Currently there is great demand for many computer chess enthusiasts from Ctg opening books. Format (fritz) and how to create 'the opening books and also how to improve their fighting against other chess engines and their respective opening books, I'll do a tutorial on this subject, to try to clarify some doubts and, on the other hand , to receive comments on this tutorial. The aim is to spread my knowledge on this subject to share with all users, enthusiasts of chesscomputer.To begin we will make two distinctions of opening books. Blog In this file we will focus on opening books for chess engines on play time ranging from 3 minutes and 16 minutes.Once we know we have to create an opening book for this pace of play, we must choose what chess engine we are using this book and therefore we have a database as current as possible on the engine chess complement the book. In this case, I'll use Rybka as a reference in creating the opening book, which in turn will be published once you have completed the tutorial. The database will be the point since we started working with the opening book to blitz. As we will use a chess engine rybka and depart from a database consisting of 85% of games (1-0 and 0-1 of the Rybka chess engine) 10% of chess games last year, in our case 2009 (also 1-0 and 0-1) and finally, 5% of games GMI year 270+ Elo Fide 2009 (1-0 and 0-1).It is necessary to verify that there are no duplicate sets in the three databases that form the opening book.

As we've said before we will create three databases with the following percentages: 85% delo games chess engine we use for the opening book. 10% of correspondence chess games, and 5% of GMI +2700 games. There will be at these games database tables with the result. web sites recommended to choose the games are the following websites: Frayer, Sedat, 40/40 CCRL, husvankempen and Pionnerchess. For correspondence chess ICCF. Regarding the GMI database there are many options, megadatabases, bigdatabases, etc. While we must ensure that this database is real and not renowned players. In our case we will use the following players.

Now we have the following database of GMI +2700 ELO Fide.

We turn to rybka database. From our database manager go and get all Rybka games that we downloaded from the websites that I have discussed and we will filter the games in response to the stated above.

and finally the database ICCF

By now we have the three bases properly filtered and proceed to copy them to a database with the name of the book we are going to build, which in this case we will assign the name "our book"

Now we have our database ready to create the opening book. Once you open the database, go to the option to create book of openings and names it.

and we will work with this book, our book, which explain how to optimize it for time control, blitz,  that we have agreed to this opening book. file in the next blog will relate in detail the steps..

The opening book has control + - during the first 50 moves, therefore, the opening has an impact on the rest of the game, and the vast majority of wins is a good transition from it to the middle game. Hence, in the 0-100 propose partdias blitz for our book, no matter the hardware as a - cpu power only thing we have to have more patience in the analysis we make to improve our opening book. Now you've created the opening book and start playing with it. While going to start learning to take a book from our database of chess engine we are going to use, in this case Rybka. After our analysis with the tool "infinite analysis" we pay particular attention to games  that has resulted draws, try searching for a line break those draws or if we find it interesting those draws if we refer to the defense as Black. In my opinion, learning the book of the databases is an issue that can be discussed here, with your comments, or in the forums. But particularly, I am in favor of playing with the book without going to improve the learning and opening book with games played (open learning) to be reduced after learning our lines and go setting manually. I think we'll get more satsifacción, since it is a more laborious, but you'll get better results. However, we will at first learning the games played by rybka won and lost.

Now we will import the games played by rybka opening book with the database we have previously developed games Rybka.

And we will add learning to the opening book to play with our book .. When we start to play our opening book leave open learning, and we'll reduce this value until it was almost entirely closed when we consider moving to manually scan the opening book ..

Now we have the book learning to start using this book in our test or chess servers. Once we have the finished book, I recommend we have sealed openings green they should undertake. In this case, we start in blocks of 500 items. that is, now we will mark in red all moves e4 except that the green will leave to make an exit test with an opening open, and hence also begin to develop our favorite defenses against this movement.

Finally I will publish our book on my website and I offer the possibility that together we can develop and submit to some tournaments (Sedat, for example, as a team). Those interested in forming a team can contact me through my website.

Jordi Ruiz