Book project: A00-E99 Libre stats from Houdini / Shredder

[Frank Quisinsky]

[/img]

Eval from Houdini / Shredder with 0,00 in green (should be delete for an opening book). But let us wait what the other 8 engines give out.

Eval to high for opening book moves in red ...
Pos ... = Position Number (10 engines analyzed 26146 Positions)
C = Houdini
D = Shredder

At the moment I am working on it ...
I added the right ECO Code of each of the position ... A00-A01 is ready. This need a while.

Loud thinking with picture, better with Libre Office!

Math experts can do much more things I can do ... think so! If I have 10 engines in that table ... I am sure, I will be happy.

Maybe clone stats are possible too (to the topic A00-E99).

Best
Frank

PS: If I am ready with A00-E99 (Information in A) I will public the first version. Maybe math experts will be help a bit?

[Frank Quisinsky]

Average of 26.146 opening evals from Houdini 5.01 Pro x64 = 0.24
Average of 26.146 opening evals from Shredder 13 x64 = 0.14

Enough ... will be later available if ready!

[Frank Quisinsky]

and Libre Office 5.3.2.2 format now available.
I am sure end of the project the statistics to the opening keys will be much better. But for the moment fully OK.

At the moment 739 / 26.146 = 0,00 evals from both engines (Houdini and Shredder). Again, such positions I will be later delete if the others have the same opinion. The contempt of my opening book later will be very high!

The stats can be found in the "statistics ... after v02.1" directory of my big download file (61.859Mb) to the project.

Menue system "New Opening Book" of my website.

Best
Frank

[Frank Quisinsky]

Hi there,

now much better ...
Can be found in an additional download for the future in my detail page to that project.

Have fun!

Best
Frank

You can be sure that more interesting stats will be follow!

[Ferdy]

I am trying to collect all positions analyzed per engine like,

Code: Select all

analyzed_pos_by_stockfish.epd
analyzed_pos_by_komodo.epd
analyzed_pos_by_houdini.epd
[others]

Perhaps you can offer this as a separate download.

From your site.

[Frank Quisinsky]

Hi Ferdinand,

I will not have the older Stockfish and Komodo analysis in the current file. Both engines analyzed other database (not the important 26.146 database). Furthermore, with other Hardware and the half of time.

With other words:
Komodo and Stockfish have to analyszed the same 26.146 positions again. I have in my brain to start this engines at last. At the moment I have two engines and eight will be follow. Fire as third is still running, Wasp 2.00 will be the next.

So I have end of the project with Komodo and Stockfish the current versions in the list.

The older Komodo and Stockfish analysis can be found in the main file to the project (under beta-v01 and beta-v02). But again, I think not longer important.

In the current file with 14Mb I have the analyses by engines in Excel. But here I can ... if you like the epd files in a separate directory too?!

Best
Frank

[Ferdy]

Hi Ferdinand,

I will not have the older Stockfish and Komodo analysis in the current file. Both engines analyzed other database (not the important 26.146 database). Furthermore, with other Hardware and the half of time.

Ahh right, I have position without Stockfish evaluations.

I am trying to create a web app, this will be good for your site to search position and compare engine search info.





After searching certain epd.

[Frank Quisinsky]

Hi Ferdinand,

that's great.
If I am ready with my project I will create a new website for Wasp and the Opening Book project. End of Mai I can added this Information on my current site.

Better is to added not the Stockfish and Komodo Analysis with 30 seconds on an other Hardware. Again, both will analyze the positions again, Looks better for the complete Project if all engines have the same conditions and analysed the same database.

Very interesting is ... better as in my Excel files ... that the first move of the Analysis is available in your table.

I am sure, if I collected more engines that we can do a lot with the material. Each day I have new ideas here ... but at the moment I have to work on 6/7 days for my professional work and can't do many things here. Important is only that the engines analysis are still running.

Best
Frank

[Gusev]

Hi Frank,

After 1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 Nc6 6. Be2, why not 6...d5!? Your ctg considers only 6...d6 and 6...Bb4.

[Frank Quisinsky]

Hi Dmitri,

B45 is "c-b" priority after the move before. Never played in one of my rating systems. In B45 a lot of balanced lines are possible, never saw that in my stats.

6. ... d5 is very rarely. I have not many written books to that line. Looking in Sahovski and found not many things here. Safarli is playing such a move vs. Areshchenko also with move transposition later in a Lasker game. I have also different of such lines with move Transposition because I have a lot to B45 three moves after ECO Code started ... started with 5. ... Nf6.

Honestly ... B45 with 6. Be2 is more or less rarely but playable ... means to the topic ... balanced line.

Can't changed that so far because all engines should analyzed the same database. At the moment I collect lines for an update of my database start of the next year. I will added B45 with different rarely lines to my ToDo.

Be sure that today all is very complicated for myself with the lines I have in the 26.146 database because many years I reject bad lines, added new lines ... more often by hand. But B45 is a code I never spend a lot of time because I am thinking I have here enough and most important things should be included. Not a A / B priority line ... so I never saw it.

Best
Frank

For such questions I have "anxiety". Not why my work is to bad, more I know that I reject during my FCP Rating List over 7000 lines, engines have problems. I am sure that around 2000 of such lines are wrong ... means that I reject the lines. I checked the played FCP Rating List games only three moves after the book ended with the eval of engines. Many different engines in my list and a lot of available engines have problems in understanding opening systems. Often I too ... I am not a Grandmaster.

For around 1 months I made a short test ... I selected out with games three moves ECO code was formed GM and correspondence games again. But here with average 2.350 and compare the results with that what I have today in my own book project files. I am sure I have around 95% of important lines and most things missed here have really reasons (more as one engine have problems with it ... or lines go to fast to draw with 3-fold.

So later after my work is ready absolutely possible that main lines select out in my database. The final result will be not a classical opening book ... more an opening book most engines haven't no problems and can started with a balanced line the game ... only that is important if we compare Ratings with here ... a strong database for it. To have over 24.000 lines, three moves after the ECO was building is really a lot and in my opinion enough. Holes are absolutely possible ... so I have anxiety if experts checked the database. But I haven't no anxiety later ... if users are playing eng-eng games because all what is included is checked more times and fully OK.