SedatChess

[Sedat Canbaz]

UPDATE 2

2nd Stage's Leader (via Unique openings): SF-PB- My Congratulations to SkyNet!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 SF-PB 221124                   23      12     389   53.3%   86.6% 
   2 Corchess 190125                20      12     386   52.8%   88.1% 
   3 Brainlearn 30                  14      13     391   52.0%   86.2% 
   4 Kookaburra 3                    4      13     389   50.6%   85.9% 
   5 Dark SisTer v2                  4      12     387   50.5%   87.1% 
   6 Contact 10 SE                   3      13     390   50.4%   85.9% 
   7 Stockfish 190125                1      12     392   50.1%   87.0% 
   8 Artemis 6 TR                   -1      12     390   49.9%   87.4% 
   9 Beast M 010125                 -4      12     387   49.4%   88.4% 
  10 DarkSeid 5.3                   -5      12     390   49.2%   88.7% 
  11 SF-PRO 311224                 -12      14     389   48.3%   83.8% 
  12 HypnoS++ 1.01                 -15      13     386   47.8%   85.8% 
  13 JudaS++ 1.01                  -31      14     388   45.5%   83.8% 

2527 of 6552 games finished.

Greetinga

[Sedat Canbaz]

UPDATE 3

In 2nd stage, via Unique Suite:
Corchess 190125 takes the lead again.. really impressive performance so far!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Corchess 190125                21       8     916   53.0%   87.0% 
   2 Brainlearn 30                  17       8     918   52.4%   86.5% 
   3 Contact 10 SE                  13       8     916   51.9%   86.0% 
   4 SF-PB 221124                   11       8     915   51.5%   86.7% 
   5 Kookaburra 3                    7       8     916   51.0%   86.5% 
   6 Stockfish 190125                3       9     916   50.4%   85.3% 
   7 Artemis 6 TR                    2       8     914   50.3%   86.3% 
   8 DarkSeid 5.3                    0       8     917   50.1%   86.2% 
   9 Dark SisTer v2                 -4       8     915   49.4%   87.9% 
  10 Beast M 010125                 -4       8     914   49.4%   85.9% 
  11 HypnoS++ 1.01                 -11       9     915   48.4%   84.9% 
  12 SF-PRO 311224                 -20       9     916   47.1%   82.0% 
  13 JudaS++ 1.01                  -33       9     914   45.2%   81.7% 

5951 of 6552 games finished.

Btw,
There will be another new but unofficial stage e.g next via UHO openings!
And the test via UHO: the played results will be not merged in 1st + 2nd..

But I wonder really
What will be influences..since is already clear, which are stronger...
Who knows? maybe UHO is become best for 3800+ about determining
After all, time will tell... anyhow, for anyone missed, the older test
Via UHO openings, I had bad experience, for more details, results:
https://sites.google.com/site/computers ... testings-1

Greetings

[Anton101]

Based on stockfish dev 180125, there has indeed been a regression from 3 to 5 Elos, it is certainly noticeable, however, it is better to have the engines updated to SF. And work on other aspects, such as Q-Learning.

Great tour as always.

[Anton101]

On the other hand, small modifications in the code suit corchess better.

[Sedat Canbaz]

On the other hand, small modifications in the code suit corchess better.

Thanks for your valuable information Mr. Anton..
Now is more clear why Corchess is doing better...

Greetings

[Sedat Canbaz]

UPDATE 4

3rd Stage (via UHO suite): Current Leader: Contact - Congrats to E. Nemeth!

Sure it is too early for any conclusions, since we need much more games..
But I like current UHO suite version. if nothing else not so much Drawish )

On other hand, there is one true and not as joke for sure:
My created recent handicapped suites draw ratio is close to UHO draws..!!
And it seems the latest Top 3800+ engines are more trained via UHO suite..
E.g some years ago..situation (error-margin) via weaker engines was huge!

However, according to my experience, e.g:
In case of tests via critical (weak) opening suites, sure here for less error margin:
It's recommended many thousands of games per player...where via strong suites,
As usual I say over: 1000 thousand games enough..this important formula matters
Also for much weaker chess engines as well...

