an STS Tool retouched

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
User avatar
Matthias Gemuh
Posts: 3238
Joined: Thu Mar 09, 2006 8:10 am
Contact:

an STS Tool retouched

Post by Matthias Gemuh » Mon Dec 26, 2011 4:06 pm

an STS Tool retouched

1. The STS EPD files must be renamed to STS01.epd, STS02.epd, ... STS49.epd
2. Play a Testsuite tournament using ChessGUI.
3. Import Testsuite result file into STS_Tool.
4. Supply known Elo Ratings of at least 2 engines (see CCRL).
5. Calculate Elo estimates.
6. Generate Html.
7. If desired, delete all data by deleting STS_Tool.ini

Sample Testsuite result files are provided to allow an immediate test of the STS Tool.

http://www.chessgui.com/chess/STS_Tool.zip

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de

peter
Posts: 1754
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: an STS Tool retouched

Post by peter » Mon Dec 26, 2011 10:11 pm

Matthias Gemuh wrote:an STS Tool retouched
Matthias, that looks like a really great and grand idea to me.
Might be the missing link between positions' tests and rating lists.
Could be even the better objective measure of analyze strength of engines, that Larry Kaufman hopes for:
http://www.talkchess.com/forum/viewtopi ... 29&t=41612
Peter.

lkaufman
Posts: 3530
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: an STS Tool retouched

Post by lkaufman » Mon Dec 26, 2011 11:18 pm

I rather like the idea, but I am skeptical that it will produce significantly different results from the current rating lists. Also I don't like that it ignores the opening.

peter
Posts: 1754
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: an STS Tool retouched

Post by peter » Mon Dec 26, 2011 11:44 pm

lkaufman wrote:I rather like the idea, but I am skeptical that it will produce significantly different results from the current rating lists. Also I don't like that it ignores the opening.
Well that would be worth a trial.
If you miss the opening, simply take just one more position to the many others, which also came from games played according to human masters' opening theory, maybe yet more important and of more interest to human masters than the openings engines produce:
add the one and only very starting position to the many other positons of interest and that's it, no?
:)
Of course, you can add all the other 5 to 8 moves long opening positions, that are used now, too, if you want to be very very sure not to loose any of the Elo points you gathered in the usual way till now, still knowing these were a result of pure arbitrary selection of starting positions too, even if you played from broad short books.
The editing of these books is the selection (bias), not the random choice the GUI or the engine does later on out of the given pool.
:)
Peter.

Post Reply