Willow 4.0 (final release for now)

[Whiskers]

but my goal is to fulfill the dream that I abandoned in the chase for ELO; to make an engine that plays with the most exciting style possible while still remaining clearly superhuman.

I wrote this already to the booot-author, but IMO this is important for anybody, who wants to build an engine with more aggressive playing-style:

Ed Schroeder used my EAS-Tool to make his legendary Rebel play more sacrifices (he refactored the learning-data, considering the EAS-Tool). And it worked! Look at my EAS-Ratinglist:
https://www.sp-cc.de/eas-ratinglist.htm

Rebel EAS played the most high sacs and the most sacs overall (except the 2 Stockfishes) and Rebel EAS gained Elo compared to Rebel 16.2

Ed Schroder wrote this:
"REBEL-EAS is an in between version based on the same, but more, neural net data as REBEL-16.2 but heavily modified during the last 5 months in order to make REBEL to play more aggressive and runs as a module under Chess System Tal 2.0
. REBEL 16.2 gains an EAS score of : 80.258
. REBEL-EAS gets an EAS score of : 160.569
As a cherish on the cake the playing strength of the neural net has increased with 20-25 elo using balanced positions. Using the TCEC positions composed by Jeroen Noomen and GM Matthew Sadler adds another 26 elo points while the draw rate dropped with almost 20%.
The REBEL-EAS approach: King Safety and Mobility were heavily used, the epd scores of King Safety and Mobility even increased with 25% with this function to get the desired effect, to play more aggressive measured with the EAS tool."

Perhaps you should try to go the same way as Ed did?

Here the link to my EAS-Tool:
https://www.sp-cc.de/files/engines_aggr ... cs_tool.7z

And her the link to the Ed Schroeder Tool (link-text is wrong, but leads to the correct site):
https://rebel7775.wixsite.com/rebel/kop ... rl-blitz-1

(Mention, the EAS-Tool needs a huge amount of games for valid EAS-scores. I strongly recommend at least 3000 played games per engine...but the more games, the better)

I know about the EAS tool but I am trying to find out what aggressive positions look like, not just aggressive games. Also, the fact that Stockfish and Torch are at the top by a country mile suggests that a large part of what EAS is measuring is engines taking advantage of tactical mistakes by other engines rather than actively seeking out an aggressive play style. Thank you for the helpful links though I will try using them.

[JohnW]

Hi all, here is Willow 4.0 after several more months of development, it's 170 elo stronger than 3.1 using balanced books and likely much more on unbalanced books like Pohl. It's currently scoring 5/8 at its debut at TCEC and I feel so proud of it, and I've loved every step of the journey.

I'm moving on to my next project after this; it'll probably draw on a lot of code from Willow, but my goal is to fulfill the dream that I abandoned in the chase for ELO; to make an engine that plays with the most exciting style possible while still remaining clearly superhuman. I'll probably post a progress thread in the vein of Leorik's devblog. See you then.

https://github.com/Adam-Kulju/Willow/releases/tag/4.0

Excuse my ignorance, but why is there a folder for 4.0 containing 3.1 exes instead of 4.0 exes?

I forgot to update the version info printed when uci is sent.
Here is a bugfix version that fixes that problem, use it instead of 4.0: https://github.com/Adam-Kulju/Willow/re ... ag/4.0.0.1

Hmm, it still says 3.1 when I try to register it in Fritz

[Frank Quisinsky]

This is often true!

But if the engines have a small move average for wins, not many own fast losses, you can guess that the first phase of play must be very aggressive. If you look at the games, you can see the same pattern of attack in most cases.

And for the mid- and endgame the statistics are more complicated.
You can do a lot ...

Example:
In how many cases you find a move in the endgame the evaluation goes up by 1 or 2 points. Here the tool from Stefan is great.

The combination of viewpoints will give us some information about the strengths.

It is interesting to look a little deeper.

I look for balanced middlegame positions for games that end in 70-90 moves (in most cases late middlegame). Now I look at how many knights, bishops and rooks are on the board. Are the pieces close to the own king. In step 2 I look at how good the pawn structure is, so carefully remove all the pieces from the board, not the pawns. Now try to evaluate the position with only the pawns on the board and compare it with the final game result.

