Author: Drexel,Michael
Date: 00:10:16 03/15/04
Go up one level in this thread
On March 14, 2004 at 19:32:23, Mike S. wrote: >I suggest that when new engines join the SSDF ratings, at least 50% of the >opponents should be from the top-10. When the Elo system underestimates the >chances of the weaker player to draw - as statistics experts have explained - >then choosing weak(er) opponents only for the important entry rating of an >engine known to be strong, must result in an underestimated IOW too low rating. > >(In the current Ruffian 2 results, the best opponent is Shredder 7/K6-450 which >is ranking #16.) > >Of course there's a reversed logic variant of that: When an engine which is in >fact mediocre only, would be tested against top opposition for the first rating. >Then, that engine would benefit from the effect and would be overrated. I'm not >aware of an example for this. > >Regards, >M.Scheidl Look at the opponents of Shredder 7.04: 1 Shredder 7.04 UCI 256MB Athlon 1200 MHz, 2808 Fritz 8 A1200 22-18 DpJuni8 A1200 28-22 Hiarcs9 A1200 22.5-19.5 Fritz 7 A1200 24-16 CT 15.0 A1200 30-10 CM9000 A1200 24-17 Junior7 A1200 24.5-17.5 Hiarcs8 A1200 29.5-14.5 Ruffia1 A1200 28-14 Dp Sjeng 1.5 12-5 Junior7 K6450 33-13 CT14 CB K6450 32-13 Ch.Ti12 K6450 35-7 Junior6 K6450 31.5-8.5 Shred5 K6-450 32-6 Frit532 K6450 32.5-7.5 Hiar732 K6450 31-5 SOS K6-2 450 43-5 Goliath K6450 39.5-4.5 Fritz532 P200 37.5-4.5 Craf18.12 K62 43.5-3.5 It played way more games against the "weak" opponents. Here is what Chessmaster 9000 did with "weak" opponents: Fritz6 K6-450 30-12 Junior6 K6450 25.5-14.5 Shred5 K6-450 34.5-7.5 Nimz732 K6450 31-9 Both simply smashed most "weak" engine/hardware combinations. This is necessary to obtain a high SSDF rating. So the strength of the 4 opponents was pefectly OK. Ruffian just didn't play enough opponents. Michael
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.