Chess variant tournament: Capablanca and similar

[enhorning]

Unambitious game:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.04.04"]
[Round "22"]
[White "Spartacus 1.02"]
[Black "Max-Plus 4.8S"]
[Result "1/2-1/2"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rnbackqbnr/pppppppppp/10/10/10/10/PPPPPPPPPP/RNBACKQBNR w KQkq - 0 1"]
[SetUp "1"]

1. h4 {+0.17/14} Nc6 {+0.17/10 14} 2. Nh3 {+0.22/14 21} g6 {+0.12/10 24} 3.
d3 {+0.26/13 16} Nh6 {+0.09/10 43} 4. g3 {+0.26/13 24} d6 {-0.06/9 20} 5.
Qi3 {-0.21/12 20} Ng4 {+0.01/9 17} 6. Qg1 {-0.02/14 32} Nh6 {-0.01/10 13}
7. Qi3 {+0.00/13 21} Ng4 {+0.00/10 20} 8. Qg1 {+0.00/16 22} Nh6
{-0.01/11 18}
{XBoard adjudication: repetition draw} 1/2-1/2

[hgm]

Your proposed list seems quite reasonable.

[enhorning]

The games from the mirrored Embassy chess starting position have finished now.

Standings:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Biha Biha TJch Nebi Nebi Joke Spar Sjaa Smir TSCP Max- Pair Ches Fair BigL ArcB Here
 1. Bihasa v3.4.3             #### 0=10 1111 1011 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111   95%  60.5 (1934.0, 1737.5)
 2. Bihasa v3.2               1=01 #### 1011 1110 1111 111= 1101 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111   92%  59.0 (1940.0, 1679.5)
 3. TJchess10x8 1.1-x64       0000 0100 #### 1111 1111 111= 1011 1111 1111 1111 1111 1111 1111 1111 1111 1111 1111   87%  55.5 (1954.0, 1469.8)
 4. NebiyuChess_1.43          0100 0001 0000 #### 1111 0010 1111 11== 1111 1111 0111 1111 1111 1111 1111 1111 1111   77%  49.0 (1980.0, 1217.5)
 5. NebiyuChess_1.44          0000 0000 0000 0000 #### 1110 1110 1111 =111 1111 1101 1111 =111 1111 1111 1111 1111   69%  44.0 (2000.0, 971.3)
 6. Joker80.np                0000 000= 000= 1101 0001 #### 0011 1010 1=11 1111 =111 1111 1111 =111 1111 1111 1111   68%  43.5 (2002.0, 1000.0)
 7. Spartacus 1.02            0000 0010 0100 0000 0001 1100 #### 1111 1=11 1111 111= 1111 =111 1111 1111 1111 1111   68%  43.5 (2002.0, 967.0)
 8. Sjaak 524                 0000 0000 0000 00== 0000 0101 0000 #### 0010 1111 1011 0110 =1=1 111= 0111 1=11 1110   45%  29.0 (2060.0, 576.5)
 9. Smirf BC-173g-X           0000 0000 0000 0000 =000 0=00 0=00 1101 #### 1101 1111 1101 1100 ==01 0111 1011 1110   43%  27.5 (2066.0, 537.0)
10. TSCPtest                  0000 0000 0000 0000 0000 0000 0000 0000 0010 #### 1011 0=01 111= 1011 1=11 1111 1111   37%  23.5 (2082.0, 356.0)
11. Max-Plus 4.8S             0000 0000 0000 1000 0010 =000 000= 0100 0000 0100 #### 011= 01== 1111 0011 1111 0111   35%  22.5 (2086.0, 428.8)
12. Pair-o-Max 4.8S           0000 0000 0000 0000 0000 0000 0000 1001 0010 1=10 100= #### 111= 1==0 0100 1110 1111   32%  20.5 (2094.0, 344.5)
13. ChessV_WinBoard           0000 0000 0000 0000 =000 0000 =000 =0=0 0011 000= 10== 000= #### 001= =111 1111 0111   30%  19.0 (2100.0, 327.3)
14. Fairy-Max 4.8S            0000 0000 0000 0000 0000 =000 0000 000= ==10 0100 0000 0==1 110= #### =011 1=01 1101   26%  16.5 (2110.0, 282.8)
15. BigLion80 2.23x WB        0000 0000 0000 0000 0000 0000 0000 1000 1000 0=00 1100 1011 =000 =100 #### 0111 1101   24%  15.5 (2114.0, 254.0)
16. ArcBishop80 1.01 WB       0000 0000 0000 0000 0000 0000 0000 0=00 0100 0000 0000 0001 0000 0=10 1000 #### 1=11   13%   8.5 (2142.0, 125.5)
17. Heretic 0.3               0000 0000 0000 0000 0000 0000 0000 0001 0001 0000 1000 0000 1000 0010 0010 0=00 ####   10%   6.5 (2150.0, 134.3)