I am working a time with GM Jörg Hickl.
Jörg wrote a book about it (Die Macht der Bauern, but in English available too). That is one of the best chess books ever (think so). Grandmaster where are working for his own evaluation with pawn structures are in most of cases not opening experts. The openings are not important and they plays most of times passive closed systems. I know that from him and different other grandmasters.

As you can see, Stockfish is not ranked 1st in the World because Dragon has the better pawn structure in the late middlegames or the earlier transposition into endgame. OK, for two years ago. A long time not compare it with the current versions of Dragon and Stockfish. Most of the games ended very quickly when open with aggressive pawns. The problem here is ... engines like Stockfish and Co. are perfect defenders. At the end of the attack the pawn structure from the attacker is bad and the game is over. Such problems have the attackers, where are not in the near of playing strength from Stockfish and Co.

We have no statistics to evaluate the pawn structure in games!!
Possible with own chess knowledge only.

King-zones is a big issue, pawn structure is a big issue.

The next point is pawn sacrifice. You give a pawn for an attack, or the quality. And that is the problem, most of these games end in a draw. A problem because the tools we have are most of time in looking wins. With draw games much more things are possible for find out strengths and weeknesses. OK, no good tools are available for all the draws.

The engine has a bad start position, but has found a way to draw. This is a good idea with UHU positions. But only one way, so many other things are also important for looking deeper.

The Velvet programmer do interesting things, alone what he gave for an information to his release 6.0.0. I am sure he is working with own tools for see this and that and try to opimate his Velvet more and more. No other engine can beat the aggressivess in the middle of the board from Velvet in closed positions with many pieces on board.

Best
Frank

Impressions: Always doubting whether impressions are right.
Clearly the most interesting point for me ...
Which weaknesses and strengths generate the available engines. This is very exciting. This is most exciting ... this is computer chess. So, what in the World is Elo?

[carldaman]

Hi all, here is Willow 4.0 after several more months of development, it's 170 elo stronger than 3.1 using balanced books and likely much more on unbalanced books like Pohl. It's currently scoring 5/8 at its debut at TCEC and I feel so proud of it, and I've loved every step of the journey.

I'm moving on to my next project after this; it'll probably draw on a lot of code from Willow, but my goal is to fulfill the dream that I abandoned in the chase for ELO; to make an engine that plays with the most exciting style possible while still remaining clearly superhuman. I'll probably post a progress thread in the vein of Leorik's devblog. See you then.

https://github.com/Adam-Kulju/Willow/releases/tag/4.0

A very worthy pursuit, to chase style not Elo! This is something I've been advocating for many years...
I wish you success!

[Whiskers]

Hi all, here is Willow 4.0 after several more months of development, it's 170 elo stronger than 3.1 using balanced books and likely much more on unbalanced books like Pohl. It's currently scoring 5/8 at its debut at TCEC and I feel so proud of it, and I've loved every step of the journey.

I'm moving on to my next project after this; it'll probably draw on a lot of code from Willow, but my goal is to fulfill the dream that I abandoned in the chase for ELO; to make an engine that plays with the most exciting style possible while still remaining clearly superhuman. I'll probably post a progress thread in the vein of Leorik's devblog. See you then.

https://github.com/Adam-Kulju/Willow/releases/tag/4.0

Excuse my ignorance, but why is there a folder for 4.0 containing 3.1 exes instead of 4.0 exes?

