Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: My thoughts on these positions for what it's worth.

Author: Mike S.

Date: 21:40:36 12/22/05

Go up one level in this thread


On December 22, 2005 at 23:52:45, Graham Banks wrote:

>On December 22, 2005 at 23:30:46, Heinz van Kempen wrote:

>>I tested the positions again. Highest value was +0.59 after Shredder 9
>>calculating for a few minutes. Show me another test suite or book that gives so
>>balanced lines. For Nunn positions Robert Allgeuer made some investigations sent
>>to me by email that they are not all balanced, some give a score of more than
>>70% to White or Black overall. Nommen Suite also has known weaknesses. I played
>>many thousand games with all of them and I am happy to have found a better
>>alternative now.

>I have already stated to Harry and others (as you're aware)that his small book
>and test suite are the fairest I've used to date.
>
>I've also stated that there is no book, either commercial or otherwise, that
>does not have dubious lines.
>Others may or may not agree with my thoughts on those positions posted. It would
>be good to get other opinions.
>
>I stand by my assertion that ideally, the evaluation for each side coming out of
>the opening should be no higher than 0.50
>Again it would be good to get some opinions on this also.

A agree that this would be perfect for a very good neutral book, but it's
difficult to achieve for every variation. At least I can say, in a book
basically generated from master games like mine, no doubt that it may contain
some dubious moves - but as a consequence of the statistical data, the
probabilities of such moves should be very small, or zero. For example, moves
which have been played only once in the database the book is based upon, will
get prob. 0.00% in Fritz with optimal or normal book settings.

I have made a test run with the PB12moves.ctg, having 200 examples checked by
both Fritz and Hiarcs: 130/200 have been smaller than +/- 0.50. Less than 10%
have had evals bigger than +/- 0.80 (19/200). This test was with book option
"normal."

http://www.computerschach.de/index.php?option=com_joomlaboard&Itemid=208&func=view&id=942&catid=3
(2nd posting)

I was satisfied with these results! :) Practical tests in matches were ok too,
so far.

Regards,
M.Scheidl


P.S.: If you are interested in that book, download is free from CSS Online

http://www.computerschach.de/index.php?option=com_remository&Itemid=52&func=fileinfo&id=51

English release notes:

http://www.computerschach.de/index.php?option=com_remository&Itemid=52&func=fileinfo&id=51



This page took 0.23 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.