Chess variant tournament: Capablanca and similar

[hgm]

Depends on how long the games end up being as well - I don't really have a good feeling for how long Capablanca games typically last.

Should not be longer than normal games (~60 moves on average between equal opponents). As the participant strength differs wildly, I expect the majority of the games to last much shorter.

And you are right; it is a bit pointless to play ChessV or BigLion against Bihasa. (This almost holds for any other engine against Bihasa too, btw...) What I did in Battle of the Goths 2012 was to have a qualifier (from which I had exempted the top 3 and bottom 3 of the previous event, plus Bihasa (which blitz tests had shown to be super-strong), and then split in two groups based on the qualifier result. The Qualifier result was

Code: Select all

                              Spar Nebi tscp Fair Here Sjaa
 1. Spartacus                 #### 10=1 1111 1111 1111 1111
                              #### 1110 1111 1111 1101 1111   91%  36.5 (668.0, 578.3)

 2. NebiyuChess               01=0 #### 0011 1111 1111 1111
                              0001 #### 1111 1111 1111 1111   81%  32.5 (700.0, 457.3)

 3. tscpgothic                0000 1100 #### 0111 1111 1111
                              0000 0000 #### 1011 1100 1101   52%  21.0 (792.0, 253.0)

 4. Fairy-Max                 0000 0000 1000 #### 111= 1110
                              0000 0000 0100 #### 1100 0111   34%  13.5 (852.0, 136.8)

 5. Heretic                   0000 0000 0000 000= #### 0010
                              0010 0000 0011 0011 #### 0011   21%   8.5 (892.0, 136.3)

 6. Sjaak                     0000 0000 0000 0001 1101 ####
                              0000 0000 0010 1000 1100 ####   20%   8.0 (896.0,  90.5)

So I finally ended up with a 'champions league' Bihasa, Joker, SMIRF, TJchess, Spartacus, Nebiyu and TSCP, and a 'consolation group' Fairy-Max, Sjaak, Heretic, BigLion, ArcBishop, ChessV. (Where the results from the qualifier kept counting and were not replayed.)

Heretic was brand new at that time, and crippled by bugs in the qualifier. After the bugs were fixed it won nearly all its games. But then it was already in the consolation group, and because its old results against the main competitors there kept counting, it did not even manage to rise to the top there. I am really curious how well it will do this time, now that the fixed version can play from the beginning.

The private version of ChessV I was running also seemed unusually buggy. As you are running the public version, it might do a lot better than in BotG 2012, where it ended last. Nebiyu and Sjaak will probably also have improved (if only because of the 64-bit hardware). Spartacus, Joker and SMIRF are still exactly the same, Fairy-Max might even be weaker, as I think I was running an especially fast PGO compile then, not my own 'gcc -O2' compile I distribute with WinBoard. Not sure about TJchess. AFAIK BigLion and ArcBishop also have not been changed. Only Max-Plus is new, but it should be a medium-class engine. With hardly more than material for evaluation, I would be surprised if it could reach the level of TSCP-Gothic.

[enhorning]

First few results:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              T B B S N P J M S S A B T C H F N
 1. TJchess10x8 1.1-x64       #           1 1           1 1      100%   4.0 (  4.5,   4.5)
 2. Bihasa v3.2                 #   1 1               1          100%   3.0 (  6.0,   6.0)
 3. Bihasa v3.4.3                 #             1             1  100%   2.0 (  1.5,   1.5)
 4. Spartacus 1.02              0   #     1         1     1       75%   3.0 (  6.0,   3.0)
 5. NebiyuChess_1.44            0     # 1                   1     67%   2.0 (  5.0,   2.0)
 6. Pair-o-Max 4.8S                   0 #             1     1     67%   2.0 (  3.0,   1.0)
 7. Joker80.np                0     0     #   1       1           50%   2.0 ( 10.0,   3.0)
 8. Max-Plus 4.8S             0             #   0 1           1   50%   2.0 (  6.5,   1.0)
 9. Smirf BC-173g-X                       0   #     0   1 1       50%   2.0 (  3.5,   0.5)
10. Sjaak 524                     0         1   #       =         50%   1.5 (  4.5,   2.3)
11. ArcBishop80 1.01 WB                     0     #       1       50%   1.0 (  2.0,   0.0)
12. BigLion80 2.23x WB              0         1     # 0           33%   1.0 (  6.0,   2.0)
13. TSCPtest                    0       0 0         1 #           25%   1.0 (  8.0,   1.0)
14. ChessV_WinBoard           0               0 =       #         17%   0.5 (  7.5,   0.8)
15. Heretic 0.3               0     0         0   0       #        0%   0.0 ( 10.0,   0.0)
16. Fairy-Max 4.8S                    0 0                   #      0%   0.0 (  4.0,   0.0)
17. NebiyuChess_1.43              0         0                 #    0%   0.0 (  4.0,   0.0)

So far, average game length is 59 moves, so games are getting done at quite a decent pace.