I forgot to update the version info printed when uci is sent.
Here is a bugfix version that fixes that problem, use it instead of 4.0: https://github.com/Adam-Kulju/Willow/re ... ag/4.0.0.1

Hmm, it still says 3.1 when I try to register it in Fritz

fixed now, try it again please

[peter]

I forgot to update the version info printed when uci is sent.
Here is a bugfix version that fixes that problem, use it instead of 4.0: https://github.com/Adam-Kulju/Willow/re ... ag/4.0.0.1

Hi!
At willow-v2-windows.exe download Windows10 Defender gives warning about
Trojan:Win32/Sabsik.FL.A!ml
Just to tell, regards

[Whiskers]

I forgot to update the version info printed when uci is sent.
Here is a bugfix version that fixes that problem, use it instead of 4.0: https://github.com/Adam-Kulju/Willow/re ... ag/4.0.0.1

Hi!
At willow-v2-windows.exe download Windows10 Defender gives warning about
Trojan:Win32/Sabsik.FL.A!ml
Just to tell, regards

I have Windows 10 as well and I also get messages about it being a potentially dangerous download when I try to download. This is a problem that has also randomly happened with some of my engine dev friends. If it's not letting you download the executable directly, you can build Willow from source using the instructions provided in the release notes.

Also, v3 is faster than v2 and should work on most machines. Only use v2 if you know v3 won't work or have already tried and failed to use v3. Thanks!

[Wolfgang]

Hi all, here is Willow 4.0 after several more months of development, it's 170 elo stronger than 3.1 using balanced books and likely much more on unbalanced books like Pohl. It's currently scoring 5/8 at its debut at TCEC and I feel so proud of it, and I've loved every step of the journey.

I'm moving on to my next project after this; it'll probably draw on a lot of code from Willow, but my goal is to fulfill the dream that I abandoned in the chase for ELO; to make an engine that plays with the most exciting style possible while still remaining clearly superhuman. I'll probably post a progress thread in the vein of Leorik's devblog. See you then.

https://github.com/Adam-Kulju/Willow/releases/tag/4.0

Test for our blitz list (4'+2") started: https://cegt.forumieren.com/t2110-testi ... 4-0nn#3842

Thanks for the new version!

[Whiskers]

This is often true!

But if the engines have a small move average for wins, not many own fast losses, you can guess that the first phase of play must be very aggressive. If you look at the games, you can see the same pattern of attack in most cases.

And for the mid- and endgame the statistics are more complicated.
You can do a lot ...

Example:
In how many cases you find a move in the endgame the evaluation goes up by 1 or 2 points. Here the tool from Stefan is great.

The combination of viewpoints will give us some information about the strengths.

It is interesting to look a little deeper.

I look for balanced middlegame positions for games that end in 70-90 moves (in most cases late middlegame). Now I look at how many knights, bishops and rooks are on the board. Are the pieces close to the own king. In step 2 I look at how good the pawn structure is, so carefully remove all the pieces from the board, not the pawns. Now try to evaluate the position with only the pawns on the board and compare it with the final game result.

I am working a time with GM Jörg Hickl.
Jörg wrote a book about it (Die Macht der Bauern, but in English available too). That is one of the best chess books ever (think so). Grandmaster where are working for his own evaluation with pawn structures are in most of cases not opening experts. The openings are not important and they plays most of times passive closed systems. I know that from him and different other grandmasters.

As you can see, Stockfish is not ranked 1st in the World because Dragon has the better pawn structure in the late middlegames or the earlier transposition into endgame. OK, for two years ago. A long time not compare it with the current versions of Dragon and Stockfish. Most of the games ended very quickly when open with aggressive pawns. The problem here is ... engines like Stockfish and Co. are perfect defenders. At the end of the attack the pawn structure from the attacker is bad and the game is over. Such problems have the attackers, where are not in the near of playing strength from Stockfish and Co.

We have no statistics to evaluate the pawn structure in games!!
Possible with own chess knowledge only.

King-zones is a big issue, pawn structure is a big issue.

The next point is pawn sacrifice. You give a pawn for an attack, or the quality. And that is the problem, most of these games end in a draw. A problem because the tools we have are most of time in looking wins. With draw games much more things are possible for find out strengths and weeknesses. OK, no good tools are available for all the draws.

The engine has a bad start position, but has found a way to draw. This is a good idea with UHU positions. But only one way, so many other things are also important for looking deeper.

The Velvet programmer do interesting things, alone what he gave for an information to his release 6.0.0. I am sure he is working with own tools for see this and that and try to opimate his Velvet more and more. No other engine can beat the aggressivess in the middle of the board from Velvet in closed positions with many pieces on board.

