Statistics to playing phases ... Komodo / Zahak and the others!

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

Frank Quisinsky
Posts: 7285
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Statistics to playing phases ... Komodo / Zahak and the others!

Post by Frank Quisinsky »

Hi there,

a bit statistic on Christmas to TOP-41 with 1:20h games on 4.4Ghz.
Elo not important ... more interesting are the differents betweens the playing phases!
Please not looking on Elo ... looking on places in the table!!

1.
Games ended between 20-59 moves ...
Resign = off games ...

Code: Select all

   # Player                                :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 151121 NN x64               :  3416.31    190    80.3  115    75     0   152.5   39.5  36.49  3154.93  23.30   35.7
   2 Dragon 2.5.1 by Komodo NN AVX2 x64    :  3389.26    155    79.0   90    65     0   122.5   41.9  39.79  3143.56  23.78   33.8
   3 Ethereal 13.25 NN PEXT x64            :  3287.21    154    64.6   48   103     3    99.5   66.9  26.75  3166.50  23.43   34.7
   4 SlowChess Blitz 2.8 NN AVX2 x64       :  3278.07    139    64.4   41    97     1    89.5   69.8  27.89  3161.03  22.79   35.5
   5 Fire 8 NN MC.3 x64                    :  3277.00    153    63.7   42   111     0    97.5   72.5  26.00  3176.05  23.14   32.0
   6 Revenge 1.0 NN x64                    :  3271.19    224    63.8   63   160     1   143.0   71.4  22.65  3164.96  23.54   36.7
   7 Koivisto 7.5 NN AVX2 x64              :  3258.12     97    59.3   25    65     7    57.5   67.0  30.74  3177.97  23.49   30.6
   8 Berserk 7 NN PEXT x64                 :  3253.50    133    59.0   29    99     5    78.5   74.4  24.68  3178.17  23.68   34.5
   9 RubiChess 2.2 NN x64                  :  3238.41    208    58.9   41   163     4   122.5   78.4  20.29  3173.73  23.13   34.6
  10 Arasan 23.0.1 NN AVX2 x64             :  3231.24    217    59.7   43   173     1   129.5   79.7  20.49  3159.37  23.20   37.6
  11 Igel 3.0.10 NN BMI2 x64               :  3217.96    290    57.4   46   241     3   166.5   83.1  17.30  3165.69  23.34   37.5
  12 rofChade 2.310 NN x64                 :  3217.33    157    56.4   28   121     8    88.5   77.1  22.31  3171.54  23.12   34.4
  13 Booot 6.5 POPCNT x64                  :  3207.27    218    56.0   38   168    12   122.0   77.1  19.98  3166.63  23.35   36.0
  14 Nemorino 6.09 NN x64                  :  3199.65    197    52.5   22   163    12   103.5   82.7  19.97  3175.89  23.39   35.7
  15 Seer 2.4.0 NN AVX2 x64                :  3196.40    190    51.1   17   160    13    97.0   84.2  20.03  3180.14  23.49   36.8
  16 Schooner 2.2 XB SSE x64               :  3194.68    170    55.6   27   135     8    94.5   79.4  21.34  3157.88  22.76   32.5
  17 Wasp 5.00 NN AVX2 x64                 :  3188.05    202    53.0   20   174     8   107.0   86.1  20.07  3170.54  23.45   37.0
  18 Weiss 2.0 PEXT x64                    :  3181.33    154    52.6   21   120    13    81.0   77.9  22.71  3165.18  23.36   32.1
  19 Xiphos 0.6 BMI2 x64                   :  3180.50    195    52.6   22   161    12   102.5   82.6  20.54  3159.15  23.09   35.9
  20 Laser 1.7 BMI2 x64                    :  3161.42    167    51.2   17   137    13    85.5   82.0  20.00  3154.69  23.47   34.6
  21 Clover 2.4 x64                        :  3158.54    176    49.4   14   146    16    87.0   83.0  21.54  3167.33  22.84   33.8
  22 Defenchess 2.3 dev BMI2 x64           :  3154.56    136    50.4   11   115    10    68.5   84.6  23.08  3153.74  22.99   33.3
  23 Fizbo 2.0 BMI2 x64                    :  3142.13    194    47.2   15   153    26    91.5   78.9  20.75  3159.39  23.26   34.8
  24 Fritz 18 (Ginkgo) x64                 :  3137.52    146    46.9   18   101    27    68.5   69.2  22.09  3160.91  22.94   35.1
  25 DanaSah 9.0 NN AVX2 x64               :  3137.35    187    45.2   14   141    32    84.5   75.4  20.44  3181.87  23.27   35.2
  26 Andscacs 0.95.123 x64                 :  3137.30    176    47.2   13   140    23    83.0   79.5  20.46  3161.29  23.19   36.9
  27 Minic 3.17 NN x64                     :  3128.56    186    43.0   13   134    39    80.0   72.0  21.56  3179.97  23.25   35.3
  28 Chiron 5 x64                          :  3127.95    179    45.0   11   139    29    80.5   77.7  21.30  3167.84  22.71   34.9
  29 Dark Toga 1.1 NN AVX2 x64             :  3125.79    128    46.1    9   100    19    59.0   78.1  24.44  3156.44  22.87   33.6
  30 Lc0 0.28.0 NN CPU-dnnl                :  3123.74    216    41.9   21   139    56    90.5   64.4  20.82  3189.40  23.80   36.7
  31 Shredder 13 POPCNT x64                :  3122.18    178    43.8   10   136    32    78.0   76.4  21.20  3165.08  23.01   36.5
  32 Marvin 5.2.0 NN AVX2 x64              :  3120.11    184    45.1    8   150    26    83.0   81.5  20.24  3163.06  22.99   35.5
  33 Zahak 8.6 AMD x64                     :  3095.78    220    39.1    9   154    57    86.0   70.0  20.90  3179.49  23.10   36.7
  34 Nirvanachess 2.5 POPCNT x64           :  3079.13    141    35.8    5    91    45    50.5   64.5  28.39  3186.57  23.52   33.3
  35 Halogen 10 NN PEXT x64                :  3072.51    196    36.0    5   131    60    70.5   66.8  22.85  3189.19  24.06   35.7
  36 Orion 0.8 NN FMA x64                  :  3069.94    222    34.9    4   147    71    77.5   66.2  21.91  3190.90  23.43   36.3
  37 Stash 31.16 x64                       :  3062.06    153    36.3    5   101    47    55.5   66.0  26.39  3171.26  23.15   35.0
  38 GullChess 3.0 Sy BMI2 x64             :  3057.67    156    32.7    9    84    63    51.0   53.8  28.33  3191.77  23.68   33.6
  39 Winter 0.9 BMI2 x64                   :  3056.63    205    33.9    8   123    74    69.5   60.0  24.23  3177.76  23.51   36.3
  40 Demolito 2021-07-09 x64               :  3032.48    114    28.5    3    59    52    32.5   51.8  36.94  3196.49  23.54   31.8
  41 Combusken 1.4.0 AMD x64               :  3024.16    181    30.1    7    95    79    54.5   52.5  29.37  3188.23  23.33   35.9

