My new openings books

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

benstoker
Posts: 342
Joined: Tue Jan 19, 2010 2:05 am

Making books for engines; was Re: My new openings books

Post by benstoker »

How do the book makers make their books? What is the general methodology? By hand?!

I would think one could take a large collection of GM games; have your favorite engine then play itself, taking the first 10 moves or so from the GM games one by one, recording the game results; let it play for several days to generate lots of game data; then, let polyglot make a book from the output PGN file, which does its thing by putting high percentages on the most winning opening lines.

Is this the method the book makers use? In any event, is this a sound method?
Rubinus wrote:Today avalaible:

SpikeTurnPH-Book.rar cca 2.9MB rar, 4.2MB unpack. New version, 9.4.2007.
User avatar
Rubinus
Posts: 1161
Joined: Thu Jan 18, 2007 4:05 pm
Location: Prague
Full name: Pavel Háse

Re: Making books for engines; was Re: My new openings books

Post by Rubinus »

benstoker wrote: ...
I would think one could take a large collection of GM games; have your favorite engine then play itself, taking the first 10 moves or so from the GM games one by one, recording the game results; let it play for several days to generate lots of game data; then, let polyglot make a book from the output PGN file, which does its thing by putting high percentages on the most winning opening lines.
Is this the method the book makers use? In any event, is this a sound method?
Around yes. Polyglot system here:
a) GM games, cca last 2 years, cca import 10-15moves
b) correspondence a comp games (length time control and good hw), if good results, import to cca 35 moves
c) merging a-book + b-book

later
d) engine games, for tunning, import cca 10 moves.
e) merging d-book + ab-book
benstoker
Posts: 342
Joined: Tue Jan 19, 2010 2:05 am

Re: Making books for engines; was Re: My new openings books

Post by benstoker »

Rubinus wrote:
benstoker wrote: ...
I would think one could take a large collection of GM games; have your favorite engine then play itself, taking the first 10 moves or so from the GM games one by one, recording the game results; let it play for several days to generate lots of game data; then, let polyglot make a book from the output PGN file, which does its thing by putting high percentages on the most winning opening lines.
Is this the method the book makers use? In any event, is this a sound method?
Around yes. Polyglot system here:
a) GM games, cca last 2 years, cca import 10-15moves
b) correspondence a comp games (length time control and good hw), if good results, import to cca 35 moves

thx
c) merging a-book + b-book

later
d) engine games, for tunning, import cca 10 moves.
e) merging d-book + ab-book