Best
Frank

Impressions: Always doubting whether impressions are right.
Clearly the most interesting point for me ...
Which weaknesses and strengths generate the available engines. This is very exciting. This is most exciting ... this is computer chess. So, what in the World is Elo?

This method works well for weaker engines, where a human knows enough about chess to notice clear weaknesses/strengths not just from game results but from actual moves in the game. I can attest to the fact that very early versions of Willow were comparatively bad at raw calculative sacrifices and endgames, but comparatively exceled at sacrificed pawns for initiative.

The problem is that this doesn't really scale to 3000+ engines. When every facet of a chess engine (except for really minor nitpicks such as obvious fortresses) is unimaginably superior to people, you can't accurately judge any strengths and weaknesses just by looking at games. The vast majority of times, if a strong engine misses something and loses it's just because some component of search prevented it from seeing the right move in this circumstance and is not any indicator of a general weakness. A good example of this is Stockfish and its ability to solve a famous queen sacrifice position (4q1kr/p6p/1prQPppB/4n3/4P3/2P5/PP2B2P/R5K1 w - - 1 24, best move Qxe5!!). On a good day, Stockfish will be able to solve this position very quickly; then a minor search patch is merged, and all of a sudden Stockfish needs to get to depth 40 in order to spot the winning move. I really do wish it was different.

One last thing: You wrote that "if the engines have a small move average for wins, not many own fast losses, you can guess that the first phase of play must be very aggressive." I think that's much more of a sign of a very powerful engine that's just pouncing on opponent mistakes. If anything you'd expect an aggressive engine to lose quite a few games quickly, as an ideal aggressive engine would often neglect its own king safety to focus on attacking the other king (and occasionally would pay the price).

[Frank Quisinsky]

Everything is fine and the position you gave is interesting, thank you!

It is a combination of impressions and small statistics.
Often questionable impressions that we automatically generate while watching the games.

Rybka is a perfect example:

I did not know that such a strong engine with around 3000 Elo produces so many fast losses and is absolutely strong in endgames. In the past I looked at the distribution by number of moves. You can also create statistics by distribution by number of pieces on the board. Optimal is a combination of both, but not easy to achieve.

Games without resign-mode (allways important for stats).
Stats from FCP-Tourney-2020 (41.000 games on 4.9 Ghz, I believe was a 40 in 20 tournament before NN aera, speaking from 2.000 games for all of the 41 engines).

