SCCT VIII - Final Standings

By Benno Hartwig Date 2011-09-01 13:59

Thanx, a very interesting tournament.
Surely good books can help significantly to win.
On the other hand with only 152 games also fortune and misfortune can be very important.

I made a gedankenexperiment, taking 77 engines of equal strength, drawn-likelyhood=0.4.
And I wanted to know: what are the mean scores of the best and the worst engines in those tounaments, caused only by accident only.
I made 1000 passes, and the mean scores of the best and the worst engines were:

Code:

  1  87,6095
  2  85,8440
  3  84,8045
  4  84,0475
  5  83,4935
  6  83,0095
  7  82,5800
  8  82,2010
  9  81,8610
 10  81,5480
...
 68  70,4370
 69  70,1330
 70  69,8075
 71  69,4035
 72  68,9915
 73  68,5150
 74  67,9205
 75  67,1855
 76  66,1875
 77  64,3695

Best and worst scores in your List are:

Code:

 01 Hitman H15a                    :  96.0/152
 02 Defender H15a                  :  94.5/152
 03 JapanTopOne H15a               :  91.0/152
...
 75 Sirena H15a                    :  61.0/152
 76 Amfetaminpro H15a              :  60.5/152
77 Jakal H15a                     :  57.0/152

We see, that the best participants in your tounament have better results, which will probably be significantly influenced by their real strength. (Distance form best to worst: 39 points)
But we see also, that the scatter caused by luckyness is very large too. (here: 23,24 points)
And we see, that it is normal and will normally be caused by accident, that the very best and very worst engines have a relative large distance to the next participants.

Benno

By Sedat Canbaz Date 2011-09-01 17:15

>Thanx, a very interesting tournament.

Not at all...its my pleasure and thanks a lot for your interesting comments

>Surely good books can help significantly to win.

Actually,the strongest (well-optimized) books,which include the latest strongest opening theory have more chances to win

>On the other hand with only 152 games also fortune and misfortune can be very important.

In my opinion, 152 games per player (total 5852 games) is quite enough number to show the real strength of the current book participants

I know very well too that more games is better and more reliable,e.g 10.000 games per player would be ideal...
For example,in case of running a such number of games,the book tour duration will continue more than 1 year or maybe 2 years
And running a such long time book competition would be meangles,due to opening theory is changing almost daily
I am quite sure too that mostly of SCCT Book Authors will be not satisfied,because almost all Authors wish their books to be updated oftenly

Plus i have no much free time to organize a such 1-2 year duration book tour

Note also that only for the current book tournament i have spent more than 20 days (mostly of my efforts are done by hand)

>And I wanted to know: what are the mean scores of the best and the worst engines in those tounaments, caused only by accident only.

Actually almost all participants are played by same engine: Houdini 1.5a x64 4c,exception the opening books

About what is meaning best and the worst,which are caused only by accident in SCCT Book Tournaments ?!:
-I strongly believe those accidents are available not only in my tournaments,in all around the world there are plenty of them

-SCCT Rules:any game which is caused by accident only:is counting as 'win',thats why we call it 'Tournament'
A little note:there is no doubt that TOP Book Makers have less accidents,they are spending many efforts to avoid those accidents

Kind Regards,
Sedat

By Benno Hartwig Date 2011-09-02 07:23

[quote="Sedat Canbaz"]In my opinion, 152 games per player (total 5852 games) is quite enough number to show the real strength of the current book participants[/quote]The spreading, which is to be expected because of accident only, is 23 points.
The influence of different strength led to a split of 39 points in your tournament, only 16 points more than pure accident-influence.
I'm afraid 152 games of each player are not enough to lead the influence of different strengths out of the rather deep swamp of accident.

Yes, I know, that you invested much work, but these engines have rather similar strength, and accident influence seems to be larger than the influence of strength variation.

You can get more convincing results only if you have larger strength differences or much more games.

Benno

By Clemens Keck Date 2011-09-02 09:21

Aber Benno

deswegen ist es doch ein Turnier und keine Rangliste

Das macht doch das ganze so spannend.

Clemens

translation:

this is one of the "spice" of a tournament. Its no ratinglist.

regards, Clemens

By Sedat Canbaz Date 2011-09-01 16:01

More Rankings/Reports by Chess Utility Programs:

*BayesElo Rating List:
http://sedat-chess.110mb.com/files/scct8_bayselo.txt
Note:SeekWanda is third with BayesElo

*ChessBase CrossTable:
http://sedat-chess.110mb.com/files/scct8_cb.htm

*Scid Standing:
http://sedat-chess.110mb.com/files/scct8_scid_rank.txt

*Scid Opening Report:
http://sedat-chess.110mb.com/files/scct8_scid.html
Note:the report includes a lot of interesting data

*EloStat Individual Statistics:
[url]http://sedat-chess.110mb.com/files/scct8_ind_stat.txt [/url]

Best Wishes,
Sedat