The result that surprised me at a quick glance was Max-Plus's win against Nebiyu 1.43 - seems Nebiyu under-estimated the passed i-pawn:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.03.29"]
[Round "1"]
[White "Max-Plus 4.8S"]
[Black "NebiyuChess_1.43"]
[Result "1-0"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rnabqkbcnr/pppppppppp/10/10/10/10/PPPPPPPPPP/RNABQKBCNR w KQkq - 0 1"]
[SetUp "1"]

{--------------
r n a b q k b c n r
p p p p p p p p p p
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
P P P P P P P P P P
R N A B Q K B C N R
white to play
--------------}
1. h3 {+0.02/10} Nc6 {+0.20/17 28} 2. e3 {+0.17/10 29} f5 {+0.15/17 28} 3.
d3 {+0.12/9 13} Nh6 {+0.30/16 29} 4. Nd2 {+0.09/11 38} Be6 {+0.20/15 29} 5.
Nf3 {+0.09/10 14} Cg6 {-0.10/16 29} 6. Bi3 {+0.09/11 41} i6 {+0.00/18 29}
7. g3 {+0.09/10 15} Bf7 {+0.10/17 29} 8. Ng5 {+0.05/11 34} Bg8
{+0.10/17 29} 9. e4 {+0.00/10 35} Rh8 {-0.15/16 29} 10. Af4 {+0.15/10 23}
fxe4 {+0.15/16 24} 11. Axg6+ {+0.45/11 18} Qxg6 {+0.00/18 25} 12. Nxe4
{+0.40/11 20} d5 {+0.05/17 25} 13. Nd2 {+0.44/11 31} Ad6 {+0.10/17 25} 14.
Ng2 {+0.57/11 26} e6 {+0.15/16 24} 15. Nf4 {+0.57/11 22} Qf7 {+0.30/18 24}
16. Bxd8 {+0.48/11 20} Rxd8 {+0.60/18 28} 17. Nf3 {+0.53/11 23} Re8
{+0.60/18 29} 18. Be2 {+0.52/10 12} e5 {+1.45/22 30} 19. Ng2 {+0.01/11 16}
e4 {+1.40/21 30} 20. dxe4 {+0.01/11 18} dxe4 {+1.30/21 31} 21. Nfh4
{-0.09/11 28} g5 {+1.60/19 31} 22. Qc3 {-0.51/10 24} Ae5 {+2.10/20 25} 23.
Qc5+ {-0.87/10 13} Qe7 {+2.40/23 24} 24. Qxe7+ {-1.33/11 17} Rxe7
{+2.45/23 26} 25. Nj3 {-1.35/12 16} Axb2 {+2.45/22 32} 26. Re1
{-1.66/13 33} Nd4 {+2.40/20 32} 27. Ne3 {-1.59/12 29} Bxa2 {+2.45/20 32}
28. Ch2 {-1.84/11 20} Nxe2 {+2.45/20 32} 29. Rxe2 {-1.91/12 20} Be6
{+2.50/19 31} 30. O-O {-1.53/11 24} Nf7 {+2.35/17 33} 31. Ng4 {-1.73/11 17}
Nd6 {+2.40/20 31} 32. f3 {-1.70/11 30} h5 {+2.10/20 33} 33. Ngi5
{-1.70/11 43} Rg8 {+2.85/19 26} 34. fxe4 {-2.03/11 32} Ae5 {+3.10/20 34}
35. h4 {-2.57/11 28} g4 {+4.00/22 35} 36. Re3 {-2.65/11 19} j6
{+4.05/21 36} 37. Rf1+ {-2.74/12 45} Ke8 {+4.20/22 34} 38. Cg2
{-2.91/13 34} jxi5 {+4.20/20 29} 39. hxi5 {-3.01/12 30} Bc4 {+4.15/21 42}
40. Rf4 {-2.90/12 28} Rg6 {+4.50/22 46} 41. Nh4 {-2.94/12 29} Rge6
{+4.40/21 29} 42. Cf2 {-3.03/11 22} Ad4 {+4.55/21 22} 43. Cg2 {-3.54/12 20}
