Dragon 2.6 elo so far

[criko]

so far 18 elo+

Code: Select all

Ryzen Threadripper PRO 3995WX
cutechess cli
8moves_v3.pgn
60s + 1s
1 thread for all engines
256 MB Hash
60 concurrent games
Ordo-calculation is fixed to Stockfish 14.1 = 3780 Elo

------------------------------------------------------------------------------------------


Games        : 77600 (finished)

White Wins   : 24136 (31.1 %)
Black Wins   : 17272 (22.3 %)
Draws        : 36192 (46.6 %)
Unfinished   : 28

White Score  : 54.4 %
Black Score  : 45.6 %



     Program              Elo    +    -   Games   Score   Av.Op.  Draws

   1 Zeus 21.0          : 3792    6    6  9000    73.5 %   3564   51.0 %
   2 Stockfish 14.1     : 3780    6    6  9000    72.2 %   3564   53.6 %
   3 SugaR AI 2.50      : 3779    6    6  6500    64.9 %   3661   67.1 %
   4 BrainLearn 15.1    : 3762    6    6  9000    70.3 %   3565   55.6 %
   5 ShashChess 20.1    : 3754    6    6  9000    69.4 %   3566   56.1 %
   6 Dragon 2.6         : 3743    8    8  4100    55.9 %   3680   71.7 %
   7 Stockfish 13       : 3735    6    6  9000    67.2 %   3567   56.0 %
   8 Dragon 2.5.1       : 3725    6    6  8500    67.2 %   3557   52.6 %
   9 Fire 8.NN.MC.3     : 3610    6    6  8100    55.1 %   3553   54.4 %
  10 Revenge 2.0        : 3571    6    6  8100    48.0 %   3574   49.1 %
  11 Slow Chess 2.8     : 3569    6    6  8600    48.7 %   3568   51.6 %
  12 Berserk 8 NN       : 3565    6    6  8600    48.1 %   3569   49.2 %
  13 Koivisto 7.9       : 3562    6    6  8600    47.9 %   3569   49.9 %
  14 Ethereal 13.25     : 3555    6    6  8600    47.0 %   3569   48.2 %
  15 Revenge 1.0        : 3483    6    6  8100    38.9 %   3574   41.1 %
  16 Wasp 5.00          : 3285    7    7  8100    20.8 %   3573   27.2 %
  17 Fritz 18           : 3280    7    7  8100    20.4 %   3573   25.2 %
  18 Chiron 5           : 3226    8    8  8100    16.3 %   3577   21.0 %
  19 Critter v1.6a      : 3164    9    9  8100    12.1 %   3580   16.8 %
Individual statistics:



6 Dragon 2.6       : 3743 4100 (+822,=2938,-340), 55.9 %

Revenge 2.0        : 100 (+ 51,= 48,-  1), 75.0 %
SugaR AI 2.50      : 500 (+  9,=427,- 64), 44.5 %
ShashChess 20.1    : 500 (+ 18,=437,- 45), 47.3 %
Slow Chess 2.8     : 100 (+ 46,= 54,-  0), 73.0 %
Ethereal 13.25     : 100 (+ 46,= 54,-  0), 73.0 %
Chiron 5           : 100 (+ 90,=  9,-  1), 94.5 %
Koivisto 7.9       : 100 (+ 52,= 47,-  1), 75.5 %
Critter v1.6a      : 100 (+ 94,=  6,-  0), 97.0 %
Fire 8.NN.MC.3     : 100 (+ 38,= 61,-  1), 68.5 %
Wasp 5.00          : 100 (+ 86,= 14,-  0), 93.0 %
BrainLearn 15.1    : 500 (+ 23,=427,- 50), 47.3 %
Stockfish 13       : 500 (+ 40,=423,- 37), 50.3 %
Stockfish 14.1     : 500 (+ 11,=426,- 63), 44.8 %
Zeus 21.0          : 500 (+ 12,=412,- 76), 43.6 %
Fritz 18           : 100 (+ 82,= 18,-  0), 91.0 %
Revenge 1.0        : 100 (+ 67,= 33,-  0), 83.5 %
Berserk 8 NN       : 100 (+ 57,= 42,-  1), 78.0 %

[Jouni]

Same version has 10,5 - 19,5 against SF dev in CCC?

