Hiarcs 12 rating after 1,000 games (FRC)

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Dann Corbit, Harvey Williamson

Spock

Hiarcs 12 rating after 1,000 games (FRC)

Post by Spock »

CCRL FRC testing of 1,000 Hiarcs 12 games has been completed. Just 48 hrs after release, the only rating list with a large number of games and a solid rating. Being an FRC fan, any new Shredder GUI FRC capable engine that comes out gets my immediate attention :)

Firstly, on the FRC list, the pure list is the one I strongly recommend people look at. With so many different versions of a small number of engines, this list is likely to give a more reliable rating. It is a separate database with only the strongest versions of each engine.

http://www.computerchess.org.uk/ccrl/40 ... _pure.html


Hiarcs 12 rating on the pure list is 2915. It replaces Hiarcs 11.1 on the pure list which previously had a rating of 2875. So Hiarcs 12 is approx 40 ELO stronger under these conditions. It is worth noting that on this list, Hiarcs 11.2 was some 20 ELO weaker than Hiarcs 11.1

Hiarcs 12 is the third strongest publicly available FRC engine behind Shredder 11 and Naum 3. In terms of the bayeselo "likelihood of superiority" calculation, it is 100% certain that Shredder 11 is stronger than Hiarcs 12, and almost 90% certain that Naum 3 x64 is stronger than Hiarcs 12. These results don't necessarily compare to normal chess - firstly FRC is different, plus there is a somewhat limited range of engines here with a wide ELO range. But results from CEGT so far suggest that our results in normal blitz chess will be similar

Being blitz, the default Sharpen PV OFF is used.

The private Rybka 2.3.2 beta FRC version is of course still number 1. You'll see that on the main page.