Bayeselo give these ratings and LOS matrix:

Code: Select all

ResultSet-EloRating>ratings
Rank Name                  Elo    +    - games score oppo. draws
   1 Bihasa v3.4.3         608  157  157    64   95%   -38    2%
   2 Bihasa v3.2           552  146  146    64   92%   -35    3%
   3 TJchess10x8 1.1-x64   444  129  129    64   87%   -28    2%
   4 NebiyuChess_1.43      281  115  115    64   77%   -18    3%
   5 NebiyuChess_1.44      182  107  107    64   69%   -11    3%
   6 Joker80.np            176  107  107    64   68%   -11    8%
   7 Spartacus 1.02        162  106  106    64   68%   -10    5%
   8 Sjaak 524             -72   97   97    64   45%     5    9%
   9 Smirf BC-173g-X       -83   97   97    64   43%     5    8%
  10 TSCPtest             -164   96   96    64   37%    10    5%
  11 Max-Plus 4.8S        -172   97   97    64   35%    11    8%
  12 Pair-o-Max 4.8S      -211   97   97    64   32%    13    8%
  13 ChessV_WinBoard      -220   94   94    64   30%    14   16%
  14 Fairy-Max 4.8S       -264   96   96    64   26%    17   14%
  15 BigLion80 2.23x WB   -294  102  102    64   24%    18    5%
  16 ArcBishop80 1.01 WB  -431  115  115    64   13%    27    5%
  17 Heretic 0.3          -493  128  128    64   10%    31    2%
ResultSet-EloRating>los
                     Bi Bi TJ Ne Ne Jo Sp Sj Sm TS Ma Pa Ch Fa Bi Ar He
Bihasa v3.4.3           71 95 99 99 99 99 99 99 99 99 99 99100100100100
Bihasa v3.2          28    87 99 99 99 99 99 99 99 99 99 99100100100100
TJchess10x8 1.1-x64   4 12    97 99 99 99 99 99 99 99 99 99 99 99100100
NebiyuChess_1.43      0  0  2    90 91 94 99 99 99 99 99 99 99 99 99100
NebiyuChess_1.44      0  0  0  9    52 60 99 99 99 99 99 99 99 99 99 99
Joker80.np            0  0  0  8 47    57 99 99 99 99 99 99 99 99 99 99
Spartacus 1.02        0  0  0  5 39 42    99 99 99 99 99 99 99 99 99 99
Sjaak 524             0  0  0  0  0  0  0    56 91 93 97 98 99 99 99 99
Smirf BC-173g-X       0  0  0  0  0  0  0 43    88 90 97 97 99 99 99 99
TSCPtest              0  0  0  0  0  0  0  8 11    54 75 80 93 97 99 99
Max-Plus 4.8S         0  0  0  0  0  0  0  6  9 45    71 76 91 96 99 99
Pair-o-Max 4.8S       0  0  0  0  0  0  0  2  2 24 28    55 79 88 99 99
ChessV_WinBoard       0  0  0  0  0  0  0  1  2 19 23 44    75 86 99 99
Fairy-Max 4.8S        0  0  0  0  0  0  0  0  0  6  8 20 24    67 98 99
BigLion80 2.23x WB    0  0  0  0  0  0  0  0  0  2  3 11 13 32    96 99
ArcBishop80 1.01 WB   0  0  0  0  0  0  0  0  0  0  0  0  0  1  3    77
Heretic 0.3           0  0  0  0  0  0  0  0  0  0  0  0  0  0  0 22

Max-Plus has finally passed Pair-o-Max, even though its LOS is only 71%.

