Author: Christopher A. Morgan
Date: 10:54:25 01/01/02
Go up one level in this thread
Blackjack is a game that can be beaten with proper play, meaning the player has about a 1-2% advantage over the house. It is well known, however, that even with best play and this advantage over the long run, a player can have long losing streaks, some lasting months while playing 3-4 hours a day, and that just due solely to the mathmatics of statistics. To test any given strategy/betting simulation the number of blackjack hands in the simulation, to be accurate, is in the hundreds of millions, not hundreds of thousands, very easily accomplished on today's computers. If you study the game of blackjack, and get some of the available simulation programs, you will understand Chris' statement very clearly. The difference among the top chess programs falls in this 1-2% range, and without playing a substantial number of games, determining which program is better is nothing more than a guess. Chris On January 01, 2002 at 13:01:01, Christophe Theron wrote: >On January 01, 2002 at 11:05:55, Eran wrote: > >>Please see the following list that comes from the Rebel homepage below. >> >>RANKING after ROUND-12 >> Tournament: Odyssey-2001 >>Place Name Sco MBch Buch Ws >>------------------------------------------------------------- >> 1 SHREDDER5, (1) 9.5 66 78½ 7 >>2-3 REBEL-CENTURY4, (10) 8.5 66 81 7 >> GAMBIT-TIGER14.6, (2) 8.5 63 77 6 >>4-7 NIMZO8, (20) 7.0 67½ 81½ 6 >> YACE BERLIN, (24) 7.0 65½ 79 4 >> CRAFTY 18.12, (19) 7.0 61½ 73½ 6 >> GANDALF432H, (11) 7.0 61 74 5 >>8-12 PHARAON 2.5, (12) 6.5 66 77 4 >> FRITZ7A, (8) 6.5 65½ 79½ 4 >> CHESSMASTER8000 1.0.4., (6) 6.5 63½ 74½ 5 >> SHREDDER4 CHESSBITS, (5) 6.5 58 69 3 >> VIRTUAL-CHESS2, (7) 6.5 52 60½ 5 >>13-16 CHESS SYSTEM TAL2.03, (14) 6.0 56 69½ 2 >> HIARCS7.01, (4) 6.0 55½ 65½ 4 >> LITTLE-GOLIATH2000V3, (16) 6.0 55 63½ 4 >> REBEL-TIGER14.6, (3) 6.0 55 63 3 >>17-20 JUNIOR7, (9) 5.5 68½ 81½ 3 >> PATZER311B, (17) 5.5 67 80 3 >> MCHESS8, (23) 5.5 63½ 73½ 4 >> GROMIT 3.8.1, (26) 5.5 58 68 3 >>21 GENIUS6.5 CZUB-STYLE, (21) 5.0 58 68 3 >>22-23 ZARKOV4.5T, (13) 4.5 61 71½ 3 >> WCHESS2000, (15) 4.5 57½ 67½ 2 >> 24 COMET B36, (18) 4.0 56½ 67½ 3 >> 25 SOCRATES X, (22) 3.5 54 62½ 3 >> 26 EUGEN7.92, (25) 1.5 55½ 65½ 0 >> >>You see that Shredder5 is still on the first place so far and Fritz7a on the >>nineth place, how come? >> >>Eran > > > >You still have to learn a lot about statistics and margin of errors... > >Short story: there is too much randomness in a tournament like this one. You'd >better trust the SSDF list, where each program has played many many games before >it gets a rating. > >If you don't believe it, run a "virtual" tournament where you decide the outcome >of the games by flipping a coin. All participants under these conditions should >be perfectly equal, but you will see one performing extremely well and the other >one performing extremely badly. How comes? Statistics... > > > > Christophe
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.