[criko]

maybe

[criko]

after 6740 still +18 ahead of Dragon2.5.1

Code: Select all

60s + 1s
1 thread for all engines
256 MB Hash
60 concurrent games
Ordo-calculation is fixed to Stockfish 14.1 = 3780 Elo

------------------------------------------------------------------------------------------


Games        : 80240 (finished)

White Wins   : 25110 (31.3 %)
Black Wins   : 17993 (22.4 %)
Draws        : 37137 (46.3 %)
Unfinished   : 28

White Score  : 54.4 %
Black Score  : 45.6 %



     Program              Elo    +    -   Games   Score   Av.Op.  Draws

   1 Zeus 21.0          : 3792    6    6  9000    73.5 %   3564   51.0 %
   2 Stockfish 14.1     : 3780    6    6  9000    72.2 %   3564   53.6 %
   3 SugaR AI 2.50      : 3779    6    6  6500    64.9 %   3661   67.1 %
   4 BrainLearn 15.1    : 3762    6    6  9000    70.3 %   3565   55.6 %
   5 ShashChess 20.1    : 3754    6    6  9000    69.4 %   3566   56.1 %
   6 Dragon 2.6         : 3743    7    7  6740    65.9 %   3587   57.6 %
   7 Stockfish 13       : 3735    5    5  9000    67.2 %   3567   56.0 %
   8 Dragon 2.5.1       : 3725    6    6  8500    67.2 %   3557   52.6 %
   9 Fire 8.NN.MC.3     : 3610    6    6  8340    54.5 %   3558   54.5 %
  10 Revenge 2.0        : 3572    6    6  8340    47.5 %   3578   49.4 %
  11 Slow Chess 2.8     : 3569    6    6  8840    48.1 %   3573   51.6 %
  12 Berserk 8 NN       : 3565    6    6  8840    47.6 %   3573   49.3 %
  13 Koivisto 7.9       : 3562    6    6  8840    47.3 %   3573   49.9 %
  14 Ethereal 13.25     : 3554    6    6  8840    46.4 %   3574   48.1 %
  15 Revenge 1.0        : 3483    6    6  8340    38.3 %   3578   40.9 %
  16 Wasp 5.00          : 3285    8    8  8340    20.4 %   3578   26.8 %
  17 Fritz 18           : 3279    7    7  8340    20.0 %   3578   24.8 %
  18 Chiron 5           : 3225    8    8  8340    15.9 %   3581   20.6 %
  19 Critter v1.6a      : 3164    8    8  8340    11.9 %   3585   16.5 %

Individual statistics:

6 Dragon 2.6       : 3743 6740 (+2498,=3883,-359), 65.9 %

Revenge 2.0        : 340 (+144,=193,-  3), 70.7 %
SugaR AI 2.50      : 500 (+  9,=427,- 64), 44.5 %
ShashChess 20.1    : 500 (+ 18,=437,- 45), 47.3 %
Slow Chess 2.8     : 340 (+155,=183,-  2), 72.5 %
Ethereal 13.25     : 340 (+176,=164,-  0), 75.9 %
Chiron 5           : 340 (+312,= 27,-  1), 95.7 %
Koivisto 7.9       : 340 (+166,=170,-  4), 73.8 %
Critter v1.6a      : 340 (+320,= 20,-  0), 97.1 %
Fire 8.NN.MC.3     : 340 (+130,=202,-  8), 67.9 %
Wasp 5.00          : 340 (+296,= 44,-  0), 93.5 %
BrainLearn 15.1    : 500 (+ 23,=427,- 50), 47.3 %
Stockfish 13       : 500 (+ 40,=423,- 37), 50.3 %
Stockfish 14.1     : 500 (+ 11,=426,- 63), 44.8 %
Zeus 21.0          : 500 (+ 12,=412,- 76), 43.6 %
Fritz 18           : 340 (+298,= 42,-  0), 93.8 %
Revenge 1.0        : 340 (+220,=116,-  4), 81.8 %
Berserk 8 NN       : 340 (+168,=170,-  2), 74.4 %

