Author: James Robertson
Date: 17:45:00 03/18/99
Go up one level in this thread
On March 18, 1999 at 13:31:26, KarinsDad wrote: >On March 18, 1999 at 11:47:35, James Robertson wrote: > >>On March 18, 1999 at 10:32:03, Pat King wrote: >> >>>On March 18, 1999 at 09:21:01, Albert Silver wrote: >>> >>>>On March 18, 1999 at 05:30:33, Charles Unruh wrote: >>>> >>>>> If Kasparov beat GM Ashley 6 games in a row what would it say about his chess? >>>>> If Yermo beat H7 6 times in a row what would it say about it's chess? >>>>> H7 needs to play a 2500 GM. Why? Because we(most of us speculate the comps >>>>>are around 2500-2550 no one(few) expect that progs are 2600+ >>>> >>>>If the program does indeed play at 2500 then it should hardly lose 6-0 but more >>>>like 4-2. >>> >>>It's expected score may be 4-2, but a 6-0 loss would not be significant. Just as >>>if I flip a coin 6 times, and get heads 4 or 5 times instead of 3 times, I >>>cannot draw a conclusion about the "fairness" of the coin. The sample size is >>>just too small. >> >>I agree that the sample space is small, but it is not _that_ small. The odds of >>getting 6-0 for flipping a coin is 2 ^ 6, or 1 in 64. But in chess we have >>draws; 3 ^ 6 is 1 in 729, or 0.13%. And at the GM level I would wager that draws >>occur more than 1/3 of the time between evenly rating players. >> >>Even though the ratings are uneven, they change the odds only tiny amount; For >>instance, Yermolinsky should win 0.679% of the points, since he is 130 points >>higher (assuming Hiarcs 7 is 2500). That increases the odds he will win all 6 >>games by less than 1/2 of a percent. >> >>James >> > >Hi James, > >Actually, I would say that the chances of Yermo going 6-0 is much smaller than >your calculations indicate. I think you misunderstand my motivation for the calculations. First of all, I made no adjustment for draw percentage, which is certainly not only 30%. Right from the start, my final number of "about 0.4%" can be thrown out the window. Then of course there are the human factors you mention below, and a million other things that alter the odds. I used the very rough estimate simply to disprove the idea that a 6-0 result means nothing. > >The reason is that if he is leading the match 3-0 (a prerequisite for winning it >6-0), he may not have the motivation to force wins in the last 3 games. He could >lose up to 2 of them and still win the match. So, unless there is some major >financial or personal (some people are really competitive) motivation for him to >win the last 3 games, he may quickly ask or try for a draw in at least one of >them. Of course... James > >So, at some point, there would be a diminishing return on working hard for wins >when draws (or even loses) would suffice. > >Also, due to the "apparent" playing strength of Hiarcs7, I doubt Yermo will even >go 3-0 in order to make that decision in the first place. For example, if Yermo >is winning 1-0 or 2-0, there is a good chance he would try for a draw in game 2 >or 3 if the position looked drawish as opposed to trying to force a win which >may not be there. In a position that he does not feel he is winning (i.e. it is >equal or he is behind), he will probably not play aggressively for a win, but he >may play aggressively for a draw. This is the real reason that most GMs have >such a high draw percentage. They often play conservatively (unlike Kasparov) >when the position looks equal in order to not lose drawn games (and of course, >the mistakes made by their opponents at that level are very small and harder to >detect). > >KarinsDad :) > >> >>> >>>A more even match with an expected outcome of 3-3 MIGHT let one say something >>>definitive in the event of a 6-0 rout.
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.