Given the results so far, Nebiyu 1.43 seems stronger than 1.44 - Bayeselo gives 90% likelihood of superiority. Given this, and given that I said I had space for one more program, I am tempted to include both versions of Nebiyu for the remainder of the tournament as well - to possibly either get higher certainty that 1.43 is indeed stronger, or to find out that it is due to random fluctuations.

The games so far can be downloaded from: https://www.dropbox.com/s/3fnskpwc5k383zq/Capa30sAB.pgn

Out of 544 games, there have been 256 White wins, 33 Draws, and 255 Black wins.

I will start the next set of games, with the cut down list of participants, tomorrow night.

[hgm]

I am very surprised by the poor performance of Heretic. The initial version I was running in BotG 2012 was severely bugged, but near the end of the qualifier I replaced it by a version 0.2, which then won virtually all remaining games in the qualifier and consolation group. It seems that this v0.3 is again a regression.

[enhorning]

I am very surprised by the poor performance of Heretic. The initial version I was running in BotG 2012 was severely bugged, but near the end of the qualifier I replaced it by a version 0.2, which then won virtually all remaining games in the qualifier and consolation group. It seems that this v0.3 is again a regression.

Hmm - what time control were you using?

Heretic did clearly better than this in my blitz tests before the tournament... it might be that it's good a blitz, but not at longer time controls?

[hgm]

I don't recall exactly what time control I was using, but I am sure it was not blitz. Must have been at least 20 min (on a 2.4GHz Core 2 Duo, 32-bit).

[enhorning]

No idea then. I have noticed it losing some games on time, but haven't kept track of how many.

[enhorning]

More castling weirdness:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.04.07"]
[Round "14"]
[White "Max-Plus 4.8S"]
[Black "Spartacus 1.02"]
[Result "1-0"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rqnbakbncr/pppppppppp/10/10/10/10/PPPPPPPPPP/RQNBAKBNCR w KQkq - 0 1"]
[SetUp "1"]

{--------------
r q n b a k b n c r
p p p p p p p p p p
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
P P P P P P P P P P
R Q N B A K B N C R
white to play
--------------}
1. c3 {+0.07/10} Af6 {-0.02/15 29} 2. Nd3 {+0.13/10 27} d6 {+0.14/14 27} 3.
Ng3 {+0.04/10 15} e5 {+0.18/14 16} 4. h3 {-0.05/11 13} Ni6 {+0.08/14 20} 5.
Bi3 {-0.04/11 14} Nh4 {-0.07/14 22} 6. f3 {+0.02/10 17} Ae7 {-0.14/14 25}
7. Ne4 {+0.12/10 14} f5 {+0.04/16 29} 8. g3 {+0.03/11 24} fxe4
{-0.07/16 31} 9. fxe4 {-0.33/12 30} i5 {-0.15/15 17} 10. gxh4 {-0.13/12 31}
ixh4 {-0.22/16 49} 11. Bh2 {-0.12/11 15} Ag5 {-0.39/15 30} 12. Nf2
{-0.21/10 14} Be6 {-0.42/14 15} 13. Qd3 {-0.15/10 16} Ne7 {-0.28/13 16} 14.
Bb3 {+0.00/11 23} Bxb3 {-0.42/14 20} 15. axb3 {-0.05/12 45} Ng6
{-0.40/14 34} 16. Af3 {-0.01/11 16} Nf4 {-0.24/15 28} 17. Axg5
{+0.15/13 25} Bxg5 {-0.28/16 44} 18. Qf3 {+0.30/12 18} Ke7 {-0.28/15 36}
19. O-O-O {+0.17/12 36}
{False illegal-move claim} 1-0

[enhorning]

Finished the games from the Schoolhouse Chess starting position:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Bi TJ Ne Ne Sp Sm Jo TS Sj Ma
 1. Bihasa v3.4.3             ## 1= 11 11 11 11 11 11 11 11   97%  17.5 (145.0, 139.0)
 2. TJchess10x8 1.1-x64       0= ## 11 10 11 =0 =0 11 11 =1   67%  12.0 (156.0,  97.0)
 3. NebiyuChess_1.43          00 00 ## 1= 01 11 11 11 1= 11   67%  12.0 (156.0,  78.0)
 4. NebiyuChess_1.44          00 01 0= ## 01 01 11 11 11 11   64%  11.5 (157.0,  74.5)
 5. Spartacus 1.02            00 00 10 10 ## 11 11 11 11 10   61%  11.0 (158.0,  72.0)
 6. Smirf BC-173g-X           00 =1 00 10 00 ## 01 00 11 10   36%   6.5 (167.0,  48.5)
 7. Joker80.np                00 =1 00 00 00 10 ## 11 00 11   36%   6.5 (167.0,  41.5)
 8. TSCPtest                  00 00 00 00 00 11 00 ## 10 11   28%   5.0 (170.0,  24.5)
 9. Sjaak 524                 00 00 0= 00 00 00 11 01 ## 10   25%   4.5 (171.0,  27.5)
10. Max-Plus 4.8S             00 =0 00 00 01 01 00 00 01 ##   19%   3.5 (173.0,  28.0)