[Chessqueen]

Is ratings all that most of you care about here ? To me newer features is more important. But since 2013 Stockfish and Komodo has been number 1 and 2 on the rating list, and Deep Blue was the best from 1996 to 2006

[criko]

I think that is very nice for you. Keep up what is important for you and much fun with your search for new features.

[lkaufman]

Same version has 10,5 - 19,5 against SF dev in CCC?

I think it's because CCC uses unbalanced openings, which magnify elo differences. With more threads, the effect of unbalanced books becomes larger and larger, as with normal books the draw percentage starts to approach 100% with the best engines on many threads (or long time controls). With opening positions near the win/draw line, a slightly better engine might consistently win with White and draw with Black, scoring 75% = +192 elo.

[Jouni]

Yes true. SF has won EVERY 16 white games vs Ethereal!

[lkaufman]

Yes true. SF has won EVERY 16 white games vs Ethereal!

If in the future rating lists use opening books like this, we'll continue to see large elo gains even at long time limits for years to come. Ratings like 5000 will be possible. What that means is a matter of opinion; it's basically like removing the draw factor from chess. Imagine if 100 yard races were run with the rule that any result within 10% would be called a draw. All the best runners would be tied with each other almost every time. Then someone might suggest that they take turns running 110 yards vs 100 yards. Suddenly half the races would have a winner. Then a runner who was 1% faster than another would consistently win when he had the shorter distance to run and draw when he had the longer distance. He would score 75%, +192 elo! This is almost exactly what we are seeing here. We have found a solution to the draw problem in computer chess. Whether that makes it a different game than normal chess can be debated. The only other problem with this is that the choice of openings is arbitrary, dependent on some arbitrary rules and decisions. It would be nice to have opening chosen by some simple rules that don't involve an engine's judging the positions. FRC (960) is an example of this, but unfortunately those positions aren't near the win/draw line. Some similar rules that did produce positions near the line would be ideal.

[Cornfed]

Yes true. SF has won EVERY 16 white games vs Ethereal!

If in the future rating lists use opening books like this, we'll continue to see large elo gains even at long time limits for years to come. Ratings like 5000 will be possible. What that means is a matter of opinion; it's basically like removing the draw factor from chess. Imagine if 100 yard races were run with the rule that any result within 10% would be called a draw. All the best runners would be tied with each other almost every time. Then someone might suggest that they take turns running 110 yards vs 100 yards. Suddenly half the races would have a winner. Then a runner who was 1% faster than another would consistently win when he had the shorter distance to run and draw when he had the longer distance. He would score 75%, +192 elo! This is almost exactly what we are seeing here. We have found a solution to the draw problem in computer chess. Whether that makes it a different game than normal chess can be debated. The only other problem with this is that the choice of openings is arbitrary, dependent on some arbitrary rules and decisions. It would be nice to have opening chosen by some simple rules that don't involve an engine's judging the positions. FRC (960) is an example of this, but unfortunately those positions aren't near the win/draw line. Some similar rules that did produce positions near the line would be ideal.

I like it. Does this "make it a different game than normal chess..."? No - it's the same rules, piece placings, etc. It's those abominations (variants) that are truly different games.

I would like to see more 'black wins' in normal chess between engines though. Can they 'only' win with White? This means you let the engines play decidedly sub-optimal (but not outright stupid) White set ups with sometimes very short opening lines. OTB and online, I've backed into reversed Benoni's and Benko Gambits more times than I can count. I win a lot simply because I understand the structures better than many.