I wrote News 202 and added your new tool now on my download area.

Again, have many thanks!

Best
Frank

[Ferdy]

All right, I will not create a command to extract such games.

I Have a comment, from your site,

If you have interest to try out it in self work you have to install:
Python & PGN-Extract

Installation of python is not required because the tool is an exe file. I have converted the python source to exe file so that people can just start the exe file immediately without installing python.
A copy of pgn-extract-17-21.exe file is required.

For developers or those interested in source code, its source in python is also provided.

Change Philippine to Philippines .

[Frank Quisinsky]

Hi Ferdinand,

I changed that directly.
On Weekend I am again not at home.

I made a short break for 5 days with my Rating List. All is up to date on my site. Next week I will check the complete database with your tool.

Again have many thanks!
This helps me a lot in the future ...

Best
Frank

[Frank Quisinsky]

Hello Ferdinand,

working with your tool since a while.

I checked all 120.800 games and found 1.050 games I have to check again. At the Moment 400 games are checked and 180 bad lines was found. From this 180 bad lines around 50% are decativated (found it with the latest test-runs, deactivated this lines directly).

With other words ...
After 400 of 1050 games I found around 100 "new" bad lines with your tool. Really a great work you do for us and people which are working on opening books. This 100 games I must overview ... because I checked each of the FCP games ... but not in detail ... most of the time very light-minded / superficial.

Thanks again!!
Thats helps me really a lot.

Best
Frank

PS:
At the moment ...
After 100 new FCP games around 1.2 games with bad lines produced. My opening stats will be better and better. I am sure if I am ready with the 1050 game database it will be perhaps 0.7 - 0.8 games per 100.

Biggest problem are engines with give a Bonus after opening book moves (ICE for an example). Eval after book moves is often over 1.00 ... in reality the line is fully OK. But very often ICE have right with his own eval. ICE helps me a lot to find bad lines ... more as each other engine.

[Ferdy]

Biggest problem are engines with give a Bonus after opening book moves (ICE for an example). Eval after book moves is often over 1.00 ... in reality the line is fully OK. But very often ICE have right with his own eval. ICE helps me a lot to find bad lines ... more as each other engine.

I try to create an engine profile (move score vs performance) for iCE. So this would give you an idea given its move eval, you can have an expected performance score based from its white games from your database. This is the plot when iCE is white and the position is in the middle game. Middle game means, both side's pieces (non-pawn) >= 2*Q + 4*R.
When its eval is from 0 to 29 cp, expect it to score at 47%. When its move eval is from 30 to 59, expect it to score at 50%, if 60 to 89 cp, it would be 55%.



Later I will try other engines.