Ag7 {+4.65/21 22} 44. j4 {-3.55/12 17} Bd5 {+4.60/22 29} 45. Ng6
{-3.48/13 19} Bxe4 {+4.50/25 29} 46. Rfxe4 {-3.15/15 29} Rxe4 {+4.45/25 29}
47. Nxe7 {-3.16/15 16} Rxe7 {+4.35/27 29} 48. Rxe7+ {-3.11/14 29} Kxe7
{+4.30/25 29} 49. Ce3+ {-3.25/15 1:17} Ae6 {+4.90/24 28} 50. j5
{-2.86/16 22} ixj5 {+2.50/25 24} 51. i6 {-2.77/16 23} Nf5 {+2.85/26 29} 52.
Ce5 {-2.83/15 20} Ng7 {+2.55/25 23} 53. i7 {-2.68/15 23} a5 {+2.50/24 30}
54. Cxa5 {-2.12/14 20} h4 {+2.45/24 30} 55. gxh4 {-1.06/14 19} Kf6
{+2.45/24 29} 56. Cxj5 {-0.82/13 17} Af5 {+1.70/24 36} 57. Cj8
{-0.64/13 28} Ni8 {+1.70/22 24} 58. Cj4 {-0.69/13 18} Ke5 {+2.35/23 24} 59.
Ch5 {-0.69/13 20} c6 {+2.25/24 30} 60. Cg5 {-0.37/15 30} b5 {+1.60/24 58}
61. Cf7+ {-0.11/16 58} Kf4 {+1.35/27 32} 62. i4 {-0.05/15 19} Nh6
{+0.55/26 27} 63. Cf8 {+0.78/16 29} b4 {+0.00/28 26} 64. i5 {+2.63/16 20}
Ni8 {+0.00/30 22} 65. h5 {+2.60/16 46} b3 {-1.10/29 1:55} 66. cxb3
{+3.04/15 21} Ke5 {-1.10/29 23} 67. Ce8+ {+3.04/14 17} Kf4 {-6.20/29 36}
68. Cf6 {+3.01/15 24} Ke5 {-2.30/29 25} 69. Cxc6+ {+3.01/15 29} Kf4
{-7.50/30 1:08} 70. Cc4+ {+3.02/15 33} Kg3 {-8.10/28 18} 71. Cd2
{+8.17/15 18} Kh4 {-6.95/27 21} 72. h6 {+8.20/16 21} Kxi5 {-7.20/28 22} 73.
Cd5 {+10.52/19 39} Kxh6 {-18.50/30 28} 74. Cxf5+ {+15.90/19 37} Kxi7
{-18.55/28 21} 75. b4 {+15.91/22 27} Nh6 {-19.20/28 21} 76. Cg7+
{+17.23/22 31} Kj6 {-18.75/25 20} 77. b5 {+18.98/21 23} Nf7 {-299.81/25 22}
78. Cxf7 {+79.91/22 22} Ki6 {-299.83/24 18} 79. b6 {+79.91/17 32} g3
{-299.85/24 21} 80. Cg7+ {+79.93/17 20} Ki5 {-299.87/23 33} 81. b7
{+79.94/19 24} Kh6 {-299.89/22 28} 82. Cxg3 {+79.95/21 23} Ki5
{-299.91/22 28} 83. b8=Q {+79.96/29 6} Kh6 {-299.93/22 29} 84. Qe8
{+79.97/29 0.3} Ki6 {-299.95/23 29} 85. Qi8+ {+79.98/29 0.1} Kh6
{-299.97/24 23} 86. Cf5# {+79.99/29 0.1}
{Xboard adjudication: Checkmate} 1-0

[hgm]

Good swindle! And a nice success for one of the few evaluation terms Fairy-Max has, pawns on 6th and 7th rank. This makes it sacrifice its j-Pawn to advance its i-Pawn to the 7th.

It seems to me that Nebiyu is a bit naive towards advanced Pawns in general. In Spartan it also often loses because it draws its Hoplites towards the center, and then lets the Persian edge Pawns walk until it is too late.