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Contact 10 SE                  23      34     341   53.4%   15.8% 
   2 Corchess 190125                21      35     340   53.1%   12.6% 
   3 SF-PB 221124                   16      34     341   52.3%   15.0% 
   4 Kookaburra 3                   15      35     342   52.2%   11.4% 
   5 Brainlearn 30                  10      34     342   51.5%   17.0% 
   6 Stockfish 190125                2      34     344   50.3%   12.8% 
   7 DarkSeid 5.3                   -1      34     341   49.9%   14.1% 
   8 HypnoS++ 1.01                  -2      34     340   49.7%   14.7% 
   9 Artemis 6 TR                   -7      34     340   49.0%   16.2% 
  10 Dark SisTer v2                 -9      34     342   48.7%   15.5% 
  11 Beast M 010125                -12      34     341   48.2%   14.4% 
  12 SF-PRO 311224                 -28      34     343   45.9%   16.0% 
  13 JudaS++ 1.01                  -28      34     343   45.9%   17.2% 

2220 of 6552 games finished.

Btw, just to be more clear:
For the current (ongoing) test: UHO 2022 8mvs +170-179 is used..

Greetings

[Sedat Canbaz]

UPDATE 5

3rd Stage (via UHO suite): Contact 10 SE is still leading !! but CorChess
And SF-PB are very close, one of them will win, but it's hard to guess..

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Contact 10 SE                  24      22     792   53.5%   16.5% 
   2 Corchess 190125                23      22     794   53.3%   14.2% 
   3 SF-PB 221124                   21      22     791   53.0%   15.8% 
   4 Kookaburra 3                    6      22     791   50.9%   14.0% 
   5 Brainlearn 30                   4      22     793   50.6%   15.6% 
   6 Stockfish 190125                2      23     792   50.3%   13.4% 
   7 DarkSeid 5.3                   -1      22     791   49.8%   14.4% 
   8 Dark SisTer v2                 -5      22     792   49.2%   16.2% 
   9 Artemis 6 TR                   -5      22     791   49.2%   16.6% 
  10 HypnoS++ 1.01                  -8      22     790   48.9%   15.2% 
  11 Beast M 010125                 -8      22     792   48.8%   16.5% 
  12 JudaS++ 1.01                  -26      22     790   46.3%   15.9% 
  13 SF-PRO 311224                 -27      22     793   46.2%   14.6% 

5146 of 6552 games finished.

[Sedat Canbaz]

UPDATE 6

3rd Stage (via UHO suite) is completed: Contact + SF-PB are Winners!
My congratulations to both eng authors plus to rest authors as well..

And what I told all of you before ?)
It's a mystery, if running small number of games..here that's why we see
Unpredictable performances.. even despite that SF team is spending BIG
Efforts over NNUE trainings (mainly they prefer UHO opening suite series)

But via Balsa and Unique suites: Which Top engine won ?
I still did not merged the both played stages, but I think
CorChess, since managed to win the recent tour stages!!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Contact 10 SE                  21      20    1008   53.1%   16.1% 
   2 SF-PB 221124                   21      20    1008   53.0%   15.8% 
   3 Corchess 190125                19      20    1008   52.8%   13.7% 
   4 Stockfish 190125                7      20    1008   51.0%   13.3% 
   5 Kookaburra 3                    5      20    1008   50.7%   13.9% 
   6 Brainlearn 30                   4      20    1008   50.6%   15.1% 
   7 Artemis 6 TR                   -2      20    1008   49.8%   16.4% 
   8 HypnoS++ 1.01                  -3      20    1008   49.6%   15.0% 
   9 DarkSeid 5.3                   -4      20    1008   49.5%   14.8% 
  10 Dark SisTer v2                 -6      20    1008   49.2%   16.0% 
  11 Beast M 010125                -11      20    1008   48.4%   16.5% 
  12 SF-PRO 311224                 -22      20    1008   46.9%   14.8% 
  13 JudaS++ 1.01                  -30      20    1008   45.6%   16.3% 

6552 of 6552 games finished.

And soon as possible,
I hope to share full (merged 1st + 2nd stages) standings !