White advantage = 47.27 +/- 2.77
Draw rate (equal opponents) = 96.49 % +/- 1.75
New for me is the booster Ethereal got.
Seems that v13.25 can play with more aggressivess as versions before.
That is great!!

2.
Games ended between 60-79 moves ...
Resign = off games ...

Code: Select all

   # Player                                :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 151121 NN x64               :  3584.74    391    90.7  318    73     0   354.5   18.7  41.65  3135.55  30.81   38.2
   2 Dragon 2.5.1 by Komodo NN AVX2 x64    :  3550.42    352    89.6  280    71     1   315.5   20.2  40.19  3131.98  30.78   38.0
   3 Fire 8 NN MC.3 x64                    :  3411.72    343    78.3  205   127    11   268.5   37.0  32.69  3147.04  31.13   37.8
   4 SlowChess Blitz 2.8 NN AVX2 x64       :  3408.81    345    79.4  214   120    11   274.0   34.8  32.67  3134.20  30.91   37.9
   5 Berserk 7 NN PEXT x64                 :  3394.82    258    75.6  147    96    15   195.0   37.2  36.14  3149.66  30.99   36.0
   6 Ethereal 13.25 NN PEXT x64            :  3384.19    251    74.7  142    91    18   187.5   36.3  34.84  3155.28  31.18   37.7
   7 Koivisto 7.5 NN AVX2 x64              :  3378.06    238    73.9  134    84    20   176.0   35.3  38.24  3148.42  31.34   36.0
   8 RubiChess 2.2 NN x64                  :  3329.06    303    70.6  147   134    22   214.0   44.2  29.57  3152.57  31.16   38.4
   9 Revenge 1.0 NN x64                    :  3327.12    312    71.3  148   149    15   222.5   47.8  30.99  3147.14  30.87   38.2
  10 Igel 3.0.10 NN BMI2 x64               :  3285.57    273    65.4  119   119    35   178.5   43.6  31.68  3159.13  31.20   37.6
  11 rofChade 2.310 NN x64                 :  3281.90    271    64.8  121   109    41   175.5   40.2  32.80  3156.59  31.01   37.6
  12 Seer 2.4.0 NN AVX2 x64                :  3244.86    253    60.9   90   128    35   154.0   50.6  31.82  3159.18  31.18   37.8
  13 Booot 6.5 POPCNT x64                  :  3244.09    269    62.8  102   134    33   169.0   49.8  30.77  3145.94  31.05   37.0
  14 Arasan 23.0.1 NN AVX2 x64             :  3240.80    300    61.0  109   148    43   183.0   49.3  29.37  3159.74  31.20   38.1
  15 Nemorino 6.09 NN x64                  :  3237.81    302    56.3   98   144    60   170.0   47.7  28.62  3177.14  31.30   38.1
  16 Lc0 0.28.0 NN CPU-dnnl                :  3195.40    264    54.9   92   106    66   145.0   40.2  31.63  3164.32  31.38   38.1
  17 Wasp 5.00 NN AVX2 x64                 :  3154.95    291    51.0   77   143    71   148.5   49.1  27.45  3160.04  31.23   37.5
  18 Weiss 2.0 PEXT x64                    :  3152.84    268    49.3   75   114    79   132.0   42.5  30.59  3175.97  31.48   36.3
  19 Schooner 2.2 XB SSE x64               :  3149.73    304    49.2   66   167    71   149.5   54.9  28.56  3163.69  31.11   37.2
  20 Xiphos 0.6 BMI2 x64                   :  3148.41    260    48.8   63   128    69   127.0   49.2  31.01  3180.32  31.18   37.5
  21 Fritz 18 (Ginkgo) x64                 :  3128.15    272    46.1   74   103    95   125.5   37.9  30.37  3169.55  31.53   37.0
  22 Minic 3.17 NN x64                     :  3126.25    280    42.7   59   121   100   119.5   43.2  29.78  3192.56  31.70   37.8
  23 Defenchess 2.3 dev BMI2 x64           :  3111.50    281    41.1   59   113   109   115.5   40.2  29.14  3184.92  31.27   37.1
  24 Laser 1.7 BMI2 x64                    :  3099.35    289    43.1   54   141    94   124.5   48.8  28.50  3167.38  31.11   38.4
  25 Clover 2.4 x64                        :  3088.29    286    37.1   50   112   124   106.0   39.2  30.11  3194.18  31.69   37.0
  26 Shredder 13 POPCNT x64                :  3086.28    313    40.4   58   137   118   126.5   43.8  28.59  3176.17  31.35   38.2
  27 Marvin 5.2.0 NN AVX2 x64              :  3081.80    300    38.5   57   117   126   115.5   39.0  28.84  3190.00  31.50   37.2
  28 Andscacs 0.95.123 x64                 :  3076.77    280    38.6   50   116   114   108.0   41.4  30.14  3176.88  31.43   37.3
  29 DanaSah 9.0 NN AVX2 x64               :  3076.11    336    39.6   61   144   131   133.0   42.9  26.79  3177.89  31.55   37.9
  30 Chiron 5 x64                          :  3074.01    269    38.8   48   113   108   104.5   42.0  30.52  3169.93  31.17   36.2
  31 Fizbo 2.0 BMI2 x64                    :  3070.03    332    38.3   52   150   130   127.0   45.2  26.97  3183.07  31.19   37.5
  32 Zahak 8.6 AMD x64                     :  3062.63    276    34.6   42   107   127    95.5   38.8  30.79  3207.27  31.87   37.3
  33 Dark Toga 1.1 NN AVX2 x64             :  3040.61    274    30.7   41    86   147    84.0   31.4  32.38  3212.81  31.93   36.0
  34 Halogen 10 NN PEXT x64                :  3016.60    338    30.8   34   140   164   104.0   41.4  28.82  3188.68  31.28   37.8
  35 Stash 31.16 x64                       :  3012.30    260    29.4   30    93   137    76.5   35.8  31.96  3202.19  31.46   36.6
  36 Demolito 2021-07-09 x64               :  2997.01    256    28.9   37    74   145    74.0   28.9  31.86  3189.55  31.52   37.7
  37 Winter 0.9 BMI2 x64                   :  2988.93    312    28.5   40    98   174    89.0   31.4  30.81  3188.82  31.47   38.5
  38 Nirvanachess 2.5 POPCNT x64           :  2966.63    330    26.8   25   127   178    88.5   38.5  30.23  3181.90  31.28   38.5
  39 GullChess 3.0 Sy BMI2 x64             :  2960.98    336    23.8   31    98   207    80.0   29.2  30.14  3209.28  31.70   37.3
  40 Orion 0.8 NN FMA x64                  :  2954.54    326    24.5   29   102   195    80.0   31.3  30.20  3193.76  31.52   38.5
  41 Combusken 1.4.0 AMD x64               :  2886.92    360    16.8   16    89   255    60.5   24.7  33.62  3207.28  31.51   37.9