[hgm]

If you do replacements, I have uploaded a new version of MaxPlus.exe, to the same link. This fixes the repetition detection of tree positions, which I had to disable in the previous version because of a bug. After 250 games this new version is leading by about 1.2 standard deviations against the previous one, which is enough to satisfy me that it is at least not worse.

[enhorning]

If you do replacements, I have uploaded a new version of MaxPlus.exe, to the same link. This fixes the repetition detection of tree positions, which I had to disable in the previous version because of a bug. After 250 games this new version is leading by about 1.2 standard deviations against the previous one, which is enough to satisfy me that it is at least not worse.

As long as they've been reasonably well tested (and uses the same file name, so I can just copy it over), I can do replacements - so, done, in this case. What I want to avoid is putting in a new version mid-tourney that might be weaker - or even worse, which might be buggier, causing technical trouble.

Had the shortest game of the tourney so far - MaxPlus fell victim to a coordinated attack from the three heavy pieces and a rook, by Smirf:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.03.29"]
[Round "5"]
[White "Max-Plus 4.8S"]
[Black "Smirf BC-173g-X"]
[Result "0-1"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rnabqkbcnr/pppppppppp/10/10/10/10/PPPPPPPPPP/RNABQKBCNR w KQkq - 0 1"]
[SetUp "1"]

{--------------
r n a b q k b c n r
p p p p p p p p p p
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
P P P P P P P P P P
R N A B Q K B C N R
white to play
--------------}
1. c3 {+0.07/10} Nh6 {+0.03/9 29} 2. d3 {+0.11/10 20} Nc6 {+0.42/10 43} 3.
h3 {+0.18/10 20} f5 {+0.93/9 26} 4. Nd2 {+0.10/10 13} e5 {+1.02/9 29} 5.
Nf3 {+0.04/11 30} Ci6 {+1.38/9 23} 6. Bi3 {-0.07/10 19} Bxi3 {+0.90/8 26}
7. jxi3 {-0.03/11 14} Ae7 {+0.94/8 20} 8. Nh4 {+0.02/11 25} Ch8
{+0.72/9 43} 9. g3 {+0.07/11 54} i5 {+0.89/9 42} 10. Nf3 {+0.03/10 24} e4
{+0.85/9 26} 11. dxe4 {+0.03/11 17} fxe4 {+0.56/9 25} 12. Nd2 {-0.06/11 21}
d5 {+0.93/9 27} 13. Ng2 {+0.04/11 22} Bf7 {+0.73/9 42} 14. Ch2
{-0.01/11 21} j5 {+0.90/9 39} 15. O-O {+0.11/11 46} Qd8 {+1.24/9 30} 16.
Cf1 {-0.09/10 21} j4 {+2.31/9 20} 17. ixj4 {-0.27/10 21} Ci6 {+2.27/8 21}
18. f3 {-0.18/9 22} Cxj4 {+8.12/9 27} 19. Kh2 {-5.57/11 22} Ag5
{+9.56/8 20} 20. Nh4 {-17.28/10 29} ixh4 {+21.38/8 34} 21. Qf2
{-18.14/11 28} hxg3+ {+22.71/8 10} 22. Kg1 {-79.94/13 26} gxf2+
{+22.71/6 1.8} 23. Kxf2 {-79.95/13 34} Ah4+ 24. Cg3 {-79.96/17 3:20} Axg3+
25. Kg2 {-79.97/23 9} Cxi2+ 26. Rh2 {-79.97/21 11} Axh2 27. fxe4
{-79.98/27 14:29} Qg5+ 28. Kh1 {-26.25/1 0.1} Ag3#
{Xboard adjudication: Checkmate} 0-1