Games ended move 01-59

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 11 BMI2 x64           :  3264.21    405    80.0  243   162     0   324.0   40.0  25.57  3010.11  17.33   36.6
   2 Houdini 6.03 Pro x64            :  3180.15    411    70.0  164   247     0   287.5   60.1  19.14  3020.92  17.23   37.3
   3 Komodo 14.0 BMI2 x64            :  3175.97    451    69.1  172   279     0   311.5   61.9  17.23  3023.64  17.46   38.3
   4 Ethereal 12.25 PEXT x64         :  3159.03    309    68.4  115   193     1   211.5   62.5  20.81  3015.32  17.22   36.8
   5 SlowChess BC 2.2 x64            :  3130.78    440    63.4  121   316     3   279.0   71.8  15.99  3028.09  17.48   38.1
   6 Xiphos 0.6 BMI2 x64             :  3124.34    396    62.6  103   290     3   248.0   73.2  16.68  3027.91  17.12   37.5
   7 Defenchess 2.2 POP x64          :  3111.56    320    60.9   76   238     6   195.0   74.4  17.45  3026.82  17.50   37.7
   8 Booot 6.4 POP x64               :  3108.41    501    61.5  115   386     0   308.0   77.0  14.31  3026.35  16.98   37.8
   9 rofChade 2.3 BMI x64            :  3106.97    297    59.1   67   217    13   175.5   73.1  17.82  3039.68  17.18   37.0
  10 Fire 7.1 POP x64                :  3101.56    382    58.4   69   308     5   223.0   80.6  15.21  3033.38  17.16   38.1
  11 Andscacs 0.95 BMI2 x64          :  3098.08    418    59.4   89   319    10   248.5   76.3  15.62  3025.10  17.38   38.4
  12 Schooner 2.2 SSE x64            :  3095.18    380    59.9   80   295     5   227.5   77.6  16.07  3021.52  17.10   37.1
  13 Laser 1.7 BMI2 x64              :  3094.24    396    58.7   71   323     2   232.5   81.6  15.15  3026.47  17.11   38.6
  14 Fizbo 2.0 BMI2 x64              :  3090.82    363    59.0   96   236    31   214.0   65.0  15.62  3024.13  17.40   38.0
  15 Fritz 17 (Ginkgo) x64           :  3078.39    379    56.1   65   295    19   212.5   77.8  14.99  3030.91  17.21   38.1
  16 Shredder 13 x64                 :  3061.41    373    53.1   43   310    20   198.0   83.1  14.71  3035.34  17.19   37.3
  17 RubiChess 1.7.3 x64             :  3060.61    338    53.1   40   279    19   179.5   82.5  15.81  3038.17  17.20   37.9
  18 Wasp 4.00 Modern x64            :  3052.24    315    53.5   34   269    12   168.5   85.4  15.52  3030.27  17.03   38.0
  19 Arasan 22.0 BMI2 x64            :  3041.70    352    51.8   44   277    31   182.5   78.7  15.54  3031.62  17.05   37.6
  20 Chiron 4 x64                    :  3039.96    325    51.5   43   249    33   167.5   76.6  15.17  3030.19  17.32   37.2
  21 Vajolet2 2.8 BMI2 x64           :  3037.02    318    50.8   31   261    26   161.5   82.1  15.42  3035.77  17.16   38.0
  22 Pedone 2.0 BMI2 x64             :  3028.60    335    50.6   43   253    39   169.5   75.5  15.48  3026.15  17.16   36.4
  23 GullChess 3.0 BMI2 x64          :  3021.22    331    45.3   21   258    52   150.0   77.9  16.14  3051.92  17.07   36.4
  24 Nirvanachess 2.4 POP x64        :  2996.07    390    43.7   16   309    65   170.5   79.2  15.18  3045.36  17.35   37.9
  25 Igel 2.5.0 BMI2 x64             :  2992.34    412    43.2   24   308    80   178.0   74.8  14.99  3044.39  17.10   38.1
  26 Demolito 2020-05-14 PEXT x64    :  2992.18    203    42.6   22   129    52    86.5   63.5  20.96  3048.31  17.04   35.2
  27 Critter 1.6a x64                :  2985.43    311    42.9   26   215    70   133.5   69.1  17.11  3039.47  17.22   37.4
  28 Nemorino 5.00 BMI2 x64          :  2984.34    313    41.7   25   211    77   130.5   67.4  17.93  3049.33  17.45   36.7
  29 Texel 1.07 BMI2 x64             :  2982.98    315    42.7   24   221    70   134.5   70.2  16.53  3042.63  17.25   38.1
  30 Protector 1.9.0 x64             :  2981.30    318    41.7   17   231    70   132.5   72.6  16.40  3047.63  17.47   36.4
  31 iCE 4.0 v853 Modern x64         :  2976.56    343    40.5   19   240    84   139.0   70.0  16.17  3049.49  17.12   38.0
  32 Equinox 3.30 x64                :  2969.67    326    38.0   15   218    93   124.0   66.9  18.20  3059.98  17.08   36.4
  33 Hannibal 1.7 x64                :  2962.97    286    36.7   10   190    86   105.0   66.4  18.68  3062.37  17.18   35.4
  34 Rodent IV 0.22 POP x64          :  2960.72    318    38.4   11   222    85   122.0   69.8  17.75  3051.55  16.99   36.8
  35 Fritz 16 (Rybka) x64            :  2960.01    388    37.1   18   252   118   144.0   64.9  16.69  3056.31  17.00   38.1
  36 Winter 0.8 x64                  :  2953.39    393    36.8   25   239   129   144.5   60.8  16.61  3051.95  16.98   37.3
  37 Monolith 2 PEXT x64             :  2916.30    375    33.2    4   241   130   124.5   64.3  18.31  3051.44  16.85   37.3
  38 Minic 2.33 x64                  :  2914.00    388    31.4   11   222   155   122.0   57.2  18.97  3057.50  16.90   36.1
  39 Senpai 2.0 BMI2 x64             :  2906.85    348    30.9   11   193   144   107.5   55.5  19.85  3057.97  17.06   36.7
  40 Combusken 1.2.0 x64             :  2883.66    396    26.6    8   195   193   105.5   49.2  20.89  3067.63  17.04   36.2
  41 SmarThink 1.98 AVX2 x64         :  2853.77    370    23.0   11   148   211    85.0   40.0  24.02  3069.67  16.98   36.0