White advantage = 90.95 +/- 3.70
Draw rate (equal opponents) = 60.74 % +/- 0.97
For me all looks normally ...

3.
Games ended between 80-99 moves ...
Resign = off games ...

Code: Select all

   # Player                                :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Dragon 2.5.1 by Komodo NN AVX2 x64    :  3589.38    268    89.6  214    52     2   240.0   19.4  51.60  3154.39  35.15   37.7
   2 Stockfish 151121 NN x64               :  3527.10    244    86.7  180    63     1   211.5   25.8  46.48  3145.69  35.18   37.4
   3 Fire 8 NN MC.3 x64                    :  3468.07    273    83.0  189    75     9   226.5   27.5  42.49  3146.15  35.30   36.8
   4 SlowChess Blitz 2.8 NN AVX2 x64       :  3428.35    242    78.7  149    83    10   190.5   34.3  41.30  3156.94  35.66   37.7
   5 Koivisto 7.5 NN AVX2 x64              :  3424.85    289    79.9  184    94    11   231.0   32.5  37.60  3146.72  35.17   37.9
   6 Ethereal 13.25 NN PEXT x64            :  3417.81    275    78.2  175    80    20   215.0   29.1  38.38  3156.03  35.48   37.4
   7 Berserk 7 NN PEXT x64                 :  3383.20    276    77.2  165    96    15   213.0   34.8  36.30  3141.94  35.41   37.5
   8 RubiChess 2.2 NN x64                  :  3334.38    244    69.1  122    93    29   168.5   38.1  37.65  3153.57  35.72   36.8
   9 Seer 2.4.0 NN AVX2 x64                :  3331.85    283    71.2  154    95    34   201.5   33.6  34.20  3152.17  35.69   38.2
  10 Nemorino 6.09 NN x64                  :  3295.80    248    65.7  117    92    39   163.0   37.1  35.76  3165.11  35.63   36.1
  11 Igel 3.0.10 NN BMI2 x64               :  3290.14    230    65.0  106    87    37   149.5   37.8  37.41  3165.10  35.75   37.5
  12 Revenge 1.0 NN x64                    :  3284.54    249    64.1  106   107    36   159.5   43.0  34.71  3168.25  35.86   37.3
  13 rofChade 2.310 NN x64                 :  3268.15    280    63.4  121   113    46   177.5   40.4  34.29  3153.27  35.87   38.0
  14 Lc0 0.28.0 NN CPU-dnnl                :  3251.42    239    64.9  111    88    40   155.0   36.8  32.74  3140.94  35.27   37.6
  15 Arasan 23.0.1 NN AVX2 x64             :  3190.43    235    51.3   72    97    66   120.5   41.3  33.94  3174.68  35.75   37.3
  16 Booot 6.5 POPCNT x64                  :  3180.66    248    49.4   75    95    78   122.5   38.3  33.61  3180.53  35.97   36.8
  17 Minic 3.17 NN x64                     :  3149.97    248    47.8   65   107    76   118.5   43.1  31.63  3173.44  35.76   38.3
  18 Xiphos 0.6 BMI2 x64                   :  3139.79    278    47.3   72   119    87   131.5   42.8  30.35  3176.19  35.95   37.2
  19 Wasp 5.00 NN AVX2 x64                 :  3134.14    260    41.0   56   101   103   106.5   38.8  34.45  3207.23  36.08   36.5
  20 Weiss 2.0 PEXT x64                    :  3131.20    257    42.2   52   113    92   108.5   44.0  33.54  3193.81  35.94   36.4
  21 Zahak 8.6 AMD x64                     :  3109.39    233    44.4   62    83    88   103.5   35.6  34.13  3167.37  35.38   37.1
  22 Shredder 13 POPCNT x64                :  3104.55    247    43.5   57   101    89   107.5   40.9  33.59  3170.67  35.88   37.2
  23 Marvin 5.2.0 NN AVX2 x64              :  3089.44    267    39.9   56   101   110   106.5   37.8  32.59  3179.61  36.02   38.4
  24 Fritz 18 (Ginkgo) x64                 :  3089.36    269    38.7   47   114   108   104.0   42.4  32.44  3190.28  36.20   37.7
  25 Schooner 2.2 XB SSE x64               :  3086.11    264    38.4   46   111   107   101.5   42.0  32.92  3190.74  35.69   36.7
  26 Defenchess 2.3 dev BMI2 x64           :  3084.08    282    38.3   48   120   114   108.0   42.6  33.20  3198.11  36.10   37.7
  27 DanaSah 9.0 NN AVX2 x64               :  3070.98    249    37.8   42   104   103    94.0   41.8  32.88  3179.67  35.64   37.2
  28 Chiron 5 x64                          :  3063.39    259    37.8   48   100   111    98.0   38.6  33.74  3181.04  35.94   38.8
  29 Laser 1.7 BMI2 x64                    :  3062.89    275    34.9   52    88   135    96.0   32.0  33.30  3203.21  36.28   38.0
  30 Clover 2.4 x64                        :  3059.32    241    36.5   39    98   104    88.0   40.7  35.42  3185.78  35.79   37.3
  31 Winter 0.9 BMI2 x64                   :  3052.94    243    38.7   57    74   112    94.0   30.5  33.73  3161.78  35.84   36.9
  32 Fizbo 2.0 BMI2 x64                    :  3037.82    255    32.7   39    89   127    83.5   34.9  34.72  3198.61  36.02   38.0
  33 Andscacs 0.95.123 x64                 :  3035.79    256    30.9   42    74   140    79.0   28.9  34.36  3210.02  36.40   37.0
  34 GullChess 3.0 Sy BMI2 x64             :  3028.48    244    35.2   36   100   108    86.0   41.0  35.28  3163.75  35.64   37.0
  35 Orion 0.8 NN FMA x64                  :  3012.99    239    32.8   26   105   108    78.5   43.9  36.53  3167.41  35.59   37.7
  36 Dark Toga 1.1 NN AVX2 x64             :  3009.17    246    31.1   39    75   132    76.5   30.5  34.40  3183.24  35.73   37.6
  37 Halogen 10 NN PEXT x64                :  3006.68    256    31.6   44    74   138    81.0   28.9  34.94  3176.85  35.79   37.3
  38 Stash 31.16 x64                       :  2986.79    234    28.6   30    74   130    67.0   31.6  38.88  3188.11  36.03   37.4
  39 Combusken 1.4.0 AMD x64               :  2985.58    250    30.6   36    81   133    76.5   32.4  35.26  3171.27  35.48   37.2
  40 Nirvanachess 2.5 POPCNT x64           :  2920.49    260    23.7   20    83   157    61.5   31.9  36.07  3177.39  35.66   38.5
  41 Demolito 2021-07-09 x64               :  2893.52    277    20.0   26    59   192    55.5   21.3  38.89  3191.02  35.92   37.9

