Author: Dann Corbit
Date: 18:12:58 01/29/99
Go up one level in this thread
Consider a 300 point difference in a 20 game match between a 2800 SGM and a 2500 GM. There are 20 points at stake. The win expectancy is .15, so .15*20 = 3 points gathered by the 2500 GM expected. So perhaps, one win and four draws would be expected in a 20 game match, roughly speaking. Now with a 500 point difference, the expectancy is now .05, so .05*20 = 1 point, perhaps two draws against a 2200 player. 18 wins and two draws or 19 wins and a loss is pretty much complete domination in a twenty game match. Now, if you go to your chess database and pick some very prolific super GM like Spasski, you will find that every so often he does lose to someone you have never heard of. But now, let's suppose that the difference was 1000 points (As if a 2800 player playing an 1800 player match of 20 games would be held). The expectancy is 0.00315231, so in 20 games this gives .06 points. In other words, if they played 200 games, the 1800 player might somehow get a draw. But in a 20 game match, we would be very astonished if the 1800 player even drew a game. And in a 200 game match we would be very surprised if the 1800 player won a game. Now It sounds pretty absurd, doesn't it -- in 400 games the 1800 player would win one, on average (or get 2 draws)? But that is because we may be thinking that "1800 is 1800" when in fact that does not need to be the case. Tal, Karpov, etc. were all 1800 players *at some time* and rising rapidly. So now you see *how* these 1800 ELO players will win against the big fish. A lifetime 1800 player is basically *never* going to win without a stupendous blunder, and even with the blunder it is not assured. So why don't we have lots of wins over GM's by 1800 players? Because once you beat up a few GM's you don't stay at 1800 any more. Certainly the bulk of the wins by lower rated players against those much higher will be by players on the way up against players on the way down.
This page took 0.01 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.