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              B T B N S P S J M T S N A B C H F
 1. Bihasa v3.2               #     1 1         1             1  100%   4.0 (  9.5,   9.5)
 2. TJchess10x8 1.1-x64         #           1 1           1 1    100%   4.0 (  5.0,   5.0)
 3. Bihasa v3.4.3                 #               1 1 1          100%   3.0 (  3.5,   3.5)
 4. NebiyuChess_1.44          0     #   1   1           1     1   80%   4.0 (  9.0,   5.0)
 5. Spartacus 1.02            0       #     1           1 = 1     70%   3.5 (  8.0,   3.5)
 6. Pair-o-Max 4.8S                 0   #       1             1   67%   2.0 (  6.0,   2.0)
 7. Smirf BC-173g-X                       # 0 1         0 1 1     60%   3.0 (  6.0,   3.0)
 8. Joker80.np                  0   0 0   1 #   1                 40%   2.0 ( 16.5,   5.0)
 9. Max-Plus 4.8S               0         0   #   0 1 1           40%   2.0 ( 10.5,   2.0)
10. TSCPtest                  0         0   0   #       1   1     40%   2.0 (  9.0,   1.0)
11. Sjaak 524                     0           1   # 0     =       37%   1.5 (  7.0,   2.5)
12. NebiyuChess_1.43              0           0   1 #             33%   1.0 (  6.5,   1.5)
13. ArcBishop80 1.01 WB           0           0       #     1     33%   1.0 (  5.0,   0.0)
14. BigLion80 2.23x WB              0 0   1     0       #         25%   1.0 ( 12.5,   3.0)
15. ChessV_WinBoard             0     =   0       =       #       25%   1.0 ( 12.0,   2.5)
16. Heretic 0.3                 0     0   0     0     0     #      0%   0.0 ( 13.5,   0.0)
17. Fairy-Max 4.8S            0     0   0                     #    0%   0.0 ( 10.0,   0.0)

[hgm]

Yes, I will be careful. I discovered Max-Plus (and all Fairy-Max versions) are prone to '(near)infinite think' immediately after promotion. The reason is that promoted material is added to captured material to determine what the move gains. But this gain is then subtracted from total material. While the promoted stuff of course should be added (for the opponent, but Fairy-Max does not account that separately). So after a promotion it suddenly grossly underestimates remaining material (it can even go negative!), with as a result that it switches off null move, and then it will try to outdo hashed positions that were created with null move on, which without null move takes infinitely longer...

This was easy to fix, by adding two Queens to total material in the root in case of promotion, to compensate for the fact that the Queen had to be added, but was subtracted in stead. I will test that version tonight, and tomorrow it should have enough games to be sure it can be safely used as replacement.

[Daniel Shawul]

Nebiyu is pretty one dimensional and its weaknesses can be exploited with the right additions. Last tournament with spartan chess, spartacus used to wind lots of games throw wing play while Nebiyu always tries to break through the center. This is not something I planned for but a side effect of limited knowledge. All it knows is to centralize all pieces, including pawns, using PST tables. This is as a result of a choice I made to make it play non-chess variants with different pieces. But for the chess branch, pawns/pieces seem to have common properties so I can gradually add eval terms.
Btw 1.43 had bugs with castling that I fixed. It tries to castle when the king is under attack. So you can remove that version if you want to reduce the number of games. I see that 3 engines have two versions. Also I have a suggestion if you want something to challenge Bihasa. Nebiyu is multi-processor so you can let it use 2 processors for the later stages. I mean I can not be responsible for the lack of MP support for other variant engines

Daniel

[enhorning]

Another surprising game (at least to me - my general impression of ChessV is that it is fairly weak) - ChessV got a draw against Nebiyu 1.44:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.03.30"]
[Round "7"]
[White "NebiyuChess_1.44"]
[Black "ChessV_WinBoard"]
[Result "1/2-1/2"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rnabqkbcnr/pppppppppp/10/10/10/10/PPPPPPPPPP/RNABQKBCNR w KQkq - 0 1"]
[SetUp "1"]