White advantage = 83.27 +/- 4.23
Draw rate (equal opponents) = 52.35 % +/- 0.93
Looks interesting ...
The same as in FCP-Tourney-2020 and FCP Tourney-2021 and FCP Rating List.
The strong point from Komodo in comparsion to Stockfish.
The different seems to be a bit higher as normally.

Komodo is in this playing phase stronger as Stockfish, more hard fights as Stockfish, avoid draws and more points with won games, using small advantatages for it. The reason is in my opinion the pawn-structure in late mid-games.

Very interesting is the style from Zahak.
A bit to compare with the older Ktulu or Combusken.
A lot of very fast lost games but is the endgame in the near the playing strength goes high.

Attacker like Booot, Wasp, Fizbo lost strength ... nothing new for me (allways the same).

4.
Games ended between 100-299 moves ...
Resign = off games ...

Code: Select all

   # Player                                :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 151121 NN x64               :  3293.80    215    62.1   52   163     0   133.5   75.8  25.03  3194.68  20.26   34.1
   2 Fire 8 NN MC.3 x64                    :  3293.01    271    64.6   83   184     4   175.0   67.9  23.48  3173.63  20.04   36.1
   3 Dragon 2.5.1 by Komodo NN AVX2 x64    :  3292.04    265    61.9   64   200     1   164.0   75.5  22.60  3191.47  20.29   35.3
   4 Koivisto 7.5 NN AVX2 x64              :  3287.45    416    65.4  132   280     4   272.0   67.3  18.53  3170.86  20.04   37.9
   5 SlowChess Blitz 2.8 NN AVX2 x64       :  3277.96    314    62.1   80   230     4   195.0   73.2  21.00  3184.03  20.15   36.7
   6 Ethereal 13.25 NN PEXT x64            :  3274.03    360    63.8  106   247     7   229.5   68.6  19.44  3169.56  20.09   37.4
   7 Berserk 7 NN PEXT x64                 :  3270.25    373    62.6  102   263     8   233.5   70.5  19.27  3174.33  19.96   37.7
   8 Nemorino 6.09 NN x64                  :  3261.02    293    64.0   99   177    17   187.5   60.4  23.03  3160.15  19.99   37.6
   9 RubiChess 2.2 NN x64                  :  3251.57    285    60.4   70   204    11   172.0   71.6  22.89  3174.35  19.96   36.2
  10 Seer 2.4.0 NN AVX2 x64                :  3249.23    314    60.4   83   213    18   189.5   67.8  20.74  3171.01  20.02   38.1
  11 Igel 3.0.10 NN BMI2 x64               :  3237.41    247    57.7   54   177    16   142.5   71.7  22.21  3173.98  19.80   35.5
  12 Lc0 0.28.0 NN CPU-dnnl                :  3214.73    321    55.5   56   244    21   178.0   76.0  19.65  3168.89  20.06   38.3
  13 rofChade 2.310 NN x64                 :  3210.78    332    54.5   57   248    27   181.0   74.7  18.37  3176.83  20.02   37.4
  14 Revenge 1.0 NN x64                    :  3204.26    255    52.4   38   191    26   133.5   74.9  21.09  3182.70  20.04   36.2
  15 Arasan 23.0.1 NN AVX2 x64             :  3187.99    288    51.7   46   206    36   149.0   71.5  20.02  3175.31  19.99   37.4
  16 Zahak 8.6 AMD x64                     :  3183.60    311    54.0   63   210    38   168.0   67.5  19.73  3154.84  19.80   36.7
  17 Minic 3.17 NN x64                     :  3181.55    326    53.5   53   243    30   174.5   74.5  19.47  3159.37  19.93   37.6
  18 Xiphos 0.6 BMI2 x64                   :  3174.88    307    49.7   43   219    45   152.5   71.3  18.17  3173.34  19.91   35.9
  19 Shredder 13 POPCNT x64                :  3150.04    302    46.5   40   201    61   140.5   66.6  19.96  3177.51  20.16   38.1
  20 Weiss 2.0 PEXT x64                    :  3146.01    361    47.2   44   253    64   170.5   70.1  17.71  3166.90  20.00   37.9
  21 Booot 6.5 POPCNT x64                  :  3145.35    305    45.6   31   216    58   139.0   70.8  18.48  3177.60  19.99   37.3
  22 Defenchess 2.3 dev BMI2 x64           :  3142.07    341    47.4   39   245    57   161.5   71.8  18.72  3164.44  19.91   37.9
  23 Fritz 18 (Ginkgo) x64                 :  3137.45    353    45.8   33   257    63   161.5   72.8  18.28  3170.94  20.08   38.4
  24 Clover 2.4 x64                        :  3133.14    337    46.7   32   251    54   157.5   74.5  18.27  3161.06  19.98   37.6
  25 Marvin 5.2.0 NN AVX2 x64              :  3131.51    289    45.2   28   205    56   130.5   70.9  20.17  3168.80  19.87   36.7
  26 Winter 0.9 BMI2 x64                   :  3130.90    280    45.0   26   200    54   126.0   71.4  20.41  3171.77  20.02   37.7
  27 Laser 1.7 BMI2 x64                    :  3128.51    309    44.3   40   194    75   137.0   62.8  19.22  3170.90  20.05   38.0
  28 Wasp 5.00 NN AVX2 x64                 :  3128.36    287    45.3   24   212    51   130.0   73.9  19.03  3163.64  19.87   37.2
  29 Halogen 10 NN PEXT x64                :  3125.90    250    45.8   25   179    46   114.5   71.6  20.73  3157.67  19.86   36.8
  30 Orion 0.8 NN FMA x64                  :  3124.58    253    46.0   19   195    39   116.5   77.1  22.04  3155.74  19.75   36.1
  31 GullChess 3.0 Sy BMI2 x64             :  3123.34    304    46.2   33   215    56   140.5   70.7  19.57  3155.65  19.82   36.6
  32 Chiron 5 x64                          :  3118.42    333    42.8   38   209    86   142.5   62.8  18.42  3173.78  19.96   37.8
  33 Fizbo 2.0 BMI2 x64                    :  3116.06    259    43.6   30   166    63   113.0   64.1  21.49  3166.27  19.77   36.0
  34 Combusken 1.4.0 AMD x64               :  3115.62    249    46.0   26   177    46   114.5   71.1  20.98  3149.74  19.88   34.9
  35 Schooner 2.2 XB SSE x64               :  3099.82    302    39.6   26   187    89   119.5   61.9  20.61  3177.90  20.11   37.7
  36 Andscacs 0.95.123 x64                 :  3098.98    328    41.3   27   217    84   135.5   66.2  19.71  3167.29  19.84   37.7
  37 Dark Toga 1.1 NN AVX2 x64             :  3097.31    392    41.6   37   252   103   163.0   64.3  17.55  3164.26  19.90   37.9
  38 DanaSah 9.0 NN AVX2 x64               :  3086.45    268    39.2   14   182    72   105.0   67.9  21.42  3165.40  19.84   36.7
  39 Stash 31.16 x64                       :  3073.44    393    37.5   20   255   118   147.5   64.9  18.61  3167.67  20.02   38.2
  40 Nirvanachess 2.5 POPCNT x64           :  3067.54    309    36.6   21   184   104   113.0   59.5  21.36  3170.42  19.98   37.7
  41 Demolito 2021-07-09 x64               :  3044.66    393    34.5   12   247   134   135.5   62.8  19.20  3165.51  19.96   37.7

White advantage = 31.25 +/- 2.37
Draw rate (equal opponents) = 79.43 % +/- 0.64
Now more clear ...
Strength from Wasp, Booot and Fizbo go down. Same for Revenge or Schooner!
Zahak and Minic hold the endgame strength.
Now better to see that the strength from Combusken goes high.
Koivisto on the same level the best engines have.

Very nice are engine can hold in all playing phases the same level.
Also not easy ...

A good example: RubiChess!!

And the short-games stats are all the time very interesting!

Arasan I saw never on such a good ranking.
Revenge is just great here!

Code: Select all

Short games statistic:
After round 26 out of 30 = 1.040 games per engine

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

1.007 of 21.320 = 4,723% : Miniatures, mate below 60 moves
  730 of 21.320 = 3,424% : Fast draws below 40 moves