Greetings

[Sedat Canbaz]

Meanwhile,
Here are new stats generated by EloStat:

General Stats by UHO suite:

Code: Select all

Games        :   6552 (finished)

White Wins   :   5543 (84.6 %)
Black Wins   :     14 ( 0.2 %)
Draws        :    995 (15.2 %)
Unfinished   :      0

White Perf.  : 92.2 %
Black Perf.  : 7.8 %

ECO A =   2674 Games (40.8 %)
ECO B =   2110 Games (32.2 %)
ECO C =    898 Games (13.7 %)
ECO D =    494 Games ( 7.5 %)
ECO E =    376 Games ( 5.7 %)

More Details (about UHO suite statistics):
Not so much changed since years... as we see,
Just only 14 (fourteen) wins are appeared by Blacks,
If I'm not wrong: 'chess' is a game for 2 (two) sides!
But FUN is perfect here, just 15.2% draws!
----------------------------------------------

General Stats by Balsa + Unique suites:

Code: Select all

[code]Games        :  13104 (finished)

White Wins   :   1247 ( 9.5 %)
Black Wins   :    388 ( 3.0 %)
Draws        :  11469 (87.5 %)
Unfinished   :      0

White Perf.  : 53.3 %
Black Perf.  : 46.7 %

ECO A =   3336 Games (25.5 %)
ECO B =   3472 Games (26.5 %)
ECO C =   4558 Games (34.8 %)
ECO D =    710 Games ( 5.4 %)
ECO E =   1028 Games ( 7.8 %)

More Details (about Balsa + Unique stats):
The overall draws are 87.5%, too much than
UHO suite's produced stats, but if nothing
Else quite normal...reason, my suites are
Very well-tuned for both sides, otherwise
I see not so fair..such as if only Whites
Will win..just in case.. I may call it as chess,
But too far from the reality..perhaps UHO is
Future but still there is a BIG room to be in
Fun as without it... at least for nowadays!

As final words, if you are going to ask Me,
UHO suite is better than Balsa or Unique?
I'd replay, yes for non-serious or fun tests:
Go for UHO suite, but for serious tours:
Undoubtedly use Balsa or Unique suites!
Not sure about you, but frankly for my side,
Quality, Strength and Time matters a lot!

And after all, why I say like that about all over?
Due to 'time' is more valuable than money! sure
Everyone is free..but in case of via weak openings:
Run correct tests, play thousands games, if not..
Waste of time and will be as mystery or lottery to
See regarding which Top chess engine is stronger!

Hope helps...and very soon overall ranking coming..

Best,
Sedat

[Sedat Canbaz]

UPDATE 6

The Champion (Balsa + Unique suites): CorChess - Congratulations to I. Ivec!
Note: Very soon all current played results will be shared at SedatChess site

And here are the overall ranking (merged both stages in one):

Code: Select all

Rank Name               Elo    +    - games score oppo. draws 
   1 Corchess 190125   3822    4    5  2016   53%  3805   89% 
   2 Contact 10 SE     3818    5    5  2016   52%  3806   88% 
   3 SF-PB 221124      3815    5    5  2016   51%  3806   88% 
   4 Kookaburra 3      3815    5    5  2016   51%  3806   87% 
   5 Brainlearn 30     3815    5    5  2016   51%  3806   88% 
   6 Stockfish 190125  3809    5    5  2016   50%  3806   88% 
   7 DarkSeid 5.3      3807    5    5  2016   50%  3807   88% 
   8 Artemis 6 TR      3806    5    5  2016   50%  3807   88% 
   9 Dark SisTer v2    3805    5    5  2016   50%  3807   89% 
  10 Beast M 010125    3801    5    5  2016   49%  3807   86% 
  11 HypnoS++ 1.01     3798    5    5  2016   49%  3807   88% 
  12 SF-PRO 311224     3790    5    4  2016   47%  3808   85% 
  13 JudaS++ 1.01      3785    4    4  2016   46%  3808   85% 

GAMES:
https://mega.nz/file/6oAkQBjR#TTKFS-5kI ... WKas8VhO14

Conditions:
2x Epyc 7B12, CuteChess, Ponder OFF, 15s+0.3s, 1 Core, 64 Hash, 4 MEN

As final conclusion,
CorChess is with clear margin ahead of the rest opponents..
And actually what does it mean? due to simply as strongest!

But what we noticed also in tour with UHO opening lines?
CotChess performed worse..if I was new in chess..I may believe..
But I'm not..so many years in testings..enough to see the reality!

And I hope also that,
All recent my tours, data, stats to be useful for our community!

Best,
Sedat