Game opening checker

[Ferdy]

Plot for ice and komodo based from your game base on white pgn for both.

[Frank Quisinsky]

Indeed interesting.
Today I try to create some stats too ...

1. I am ready with the opening book corrections!

database with 1050 critical opening lines I found with your tool (much more but only this 1050 lines are to check).

168 lines deactivcated
244 lines found in the past and deactivated with "F" before I am looking again.
And all the other lines included in the 1050 game database can be play and should be OK.

Games with clearly book loses = in my opinion around 30


Maybe later I will replay all this games.
At the moment I have no interest to install the old engine configurations on one of my systems again. I think with 30 games I can life.

2. Chess Game Analyzer 3.4 is now on my webpage.

Best
Frank

[Ferdy]

If you can export last position of your book, by fen, epd or pgn, then it is easy to see which bad opening lines are in the book by analyzing the fen or epd or last position of every game in the pgn.

[Frank Quisinsky]

Hi Ferdinand,

the only way is *.dat (Text Format).
If you like I can do it and add the link with my book ín *.dat format.

Best
Frank

[Ferdy]

Hi Ferdinand,

the only way is *.dat (Text Format).
If you like I can do it and add the link with my book ín *.dat format.

Best
Frank

Let me have that file I will try to check what is in that file.

[Frank Quisinsky]

Hi Ferdinand,

If you have interest to have a look ...
Please download my FCP Live Book 20151008 (5Mb).
*.dat files included!
http://www.amateurschach.de/main/_download.htm

"Main Opening Book"
- 295.293 moves included (FCP Live Book)
= 49.994 lines!
= 293 lines deactivated with "E ... fast draw games"
= 2.971 lines deactivated with "F ... bad opening lines"
= 316 lines deactivated with "G ... unclear"

"Small Opening Book, should be good enough for eng-eng matches"
- 89.492 moves included (FCP Opening Book 1.84), moves with priorities only included!
= 38.169 lines!

Before you find a way to create a file I can analyze all end positions ...
All the "F" and "E" lines must be select out.

But I think it's not possible to do that ...
Thinking a long time about it ...

If the book is clear, I must not checked each of the played FCP games. All the priorities are set by Hand and many lines are set by Hand too. My work over the latest years, I set around 90.000 priorities by hand.

Best
Frank

My small opening book is only ...
If I say to Shredder GUI ...
"Weed Opening Book" ... means from my "Main Opening Book".
With the results that only moves with priorities are included. Most important lines are in it. Of course, under lines I not checked aren't included ... if you compare it with "FCP Main Book".

[Frank Quisinsky]

Hi Ferdinand,

have a look in the other message.
All information you need I think ...

Best
Frank

[Ferdy]

I got the file Frank. I will study this format, and try to convert it to pgn.

[Frank Quisinsky]

Hi Ferdinand,

shortly ... in additional!

"Main Opening Book"
= 49.994 lines!

"Small Opening Book"
= 38.169 lines!

Different = 11.825 lines!
I am very proud for this results after around 2 1/2 years hard work.

That 11.825 lines I never noticed for the following reasons:

1. Lines are good, produced no problems!
2. Lines very rarely played, so I can't noticed if bad lines are in it if I checked the games from an current test-run by hand.

Means it's possible that from this 11.825 lines bad lines in it. That is what I do daily. I checked the new games. If I noticed with eval of engines a bad line I am looking in my book and set "f" and overwork directly the priorities.

I believe from this 11.825 should around 500 - 750 bad lines in it (most C30-C39 other critical openings) ... if I am looking in my own stat files to my opening book. At the moment 1.2 games from 100 played games produced bad lines. And the bad lines I have in my book are in most of cases very rarely on the board because ... lines have no priorities!

---

If you able to create from the *.dat" file a file with end positions (EPD / PGN should be optimal) from all 49.994 lines I can do the following things ...

50.000 lines with 1-minute 1-core engine analyzes = 50.000 minutes : 60 : 24 = 34,722 days

With a 2-core engine analyzes on 14 cores = 7 different engines can search after bad lines ...

Best engines for it:

1. Komodo
2. Stockfish
3. Spark
4. Junior
5. Hakkapeliitta with 1 Core
6. ICE
7. NirvanaChess

For an example!

Time:
34,722 days : 7 = 4.960 days!

I can find out with "Notepad" text editor all the bad lines from this 7 engine analyzes. I copy all what 7 engines found in a new database, search for doubles and can deactivated all the bad lines in one step by Hand in my book.

And the book is clean / ready and I must not checkend new games for bad lines again and again. With the final result that around 46.000 balanced positions can be play with the priorties GM-theory have ...

In my opinion ...
This is the perfect opening book for testing engines, all 500 ECO codes are played and in around 20-35% engines find out lines included in the book (no played by GMs after move 6-10). That's really interesting because that is the main deal my book have ... To 25-35% opening theory optimated by best engines the World have ... and engines have a higher playing Level GMs have.

Best
Frank

[Ferdy]

It is not easy to convert the .dat file to pgn, but it is possible.

Another way of collecting games from that book is by playing games until the book moves or after 1 move out of book. This can be done by playing engines and set the adjudication aggressively like win after 1 cp advantage or draw after 300cp or less eval, if the gui supports such options. The result of such games are not important but only the games.

If the gui does not support such adjudication options, then I will make an engine that will resign after book moves. In this way more games will be generated.