                                      won59 lost59 draw39        won59w won59b lost59w lost59b
01. Stockfish 151121 NN x64             115      0     25            84     31       0       0
02. Dragon 2.5.1 by Komodo NN AVX x64    90      0     16            67     23       0       0
03. Revenge 1.0 NN x64                   63      1     40            47     16       1       0
04. Ethereal 13.25 NN PEXT x64           48      3     37            35     13       0       3
05. Igel 3.0.10 NN BMI2 x64              46      3     64            37      9       0       3
06. Arasan 23.0.1 NN AVX2 x64            43      1     40            36      7       1       0
07. Fire 8 NN MC.3 x64                   42      0     30            32     10       0       0
08. SlowChess Blitz 2.8 NN AVX2 x64      41      1     28            26     15       1       0
09. RubiChess 2.2 NN x64                 41      4     46            23     18       1       3
10. Booot 6.5 POPCNT x64                 38     12     47            29      9       2       9
11. Berserk 7 NN PEXT x64                29      5     37            21      8       0       5
12. rofChade 2.310 NN x64                28      8     47            17     11       0       8
13. Schooner 2.2 XB SSE x64              27      8     31            16     11       2       6
14. Koivisto 7.5 NN AVX2 x64             25      7     23            16      9       1       6
15. Nemorino 6.09 NN x64                 22     12     38            16      6       3       9
16. Xiphos 0.6 BMI2 x64                  22     12     49            20      2       2      10
17. Weiss 2.0 PEXT x64                   21     14     31            11     10       3      11
18. Lc0 0.28.0 NN CPU-dnnl               21     56     49            16      5      22      34
19. Wasp 5.00 NN AVX2 x64                20      8     46            11      9       1       7
20. Fritz 18 (Ginkgo) x64                18     27     18            13      5       7      20
21. Laser 1.7 BMI2 x64                   17     13     43             9      8       7       6
22. Seer 2.4.0 NN AVX2 x64               17     13     45            10      7       2      11
23. Fizbo 2.0 BMI2 x64                   15     26     30             5     10       6      20
24. Clover 2.4 x64                       14     16     46            11      3       2      14
25. DanaSah 9.0 NN AVX2 x64              14     32     41            11      3      12      20
26. Andscacs 0.95.123 x64                13     23     34             6      7       7      16
27. Minic 3.17 NN x64                    13     39     48             9      4       8      31
28. Defenchess 2.3 dev BMI2 x64          11     10     39            10      1       1       9
29. Chiron 5 x64                         11     29     30             8      3       5      24
30. Shredder 13 POPCNT x64               10     32     32             9      1       5      27
31. Dark Toga 1.1 NN AVX2 x64             9     19     23             7      2       4      15
32. Zahak 8.6 AMD x64                     9     57     43             8      1      24      33
33. GullChess 3.0 Sy BMI2 x64             9     63     21             8      1      23      40
34. Marvin 5.2.0 NN AVX2 x64              8     26     35             6      2      11      15
35. Winter 0.9 BMI2 x64                   8     74     28             6      2      23      51
36. Combusken 1.4.0 AMD x64               7     79     29             6      1      28      51
37. Nirvanachess 2.5 POPCNT x64           5     45     20             4      1      15      30
38. Stash 31.16 x64                       5     47     33             4      1      11      36
39. Halogen 10 NN PEXT x64                5     60     40             4      1      22      38
40. Orion 0.8 NN FMA x64                  4     71     45             4      0      14      57
41. Demolito 2021-07-09 x64               3     52     13             2      1      14      37

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

  282 of 21.320 = 1,322% : Miniatures, mate below 50 moves
  234 of 21.320 = 1,097% : Fast draws below 30 moves

                                      won49 lost49 draw29        won49w won49b lost49w lost49b
01. Stockfish 151121 NN x64              36      0      9            26     10       0       0
02. Revenge 1.0 NN x64                   27      0     11            20      7       0       0
03. Dragon 2.5.1 by Komodo NN AVX x64    26      0      3            15     11       0       0
04. Arasan 23.0.1 NN AVX2 x64            16      0     15            14      2       0       0
05. SlowChess Blitz 2.8 NN AVX2 x64      12      0     11             8      3       0       0
06. Berserk 7 NN PEXT x64                12      2     16             8      4       0       2
07. Fire 8 NN MC.3 x64                   11      0     11             8      3       0       0
08. Booot 6.5 POPCNT x64                 10      2     13             5      5       1       1
09. Igel 3.0.10 NN BMI2 x64               9      0     19             7      2       0       0
10. RubiChess 2.2 NN x64                  9      3     18             4      5       1       2
11. Koivisto 7.5 NN AVX2 x64              9      4      7             5      4       1       3
12. Ethereal 13.25 NN PEXT x64            8      1     16             6      2       0       1
13. Schooner 2.2 XB SSE x64               7      1      9             4      3       1       0
14. Laser 1.7 BMI2 x64                    7      3     13             4      3       0       3
15. Weiss 2.0 PEXT x64                    7      5     12             3      4       1       4
16. Xiphos 0.6 BMI2 x64                   6      3     17             5      1       1       2
17. Lc0 0.28.0 NN CPU-dnnl                6     22     12             4      2       8      14
18. Fizbo 2.0 BMI2 x64                    5      3      7             2      3       2       1
19. Nemorino 6.09 NN x64                  5      4      9             3      2       0       4
20. Clover 2.4 x64                        5      5     14             4      1       0       5
21. DanaSah 9.0 NN AVX2 x64               5     11     16             4      1       4       7
22. Winter 0.9 BMI2 x64                   5     12      8             5      0       5       7
23. Defenchess 2.3 dev BMI2 x64           4      2     15             4      0       1       1
24. Andscacs 0.95.123 x64                 4      5     12             1      3       0       5
25. GullChess 3.0 Sy BMI2 x64             4     24      5             3      1       9      15
26. Shredder 13 POPCNT x64                3      2     11             3      0       0       2
27. Seer 2.4.0 NN AVX2 x64                3      2     17             2      1       0       2
28. Fritz 18 (Ginkgo) x64                 3      8      7             3      0       3       5
29. Demolito 2021-07-09 x64               3     17      5             2      1       6      11
30. Wasp 5.00 NN AVX2 x64                 2      0     16             0      2       0       0
31. rofChade 2.310 NN x64                 2      4     19             1      1       0       4
32. Chiron 5 x64                          2      5      6             1      1       3       2
33. Nirvanachess 2.5 POPCNT x64           2     14      7             2      0       6       8
34. Dark Toga 1.1 NN AVX2 x64             1      5      8             1      0       0       5
35. Minic 3.17 NN x64                     1      6     14             1      0       1       5
36. Stash 31.16 x64                       1     17      8             0      1       5      12
37. Orion 0.8 NN FMA x64                  1     17     12             1      0       2      15
38. Halogen 10 NN PEXT x64                1     18     14             1      0       6      11
39. Combusken 1.4.0 AMD x64               1     24      6             1      0       7      17
40. Zahak 8.6 AMD x64                     1     26      9             1      0      10      16
41. Marvin 5.2.0 NN AVX2 x64              0      5     11             0      0       2       3
At the moment I disucss a parameter combination for Wasp with John Stanback.
NN start with x-pieces on board (kings safty with or without NN is very strong in case of Wasp).
Positive contempt with many pieces on board only.
The combination from this two settings = Speculative = on.

