Chess variant tournament: Makruk (without endgame counting)

[Ferdy]

No he does not! Ferd you have got to loose for once

I am fixing the knq-k ending, as well as kqqq-kp and similar stuff where all q has same colors. Also trying to add your egbb even for the rook and knights ending .

[musketeerchess]

Hi
is it possible to implment my chess variant. MusketeerChess

It's available on musketeerchess.com

i already programmed it at zillions, but it does not support autoplay.

[enhorning]

Standings after 3 cycles (90 games out of 150):

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Neb Bil HaC Fai Sja Fai
 1. NebiyuChess_1.45          ### 01= 1=1 111 1=1 111
                              ### 011 1== =1= 11= 11=   78%  23.5 (399.0, 296.3)

 2. Bilis v0.14               10= ### 111 1=1 111 111
                              100 ### 1== 11= =1= 111   78%  23.5 (399.0, 283.8)

 3. HaChu 0.17                0=0 000 ### ==0 10= ==0
                              0== 0== ### =1= 101 ==1   40%  12.0 (468.0, 157.0)

 4. Fairy-Max 4.8S            000 0=0 ==1 ### 0=0 ===
                              =0= 00= =0= ### =1= ===   35%  10.5 (477.0, 139.3)

 5. Sjaak 524                 0=0 000 01= 1=1 ### 10=
                              00= =0= 010 =0= ### 0==   35%  10.5 (477.0, 138.8)

 6. Fairy-Max 4.8S+           000 000 ==1 === 01= ###
                              00= 000 ==0 === 1== ###   33%  10.0 (480.0, 116.0)

Nebiyu and Bilis are tied, both at 23.5 points!

PGN with games at: https://www.dropbox.com/s/l0o3e8ob8b6rp7v/makruk.pgn

[enhorning]

Standings after 107 / 150 games:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Bili Nebi HaCh Fair Sjaa Fair
 1. Bilis v0.14               #### 10=1 1111 1111 111= 1=11
                              #### 0010 ==   111= 1=11 1=     78%  28.0 (579.0, 410.0)

 2. NebiyuChess_1.45          01=0 #### 1=11 1111 1=11 111=
                              1101 #### ==== 1=   1=   1=11   76%  27.5 (586.0, 421.0)

 3. HaChu 0.17                0000 0=00 #### ==0= 10=1 ==0=
                              ==   ==== #### =10= 01   1===   40%  14.5 (659.0, 232.3)

 4. Fairy-Max 4.8S+           0000 0000 ==1= #### 01=1 ====
                              000= 0=   =01= #### ===  ==     36%  12.5 (664.5, 179.0)

 5. Sjaak 524                 000= 0=00 01=0 10=0 #### 1=1=
                              0=00 0=   10   ===  #### 0==1   36%  12.5 (659.5, 189.3)

 6. Fairy-Max 4.8S            0=00 000= ==1= ==== 0=0= ####
                              0=   0=00 0=== ==   1==0 ####   33%  12.0 (679.0, 188.5)

This is the point at which Bilis got upgraded from v0.14 to v1.0.

PGN with games at : https://www.dropbox.com/s/l0o3e8ob8b6rp7v/makruk.pgn

[enhorning]

Standings after 4 cycles (120 out of 150 games):

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Bili Bili Nebi HaCh Sjaa Fair Fair
 1. Bilis v1.0                ####           11             11
                              ####                                100%   4.0 ( 56.0,  56.0)

 2. Bilis v0.14                    #### 10=1 1111 111= 1111 1=11
                                   #### 0010 ==   1=11 111= 1=     78%  28.0 (648.0, 458.8)

 3. NebiyuChess_1.45               01=0 #### 1=11 1=11 1111 111=
                                   1101 #### ==== 1=11 1=1= 1=11   77%  31.0 (680.0, 502.5)

 4. HaChu 0.17                00   0000 0=00 #### 10=1 ==0= ==0=
                                   ==   ==== #### 010= =10= 1===   37%  15.0 (760.0, 266.3)

 5. Sjaak 524                      000= 0=00 01=0 #### 10=0 1=1=
                                   0=00 0=00 101= #### ==== 0==1   36%  14.5 (812.0, 234.8)

 6. Fairy-Max 4.8S+                0000 0000 ==1= 01=1 #### ====
                                   000= 0=0= =01= ==== #### ====   36%  14.5 (812.0, 229.8)

 7. Fairy-Max 4.8S            00   0=00 000= ==1= 0=0= ==== ####
                                   0=   0=00 0=== 1==0 ==== ####   32%  13.0 (776.0, 220.5)

So, Bilis is just one point ahead of Nebiyu.

PGN file is at : https://www.dropbox.com/s/l0o3e8ob8b6rp7v/makruk.pgn

[enhorning]

It's finished.

