Dragon 2 by Komodo - chess960 testing completed, +14 Elo.

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
Modern Times
Posts: 2895
Joined: Thu Jun 07, 2012 9:02 pm

Dragon 2 by Komodo - chess960 testing completed, +14 Elo.

Post by Modern Times » Wed May 05, 2021 3:07 pm

Not TCEC for sure, instead 1CPU blitz as usual. AVX2 exe used.

+14 Elo on these lists that use bayeselo, or if you prefer Ordo which expands the ratings comparatively, then it is +18 I think with that. All subject to statistical error margins.

http://ccrl.chessdom.com/ccrl/404FRC/

http://ccrl.chessdom.com/ccrl/404FRC/cg ... librate=no
.

Opinions expressed here are my own, and not necessarily those of the CCRL Group.

lkaufman
Posts: 4877
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Dragon 2 by Komodo - chess960 testing completed, +14 Elo.

Post by lkaufman » Thu May 06, 2021 12:58 am

Modern Times wrote:
Wed May 05, 2021 3:07 pm
Not TCEC for sure, instead 1CPU blitz as usual. AVX2 exe used.

+14 Elo on these lists that use bayeselo, or if you prefer Ordo which expands the ratings comparatively, then it is +18 I think with that. All subject to statistical error margins.

http://ccrl.chessdom.com/ccrl/404FRC/

http://ccrl.chessdom.com/ccrl/404FRC/cg ... librate=no
Nearly tied with SF13, can't complain! We didn't change anything that would be specific to chess960 in between the initial Dragon release and Dragon 2, so that would imply a similar gain of 14 elo CCRL or 18 elo CEGT in the standard chess blitz ratings, which would exceed our estimated ten elo. But margins of error are large, so I'm not predicting that this will be the case.
Komodo rules!

Post Reply