Best
Frank
Frank Quisinsky
Posts: 7285
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Statistics to playing phases ... Komodo / Zahak and the others!

Post by Frank Quisinsky »

The secret for I like or I think the programmers should try to improve ...

At first ...
Please forget the boring rating list (thas is very boring for me ...)
Have more a look on MoveAvg (move average) ...

To which time an engine is ready?
If all playing phases the engine can play with the same level (more human like will be ... strong in the beginning, with weaknesses in endgames). More computer chess will be ... bad in mid-games and strong in endgames.

Please have a look on RubiChess again ...
In all playing phases the same level. Same place in short game stats.

And the move average ... is great!!
85.8 only!!

FANTASTIC, maybe a bit to many short draws in the mid-games. So the programmer can try to make it better. But all in all ... I like RubiChess
for follow the games in live mode.

The boring and horrible endgames ...
Engines should be able to give endgames in absolutly clear position very fast draw.
I lost here so many time engines are thinking minutes in 0.00 positions or fighting in dead draw positions many hours (1:20h games). A good example is Koivisto but many others do that. I lost time ressources for more games. Different engines need a minute before a move like mate 2 will be play. Programmers should more looking on the end of games for interesting improvements. With more time more games on higher niveau can be play.

So, move average is very important ... a chess program is with a lesser move average in my opinion more complete. Stockfish with a fantastic statistic here. Igel are strong, Booot are strong. But the programmer of Igel should avaoid all the short draws in the mid-games. Komodo made this better as all the others (avoid draw in the mid-games).

Here the boring Elo list.
Again ... have a look on move-average!

Berserk to high ...
Ethereal to high ...
Koivisto to high ...

A lot of room for improvements!!

Code: Select all

  FCP Tourney-2022
  ----------------

  Date           : December 25th, 2021 (14:45)
  Games          : 21.320
  Version        : 26.0, after R26
  Elo            : Shredder 13 POPCNT x64 = 3125
  Conditions     : http://www.amateurschach.de/main/_fcp-tourney-2022.htm

  Generated with : Ordo 1.2.6 x64 (Ratings for chess and other games)
  https://sites.google.com/site/gaviotachessengine/ordo
  by Miguel A. Ballicora (Argentina)



   # Player                                 :      Elo  Games  Score%   won  draw  lost  Points  Draw%   Error   OppAvg   OppE  MoveAvg
  -------------------------------------------------------------------------------------------------------------------------------------
  01. Stockfish 151121 NN x64               :  3465.17  1.040    81.9   665   374     1   852.0   36.0   18.84  3170.99  13.63     82.1
  02. Dragon 2.5.1 by Komodo NN AVX2 x64    :  3453.47  1.040    81.0   648   388     4   842.0   37.3   16.72  3171.28  13.69     87.9
  03. Fire 8 NN MC.3 x64                    :  3376.21  1.040    73.8   519   497    24   767.5   47.8   15.59  3173.21  13.71     86.1
  04. SlowChess Blitz 2.8 NN AVX2 x64       :  3359.03  1.040    72.0   484   530    26   749.0   51.0   15.19  3173.64  13.72     89.5
  05. Koivisto 7.5 NN AVX2 x64              :  3347.75  1.040    70.8   475   523    42   736.5   50.3   14.52  3173.92  13.74     97.3
  06. Ethereal 13.25 NN PEXT x64            :  3343.31  1.040    70.3   471   521    48   731.5   50.1   14.61  3174.03  13.74     92.5
  07. Berserk 7 NN PEXT x64                 :  3333.22  1.040    69.2   443   554    43   720.0   53.3   14.70  3174.28  13.74     93.8
  08. RubiChess 2.2 NN x64                  :  3296.97  1.040    65.1   380   594    66   677.0   57.1   14.03  3175.19  13.75     85.8
  09. Revenge 1.0 NN x64                    :  3281.96  1.040    63.3   355   607    78   658.5   58.4   13.63  3175.57  13.76     84.0
  10. Seer 2.4.0 NN AVX2 x64                :  3268.79  1.040    61.7   344   596   100   642.0   57.3   13.94  3175.90  13.76     86.2
  -------------------------------------------------------------------------------------------------------------------------------------
  11. Igel 3.0.10 NN BMI2 x64               :  3264.84  1.040    61.3   325   624    91   637.0   60.0   13.74  3175.99  13.76     80.4
  12. Nemorino 6.09 NN x64                  :  3254.65  1.040    60.0   336   576   128   624.0   55.4   13.31  3176.25  13.77     85.4
