Xiangi Clash 2010 (live broadcast)

[hgm]

Code: Select all

Bachelors Division cross table, sorted by score percentage, Buchholz, SB

                              Ny Ul Tl Hi Co Ju An Sa
 1. NymphChess 0.159          ## 1= 11 10 10 01 10 11   68%   9.5 ( 93.0,  64.3)
 2. Ultimate Fight 0.26.22    0= ## 1= 00 11 11 1= 11   68%   9.5 ( 93.0,  59.3)
 3. Tlxj 0.3                  00 0= ## 11 00 11 11 10   54%   7.5 ( 97.0,  46.3)
 4. Hice 0.7.5                01 11 00 ## 01 =0 =1 =0   46%   6.5 ( 99.0,  48.5)
 5. Coony 1.05                01 00 11 10 ## 0= =1 00   43%   6.0 (100.0,  42.3)
 6. Jupiter 0.91              10 00 00 =1 1= ## 01 01   43%   6.0 (100.0,  39.3)
 7. Anita 0.2                 01 0= 00 =0 =0 10 ## 11   39%   5.5 (101.0,  37.5)
 8. Saola 1.7w                00 00 01 =1 11 10 00 ##   39%   5.5 (101.0,  35.3)

This division was plagued by time losses. Part of this was my fault. I had replaced WinBoard for a newer version to adapt to a bug of ChessV that could not be fixed in time, but I had forgotten that this newer version also does not count checks and check-evasions in the 50-move count. Now it turns out that many XQ engines simply stop after 50 non-captures (and then lose on time). Of course when no Pawn pushes are made, this would really be a 50-move draw. If Pawn pushes had been made, the time loss is justified. But this new WB sometimes did noot grant a draw in the case without Pawn pushes, because it was not counting the checks.

Now the latter was in violation of the rules I had specified before the tourney. So I have corrected these results by hand to 50-move draws. The above cross table shows these corrected results.

[hgm]

Code: Select all

Masters Disivision cross table, sorted by score percentage, Buchholz, SB

                              Bu El Cy Ey Ha Bl Sw YS
 1. BugChess 0.1d5            ## 11 1= 11 == 11 11 10   82%  11.5 ( 89.0,  74.8)
 2. Elephant Eye 3.15         00 ## 11 0= 1= =1 11 10   61%   8.5 ( 95.0,  50.5)
 3. Cyclone 1.01              0= 00 ## 1= 11 10 1= 11   61%   8.5 ( 95.0,  49.5)
 4. EyChessU 1.883e           00 1= 0= ## 1= =0 11 11   57%   8.0 ( 96.0,  45.5)
 5. HaQiKi D 1.7dh            == 0= 00 0= ## 10 10 11   43%   6.0 (100.0,  37.3)
 6. Blcx 0.56                 00 =0 01 =1 01 ## 00 10   36%   5.0 (102.0,  34.8)
 7. Swallow 0.9.1             00 00 0= 00 01 11 ## 10   32%   4.5 (103.0,  24.3)
 8. YSSY                      01 01 00 00 00 01 01 ##   29%   4.0 (104.0,  29.5)

YSSY is really losing on time to much. It is a very strong engine, that can beat any opponent. But it just takes a little bit too long for its 40th or 80th move...

[hgm]

Binghewusi 2.44 is the undoubted champion of this first edition of The Xiangqi Clash! It left its competitors ahead by a wide margin:

Code: Select all

Play-Offs cross table, sorted by score percentage, Buchholz, SB

                              Bing Kou  3DCh XQ S
 1. Binghewusi 2.44           #### 111= 11== 1111
                              #### 11=0 00=0 =111   71%  17.0 (248.0, 169.0)

 2. Kou 0.1.3                 000= #### 1110 101=
                              00=1 #### =100 1111   54%  13.0 (280.0, 129.0)

 3. 3DChess 0.92              00== 0001 #### 1=00
                              11=1 =011 #### 100=   46%  11.0 (296.0, 143.0)

 4. XQ Spirit 2010.05.13      0000 010= 0=11 ####
                              =000 0000 011= ####   29%   7.0 (328.0,  83.0)

Running all games through BayesElo, with offset 2000, produces the following rating list:

Code: Select all

   1 Binghewusi 2.44         2575   98   91    38   70%  2432   24%
   2 BugChess 0.1d5          2512  117  107    28   71%  2360   21%
   3 Kou 0.1.3               2504   95   93    38   57%  2451   18%
   4 3DChess 0.92            2482   96   95    38   53%  2457   16%
   5 Cyclone 1.01            2403  110  108    28   54%  2375   21%
   6 XQ Spirit 2010.05.13    2393   92   94    38   45%  2434   21%
   7 EyChessU 1.883e         2384  107  106    28   52%  2378   25%
   8 Elephant Eye 3.15       2360  105  102    28   57%  2319   29%
   9 HaQiKi D 1.7dh          2344  105  104    28   52%  2321   25%
  10 Swallow 0.9.1           2314  106  107    28   48%  2325   18%
  11 Ultimate Fight 0.26.22  2289  108  106    28   55%  2247   18%
  12 NymphChess 0.159        2263  116  114    28   52%  2250    4%
  13 Blcx 0.56               2242  117  116    28   55%  2195   18%
  14 Tlxj 0.3                2242  105  106    28   46%  2253   21%
  15 YSSY                    2223  119  135    28   29%  2401    0%
  16 Thougth 1.1             2196  133  130    28   63%  2052    4%
  17 Saola 1.7w              2195  112  121    28   32%  2323   14%
  18 Hice 0.7.5              2183  115  112    28   57%  2121   14%
  19 Coony 1.05              2160  112  111    28   54%  2124   14%
  20 Jupiter 0.91            2150  112  113    28   52%  2126   11%
  21 Anita 0.2               2150  110  110    28   52%  2126   18%
  22 XQWLight                1920  152  138    30   78%  1518    3%
  23 MRSD [9]                1878  122  128    28   41%  1957    4%
  24 Jiaolong 1.0            1865  118  125    28   39%  1959    7%
  25 Silent Thinker 0.02     1850  143  137    30   72%  1528    3%
  26 Folium                  1803  123  132    28   30%  1968   11%
  27 TJxiangqi 0.03 (+book)  1736  130  134    30   60%  1543   13%
  28 MaxQi 4.8H              1476  144  169    30   35%  1578    3%
  29 ChessV 0.9.4x           1290  146  130    40   71%  1073    3%
  30 HoiXiangqi 10.3         1230  140  128    40   65%  1085    0%
  31 VSCCP [5]                953  118  119    40   38%  1140   10%
  32 Tsito 0.8.4w             733  128  107    40   14%  1184    8%
  33 Qianhong [3]             704  136   84    40   13%  1190    0%

[TonyJH]

It's interesting to see the BayesElo output. Nice result for HaQiKi D.
How did you make TJxiangqi use a book? (If that's what +book means.)

[hgm]

I used the WinBoard GUI book. The book was the XQWLight book (converted to Polyglot format by me, and included in the WB install), which was also used by all other engines that did not have an own book. (Except ChessV, which, due to a trivial bug could not be used in force mode.)

For TJxiangqi I mentioned it in the name, because I first accidentally played it without book, and needed a different name to remove the games later from the . This really makes an enormous difference, just like for my own engine. I am afraid our engines do not understand much of XQ opening play...