1. Ad3 {+0.15/18} Nh6 {+0.29/10 29} 2. Nh3 {+0.05/18 29} e5 {+0.28/10 20}
3. Nc3 {+0.10/17 22} f5 {+0.23/9 28} 4. e4 {+0.20/16 22} fxe4 {+0.19/9 29}
5. Qxe4 {+0.20/15 29} Ad6 {+0.12/9 29} 6. Qf3+ {+0.80/16 25} Cf7
{+0.12/8 29} 7. Cg3 {+1.65/20 27} Cxf3 {+0.38/9 20} 8. Cxf3+ {+1.25/21 39}
Nf7 {+0.21/8 28} 9. Ne4 {+1.30/21 30} Ae7 {-0.43/11 22} 10. Neg5
{+1.25/20 25} g6 {-0.36/10 30} 11. Nxf7 {+1.25/20 27} Bxf7 {-0.44/10 30}
12. Cxe5 {+1.15/18 28} d6 {-0.45/9 30} 13. Ce3 {+1.35/16 29} Nd7
{-0.52/8 29} 14. f4 {+1.30/18 29} Nc5 {-0.51/9 26} 15. Ae2 {+1.25/18 25}
Ne6 {-0.46/9 25} 16. d4 {+1.30/16 24} Af5 {-0.28/9 30} 17. Cxf5
{+1.10/18 30} gxf5 {+0.08/10 30} 18. Ad3 {+1.25/19 30} Ng7 {-0.13/8 30} 19.
Bf3 {+1.25/20 30} c6 {-0.04/9 27} 20. Re1 {+1.15/19 30} Qd7 {+0.11/8 30}
21. a4 {+1.15/19 23} O-O {+0.14/8 30} 22. Be3 {+1.20/19 27} Bh5
{+0.25/8 30} 23. c3 {+1.35/20 27} Bxf3 {+0.75/9 29} 24. gxf3 {+1.35/17 8}
Bh4 {+0.75/9 27} 25. Re2 {+1.30/22 33} Ne8 {+0.74/8 30} 26. i3
{+1.60/20 33} Bf6 {+0.95/9 30} 27. Ke1 {+1.45/19 33} Nc7 {+1.14/9 30} 28.
Nj4 {+1.45/20 33} Nd5 {+1.25/10 29} 29. Nh5 {+1.35/22 29} Nxe3
{+1.14/10 30} 30. Nxf6 {+1.35/24 34} Qf7 {+1.15/10 25} 31. Rxe3
{+1.40/25 34} Qxf6 {+1.08/10 31} 32. Kd2 {+1.35/25 28} Rae8 {+1.04/9 32}
33. Rje1 {+1.30/24 32} Rxe3 {+1.09/10 32} 34. Rxe3 {+1.30/24 36} Rg8
{+1.04/10 29} 35. Re2 {+1.25/22 29} Qh6 {+1.04/8 33} 36. h4 {+1.20/23 30}
Qg6 {+0.99/9 33} 37. Ac4 {+1.40/21 33} d5 {+0.95/9 33} 38. Ae5
{+0.00/24 34} Qg1 {+0.97/10 33} 39. b3 {+0.00/24 48} Rg2 {+1.90/11 32} 40.
c4 {+0.00/28 53} Rxe2+ {+2.50/11 33} 41. Kxe2 {+0.00/28 22} Qh2+
{+2.50/11 29} 42. Ke3 {+0.00/27 29} Qc2 {+2.32/11 29} 43. Ag7+
{+0.00/27 29} Kj8 44. Af6 {+0.00/27 29} Qxb3+ {+2.32/10 29} 45. Kf2
{+0.00/29 29} Qc2+ {+2.32/11 30} 46. Kg1 {+0.00/28 29} Ki8 {+2.32/11 30}
47. Ag7+ {+0.00/31 29} Kj8 48. Af6 {+0.00/33 26} Ki8 {+2.32/10 31} 49. Ag7+
{+0.00/34 24} Kj8 50. Af6 {+0.00/35 29}
{XBoard adjudication: repetition draw} 1/2-1/2

Fairy-Max also got a draw against Joker in a long endgame, with passed rook-pawns for both sides - ended up being a draw by the 50-move rule after 157 moves.

Code: Select all

ResultSet-EloRating>ratings
Rank Name                  Elo    +    - games score oppo. draws
   1 Bihasa v3.2           326  223  223     5  100%    37    0%
   2 Bihasa v3.4.3         252  229  229     5  100%   -46    0%
   3 TJchess10x8 1.1-x64   153  191  191     6   83%   -51    0%
   4 NebiyuChess_1.44      145  171  171     7   79%   -21   14%
   5 Spartacus 1.02        108  183  183     6   75%   -34   17%
   6 Pair-o-Max 4.8S        17  199  199     5   40%    71    0%
   7 Joker80.np             -7  170  170     6   42%    46   17%
   8 TSCPtest              -12  191  191     6   50%   -12    0%
   9 BigLion80 2.23x WB    -29  174  174     6   42%    23   17%
  10 Smirf BC-173g-X       -46  186  186     6   50%   -47    0%
  11 Sjaak 524             -48  188  188     5   50%   -45   20%
  12 NebiyuChess_1.43      -54  203  203     5   40%     1    0%
  13 ChessV_WinBoard       -69  168  168     6   25%    50   50%
  14 Fairy-Max 4.8S        -73  186  186     5   20%    90   40%
  15 Max-Plus 4.8S        -117  182  182     6   33%   -20    0%
  16 ArcBishop80 1.01 WB  -235  203  203     5   20%   -55    0%
  17 Heretic 0.3          -310  229  229     6    0%    19    0%