TJChess performed worse than it had previously - could be a fluke, or could be that having two of the power pieces start so far out on the flanks did not suit it.

So, combining that with the first two starting positions gives us these standings:
1. Bihasa 3.4.3 - 78.0
2. TJChess10x8 - 67.5
3. Nebiyu 1.43 - 61.0
4. Nebiyu 1.44 - 55.5
5. Spartacus - 54.5
6. Joker - 50.0
7. Smirf - 34.0
8. Sjaak - 33.5
9. TSCP - 28.5
10. MaxPlus - 26.0

PGN for this latest round of games can be downloaded from: https://www.dropbox.com/s/9hpf0wd4cl6fg1e/Capa30sC.pgn

Next up is the unbalanced starting position, with all three power pieces on one flank.

[enhorning]

Hmm, why did Max-Plus not see this checkmate coming, after spending 1 minute on the move before?

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.04.09"]
[Round "13"]
[White "Smirf BC-173g-X"]
[Black "Max-Plus 4.8S"]
[Result "1-0"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rbbnnkcaqr/pppppppppp/10/10/10/10/PPPPPPPPPP/RBBNNKCAQR w KQkq - 0 1"]
[SetUp "1"]

{--------------
r b b n n k c a q r
p p p p p p p p p p
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
P P P P P P P P P P
R B B N N K C A Q R
white to play
--------------}
1. Nf3 {+0.31/9} Nf6 {+0.11/10 17} 2. d4 {+0.35/9 27} Ne6 {+0.10/11 25} 3.
c4 {+0.42/9 26} d5 {+0.13/11 40} 4. cxd5 {+0.94/10 23} Nxd5 {-0.11/11 27}
5. e4 {+0.90/9 27} Nf6 {-0.09/11 18} 6. e5 {+1.09/9 20} Nd7 {-0.26/11 19}
7. Ag3 {+1.52/9 27} g6 {+0.06/10 20} 8. h3 {+1.32/9 28} c5 {+0.15/11 14} 9.
d5 {+1.40/9 27} Ng7 {-0.05/11 19} 10. d6 {+1.55/8 21} f6 {-0.01/11 32} 11.
exf6 {+2.13/9 28} Cxf6 {+0.36/11 15} 12. dxe7+ {+2.37/9 28} Ke8
{+0.83/11 21} 13. Bf4 {+2.45/11 45} Nh5 {+1.28/12 14} 14. Bxb8
{+2.58/11 44} Nxg3+ {+0.76/11 19} 15. Bxg3 {+2.73/10 43} b6 {+0.85/11 38}
16. Bd3 {+3.23/10 39} Cd5 {+0.54/11 23} 17. Bc4 {+3.49/10 19} Cb4
{-0.26/11 17} 18. Rc1 {+3.76/11 42} Ba6 {+0.61/11 15} 19. Bxa6
{+3.85/11 38} Cxa2 {+0.44/12 18} 20. Ce2 {+4.73/10 20} Cxa6 {-1.81/12 1:04}
21. Nc3 {+5.51/10 26} Ne5 {-2.36/10 13} 22. Bxe5 {+8.63/9 19} Ai6
{-7.42/12 32} 23. Ce4 {+10.52/9 19} b5 {-9.36/12 36} 24. Bf6 {+12.86/9 35}
Axf3 {-10.71/10 24} 25. gxf3 {+21.15/10 17} Qg8 {-18.20/10 13} 26. Qd6
{+22.04/10 18} Qc4+ {-19.98/12 17} 27. Cxc4 {+25.28/11 20} Cxd6
{-11.73/14 1:00} 28. Cxd6#
{Xboard adjudication: Checkmate} 1-0