Final crosstable:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Bilis Nebiy Bilis HaChu Sjaak Fairy Fairy
 1. Bilis v1.0                ##### =1          111=  11    11=1  11
                              #####                                       89%  12.5 (292.0, 254.0)

 2. NebiyuChess_1.45          =0    ##### 01=01 1=11= 1=111 111=1 11111
                                    ##### 101   ===11 =1111 =1111 =1=11   79%  39.5 (949.0, 733.0)

 3. Bilis v0.14                     10=10 ##### 1111= 111=1 1=111 11111
                                    010   ##### =     =11   =     11=     78%  28.0 (803.0, 563.3)

 4. HaChu 0.17                000=  0=00= 0000= ##### 10=10 ==0=1 ==0==
                                    ===00 =     ##### 10=1= ===== 10=1=   39%  19.5 (1118.0, 394.3)

 5. Sjaak 524                 00    0=000 000=0 01=01 ##### 1=1=0 10=0=
                                    =0000 =00   01=0= ##### ==101 ===1=   35%  17.5 (1169.0, 337.3)

 6. Fairy-Max 4.8S            00=0  000=0 0=000 ==1=0 0=0=1 ##### =====
                                    =0000 =     ===== ==010 ##### ====1   34%  17.0 (1143.0, 329.3)

 7. Fairy-Max 4.8S+           00    00000 00000 ==1== 01=1= ===== #####
                                    =0=00 00=   01=0= ===0= ====0 #####   32%  16.0 (1184.0, 315.0)

... or with the two version of Bilis joined into one entry:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Bilis Nebiy HaChu Sjaak Fairy Fairy
 1. Bilis v0.14 / v1.0        ##### 10=10 1111= 111=1 1=111 11111
                              ##### 010=1 =111= =1111 =11=1 11=11   81%  40.5 (1095.0, 817.3)

 2. NebiyuChess_1.45          01=01 ##### 1=11= 1=111 111=1 11111
                              101=0 ##### ===11 =1111 =1111 =1=11   79%  39.5 (1105.0, 803.3)

 3. HaChu 0.17                0000= 0=00= ##### 10=10 ==0=1 ==0==
                              =000= ===00 ##### 10=1= ===== 10=1=   39%  19.5 (1305.0, 420.8)

 4. Sjaak 524                 000=0 0=000 01=01 ##### 1=1=0 10=0=
                              =0000 =0000 01=0= ##### ==101 ===1=   35%  17.5 (1325.0, 349.8)

 5. Fairy-Max 4.8S            0=000 000=0 ==1=0 0=0=1 ##### =====
                              =00=0 =0000 ===== ==010 ##### ====1   34%  17.0 (1330.0, 355.8)

 6. Fairy-Max 4.8S+           00000 00000 ==1== 01=1= ===== #####
                              00=00 =0=00 01=0= ===0= ====0 #####   32%  16.0 (1340.0, 321.3)

A close race between Bilis and Nebiyu, all the way until the end!

I'm surprised we only saw two draws between Nebiyu and Bilis - I would have expected more draws between the top two programs.
Bayeselo gives:

Code: Select all

ResultSet-EloRating>ratings
Rank Name               Elo    +    - games score oppo. draws
   1 Bilis v1.0         197  129  129    14   89%   -77   21%
   2 NebiyuChess_1.45   138   64   64    50   79%   -64   26%
   3 Bilis v0.14        131   75   75    36   78%   -61   22%
   4 HaChu 0.17         -87   59   59    50   39%   -17   50%
   5 Sjaak 524         -120   62   62    50   35%   -13   34%
   6 Fairy-Max 4.8S    -120   59   59    50   34%   -10   52%
   7 Fairy-Max 4.8S+   -139   60   60    50   32%    -9   48%

PGN with games is at : https://www.dropbox.com/s/l0o3e8ob8b6rp7v/makruk.pgn

Suggestions for possible variants to run in the future are welcome... is it time to revisit some of the ones I ran tournaments for back in spring?

[Daniel Shawul]

That was a close one! Nebiyu played better than I expected. Bilis probably could have destroyed the field with the newer version but still it was a good match. For your next variants I suggest minishogi or judkins shogi., which are exciting variants with no draws. I don't know how many engines you will get but there are some from chess programmers alone. The programming is also distinctly different because of drops. If not those variants, I suggest crazyhouse or any other variant with piece drops..

[Ferdy]

Nebiyu's speed is swift, I may try one day to remove eval mobility and see how it fares with this engine.
Reviewing the games, it seems like pushing the opponent's king in the corner or edge of the board can help a lot in this variant.

I hope to improve the crazyhouse by using bitboard, calculation of attack and defense especially with those drops
are expensive with current version. I also started working on shogi 9x9 but is not done yet.

Future tour could be xiangqi, shogi, and grand.

Thanks again for running these chess variants.

[Ferdy]

Hi
is it possible to implment my chess variant. MusketeerChess

It's available on musketeerchess.com

i already programmed it at zillions, but it does not support autoplay.

Once HGM will implement this in winboard, I will probably make one, I already have a seirawan chess, this could not be difficult to adapt.
Perhaps we can make a sample correspondence game with this variant once I am done before winboard implementation.

[enhorning]

For your next variants I suggest minishogi or judkins shogi., which are exciting variants with no draws.

Isn't repetition a draw? At least, that has been the rule when I've played minishogi on turn-based servers.

I'd quite like to run Judkins Shogi - what programs are there that support that through winboard?