[enhorning]

Standings before I head to bed - almost a quarter of the games from the Capablanca setup are done now, and the rank order is starting to look a bit more like expected:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              B B T N N S T S J P F B S M C A H
 1. Bihasa v3.2               #     1   1 1   1 1 1 1            100%   7.0 ( 27.0,  27.0)
 2. Bihasa v3.4.3               # 1   1 1 1 1         1     1    100%   7.0 ( 27.0,  27.0)
 3. TJchess10x8 1.1-x64         0 #   1       1       1 1 1 1 1   87%   7.0 ( 21.5,  14.5)
 4. NebiyuChess_1.44          0     #         1 1 1 1   1 =   1   81%   6.5 ( 22.5,  14.8)
 5. NebiyuChess_1.43            0 0   # 1   1         1 0   1     57%   4.0 ( 28.0,  12.0)
 6. Spartacus 1.02            0 0     0 #     1     1   1 =   1   56%   4.5 ( 27.5,   8.8)
 7. TSCPtest                  0 0         #   0 0   1   1 1   1   50%   4.0 ( 26.5,   6.0)
 8. Smirf BC-173g-X             0     0     # 0     0 1 1 1   1   50%   4.0 ( 23.0,   6.0)
 9. Joker80.np                0   0 0   0 1 1 # 1 =               44%   3.5 ( 39.0,  12.5)
10. Pair-o-Max 4.8S           0     0     1   0 # 1 0         1   43%   3.0 ( 26.5,   7.0)
11. Fairy-Max 4.8S            0     0         = 0 # =     1   1   43%   3.0 ( 24.0,   4.5)
12. BigLion80 2.23x WB        0     0   0 0 1   1 = #             36%   2.5 ( 32.0,   8.5)
13. Sjaak 524                   0 0   0     0         # 1 = 1     36%   2.5 ( 26.5,   3.8)
14. Max-Plus 4.8S                 0 0 1 0 0 0         0 #   1     25%   2.0 ( 33.5,   5.0)
15. ChessV_WinBoard               0 =   = 0 0     0   =   #       21%   1.5 ( 31.5,   6.8)
16. ArcBishop80 1.01 WB         0 0   0               0 0   # 1   17%   1.0 ( 22.5,   0.0)
17. Heretic 0.3                   0 0   0 0 0   0 0         0 #    0%   0.0 ( 33.0,   0.0)

[enhorning]

Poor Pair-o-Max got mauled by Bihasa in this miniature game:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2013.03.30"]
[Round "11"]
[White "Pair-o-Max 4.8S"]
[Black "Bihasa v3.4.3"]
[Result "0-1"]
[TimeControl "40/1200"]
[Variant "capablanca"]
[FEN "rnabqkbcnr/pppppppppp/10/10/10/10/PPPPPPPPPP/RNABQKBCNR w KQkq - 0 1"]
[SetUp "1"]

{--------------
r n a b q k b c n r
p p p p p p p p p p
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
P P P P P P P P P P
R N A B Q K B C N R
white to play
--------------}
1. Nh3 {+0.10/9} d6 {-0.24/16 49} 2. Ci3 {+0.00/8 13} Nh6 {+0.28/17 54} 3.
Nc3 {+0.22/9 27} e5 {+0.32/16 30} 4. Cxi7 {-0.06/10 28} Ri8 {+0.60/18 37}
5. Ch5 {-0.06/10 18} f5 {+0.40/17 51} 6. Cg3 {+0.38/9 16} f4 {+0.96/15 27}
7. Cf3 {+0.07/9 28} Nf5 {+2.44/14 30} 8. i3 {+0.54/9 34} Nd4 {+5.72/18 39}
9. Cd3 {-2.34/10 17} Bc4 {+5.76/16 32} 10. Cb4 {-2.89/10 19} f3
{+9.90/14 29} 11. Ni1 {-4.12/9 1:24} Qi4 {+20.36/16 31} 12. d3 {-0.85/9 36}
fxg2+ {+299.91/16 26} 13. Nxg2 {-79.96/10 15} Ah3 {+299.93/16 26} 14. f4
{-79.97/16 18} Axg2+ {+299.95/4 0.1} 15. Kf2 {-79.98/28 0.1} Ah3+
{+299.97/4 0.1} 16. Kg3 {-79.99/28 0.1} Cg6# {+299.99/4 0.1}
{Xboard adjudication: Checkmate} 0-1