  13. rofChade 2.310 NN x64                 :  3253.48  1.040    59.9   327   591   122   622.5   56.8   13.41  3176.28  13.77     90.4
  14. Arasan 23.0.1 NN AVX2 x64             :  3222.33  1.040    56.0   270   624   146   582.0   60.0   13.03  3177.06  13.78     85.6
  15. Lc0 0.28.0 NN CPU-dnnl                :  3212.09  1.040    54.7   280   577   183   568.5   55.5   13.07  3177.31  13.78     90.4
  16. Booot 6.5 POPCNT x64                  :  3200.03  1.040    53.1   246   613   181   552.5   58.9   13.07  3177.61  13.78     86.6
  17. Xiphos 0.6 BMI2 x64                   :  3170.79  1.040    49.4   200   627   213   513.5   60.3   13.07  3178.35  13.78     87.6
  18. Minic 3.17 NN x64                     :  3155.07  1.040    47.4   190   605   245   492.5   58.2   13.07  3178.74  13.78     89.7
  19. Weiss 2.0 PEXT x64                    :  3154.70  1.040    47.3   192   600   248   492.0   57.7   13.02  3178.75  13.78     91.2
  19. Wasp 5.00 NN AVX2 x64                 :  3154.70  1.040    47.3   177   630   233   492.0   60.6   12.96  3178.75  13.78     86.3
  -------------------------------------------------------------------------------------------------------------------------------------
  21. Schooner 2.2 XB SSE x64               :  3134.43  1.040    44.7   165   600   275   465.0   57.7   12.72  3179.25  13.79     87.3
  22. Fritz 18 (Ginkgo) x64                 :  3130.29  1.040    44.2   172   575   293   459.5   55.3   12.97  3179.36  13.78     91.8
  23. Defenchess 2.3 dev BMI2 x64           :  3125.76  1.040    43.6   157   593   290   453.5   57.0   13.40  3179.47  13.77     91.5
  24. Zahak 8.6 AMD x64                     :  3125.38  1.040    43.6   176   554   310   453.0   53.3   12.85  3179.48  13.78     86.8
  25. Shredder 13 POPCNT x64                :  3125.00  1.040    43.5   165   575   300   452.5   55.3   13.27  3179.49  13.77     87.2
  26. Laser 1.7 BMI2 x64                    :  3117.80  1.040    42.6   163   560   317   443.0   53.8   12.74  3179.67  13.79     87.7
  27. Clover 2.4 x64                        :  3114.38  1.040    42.2   135   607   298   438.5   58.4   12.61  3179.76  13.79     88.9
  28. Marvin 5.2.0 NN AVX2 x64              :  3112.10  1.040    41.9   149   573   318   435.5   55.1   13.36  3179.81  13.77     86.8
  29. Chiron 5 x64                          :  3104.46  1.040    40.9   145   561   334   425.5   53.9   13.40  3180.00  13.77     89.4
  30. DanaSah 9.0 NN AVX2 x64               :  3097.55  1.040    40.0   131   571   338   416.5   54.9   12.85  3180.18  13.78     84.1
  -------------------------------------------------------------------------------------------------------------------------------------
  31. Fizbo 2.0 BMI2 x64                    :  3096.40  1.040    39.9   136   558   346   415.0   53.7   12.94  3180.21  13.78     84.2
  32. Andscacs 0.95.123 x64                 :  3089.06  1.040    39.0   132   547   361   405.5   52.6   12.73  3180.39  13.79     89.3
  33. Dark Toga 1.1 NN AVX2 x64             :  3071.06  1.040    36.8   126   513   401   382.5   49.3   13.57  3180.84  13.77     95.5
  34. Winter 0.9 BMI2 x64                   :  3067.90  1.040    36.4   131   495   414   378.5   47.6   13.73  3180.92  13.76     85.6
  35. Halogen 10 NN PEXT x64                :  3061.13  1.040    35.6   108   524   408   370.0   50.4   13.58  3181.09  13.77     84.5
  36. GullChess 3.0 Sy BMI2 x64             :  3051.08  1.040    34.4   109   497   434   357.5   47.8   13.61  3181.34  13.76     88.2
  37. Orion 0.8 NN FMA x64                  :  3047.02  1.040    33.9    78   549   413   352.5   52.8   13.53  3181.44  13.77     82.5
  38. Stash 31.16 x64                       :  3042.12  1.040    33.3    85   523   432   346.5   50.3   13.89  3181.56  13.76     93.1
  39. Nirvanachess 2.5 POPCNT x64           :  3014.49  1.040    30.1    71   485   484   313.5   46.6   14.17  3182.25  13.75     87.7
  40. Combusken 1.4.0 AMD x64               :  3008.02  1.040    29.4    85   442   513   306.0   42.5   14.12  3182.41  13.75     85.1
  -------------------------------------------------------------------------------------------------------------------------------------
  41. Demolito 2021-07-09 x64               :  3000.60  1.040    28.6    78   439   523   297.5   42.2   14.61  3182.60  13.74     96.7
  -------------------------------------------------------------------------------------------------------------------------------------

