Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rebel-Anand prediction stats

Author: blass uri

Date: 02:01:52 07/07/98

Go up one level in this thread


On July 07, 1998 at 02:25:43, Ed Schröder wrote:

>>Posted by James B. Shearer on July 06, 1998 at 18:28:27:
>
>>      The Rebel-Anand match prediction stats on the Rebel web site haven't been
>>updated in a long time.  What's the story?
>>                         James B. Shearer
>
>The story is as the page says, next update after 500 predictions. I have
>498 now so the update will come soon. Here is the 498 story....
>
>-----------------------------------------------
>Rebel vs Anand  0 - 8       3     0.6%
>Rebel vs Anand  ½ - 7½      0     0.0%
>Rebel vs Anand  1 - 7       6     1.2%
>Rebel vs Anand 1½ - 6½     20     4.0%
>Rebel vs Anand  2 - 6      37     7.4%
>Rebel vs Anand 2½ - 5½     79    15.9%
>Rebel vs Anand  3 - 5     104    20.9%
>Rebel vs Anand 3½ - 4½     66    13.3%
>Rebel vs Anand  4 - 4      41     8.2%
>Rebel vs Anand 4½ - 3½     49     9.8%
>Rebel vs Anand  5 - 3      51    10.2%
>Rebel vs Anand 5½ - 2½     20     4.0%
>Rebel vs Anand  6 - 2      11     2.2%
>Rebel vs Anand 6½ - 1½      3     0.6%
>Rebel vs Anand  7 - 1       1     0.2%
>Rebel vs Anand 7½ - ½       2     0.4%
>Rebel vs Anand  8 - 0       1     0.0%
1 prediction 8-0 for rebel is 0.2% and not 0.0%
>
>Number of predictions 498
number of predictions 494
>---------------------------------------------
>
>The current input of 498 is a bit disappointing on the other hand the
>contest didn't get much attention on my home page. I will put it on
>top of the news section to get more attention.

I believe 1 of the reasons for the disappointing input of 498
is that humans prefer to see what other think before they do their prediction
and for that reason delay their prediction to the last day.

it is logical because the best prediction is not the result you think
has the highest probability but the result you can earn in evarage most of it
and if someone think the probability for 5:3 for rebel is 30% and
the probability for 5.5:2.5 for Rebel is 20% and the number of humans who guess
5:3 is twice the number of humans who guess 5.5:2.5 then it is better to
guess 5.5:2.5.

I think it is better to predict a win for rebel because the value
of rebel10 will be higher if rebel10 wins.

Uri
>
>The direct URL is:
>
>http://www.rebel.nl/contest.htm
>
>The expiration date is July 20 (one day before the match starts)
>
>- Ed -



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.