  MoveAvg (average)           : 178.19 / 89.095
  White advantage             : 64.45  +/- 1.57
  Draw rate (equal opponents) : 69.87% +/- 0.43

  Games        :  21.320 (finished)
  White Wins   :   6.624 (31.1%)
  Black Wins   :   3.400 (15.9%)
  Draws        :  11.296 (53.0%)
  Unfinished   :       0

  White Perf.  : 57.6%
  Black Perf.  : 42.4%

  ECO A00-A99  :   4.065 Games (19.1%)
  ECO B00-B99  :   4.736 Games (22.2%)
  ECO C00-C99  :   4.802 Games (22.5%)
  ECO D00-D99  :   3.647 Games (17.1%)
  ECO E00-E99  :   4.070 Games (19.1%)
  
Frank Quisinsky
Posts: 7285
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Statistics to playing phases ... Komodo / Zahak and the others!

Post by Frank Quisinsky »

And the last message on Christmas ...

With many opponents and many time I am very sure that Komodo is stronger as Stockfish.
If I produce 5:30h games (1-core, 4.4Ghz) Komodo have the same strength Stockfish have.
Can see it if I made analyzes with my blitz results (never public) and now the 1:20h games.
Very easy to see that with programs like Excel because the playing phase Komodo have an advantage is with longer time controls more and more important.

This explain ...

Wasp for an example:
In rating systems with fast time controls Wasp is 100-110 Elo stronger.
And with more time the advantage go down to 70-80 Elo (Wasp 4.5 to 5.0) ... and the reason is the weak point Wasp have in endgames.
But in case of Wasp not important because the human-like-style is much more important for Wasp. And this playing style is just great!!

Elo isn't important but the playing style is important.
To try to make the programs more "complete" for looking in live mode the games that is my wish for the next year.
A lot of room for improvments to work on "time-ressources" with a shorter move-average. Time "fritter away" with thinking in clear mate or draw positions by so many engines!!

Go the way the RubiChess programmers go and it is much more interesting to follow endgames.

Best
Frank
Ferdy
Posts: 4853
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: Statistics to playing phases ... Komodo / Zahak and the others!

Post by Ferdy »

Interesting to see is the stats on which engine wins on lower number of moves, and which engine loses on higher number of moves with the condition that engines play the same opening. Example, e1 vs e2 using opening1, reverse it. Then e1 vs e3 using opening1 ... then e2 vs e3 using opening1 .... Then in the next round use a different opening say opening2 and so on. Selection of most common openings would be enough.