White advantage = 40.50 +/- 2.09
Draw rate (equal opponents) = 96.17 % +/- 1.48

Games ended move 60-79

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 11 BMI2 x64           :  3477.88    731    92.1  617   113     1   673.5   15.5  33.41  2998.59  23.17   38.9
   2 Komodo 14.0 BMI2 x64            :  3346.86    721    84.5  507   204    10   609.0   28.3  26.23  3004.62  23.26   39.0
   3 Houdini 6.03 Pro x64            :  3325.89    721    82.8  477   240     4   597.0   33.3  25.27  3004.04  23.25   38.9
   4 Ethereal 12.25 PEXT x64         :  3294.55    507    80.9  338   144    25   410.0   28.4  29.54  2998.92  23.29   38.1
   5 SlowChess BC 2.2 x64            :  3248.52    614    76.9  374   196    44   472.0   31.9  24.48  3005.00  23.27   38.3
   6 Fire 7.1 POP x64                :  3220.34    580    74.0  322   214    44   429.0   36.9  25.95  3009.65  23.41   38.5
   7 Xiphos 0.6 BMI2 x64             :  3188.33    631    70.3  314   259    58   443.5   41.0  23.01  3016.42  23.53   38.9
   8 Booot 6.4 POP x64               :  3179.22    689    69.9  326   311    52   481.5   45.1  21.02  3009.49  23.47   38.8
   9 Schooner 2.2 SSE x64            :  3173.71    610    69.2  293   258    59   422.0   42.3  22.42  3014.51  23.43   38.3
  10 rofChade 2.3 BMI x64            :  3167.84    520    66.5  252   188    80   346.0   36.2  24.39  3022.85  23.66   37.8
  11 Laser 1.7 BMI2 x64              :  3142.60    558    64.0  239   236    83   357.0   42.3  22.88  3025.13  23.59   39.1
  12 Fritz 17 (Ginkgo) x64           :  3139.45    616    64.9  269   262    85   400.0   42.5  22.15  3017.92  23.44   38.6
  13 Andscacs 0.95 BMI2 x64          :  3135.77    605    61.6  240   265   100   372.5   43.8  21.99  3037.69  23.47   38.7
  14 Defenchess 2.2 POP x64          :  3123.54    503    61.6  204   212    87   310.0   42.1  23.88  3031.67  23.58   38.6
  15 Shredder 13 x64                 :  3104.60    565    59.8  214   248   103   338.0   43.9  22.81  3025.80  23.63   38.5
  16 RubiChess 1.7.3 x64             :  3086.44    564    57.4  210   227   127   323.5   40.2  22.20  3029.11  23.57   38.9
  17 Fizbo 2.0 BMI2 x64              :  3074.94    566    54.9  214   194   158   311.0   34.3  23.04  3035.69  23.63   38.7
  18 Arasan 22.0 BMI2 x64            :  3032.46    589    49.7  178   229   182   292.5   38.9  21.94  3040.24  23.76   38.3
  19 GullChess 3.0 BMI2 x64          :  3024.44    576    48.8  165   232   179   281.0   40.3  21.42  3042.01  23.62   38.9
  20 Wasp 4.00 Modern x64            :  3008.06    570    46.8  159   215   196   266.5   37.7  22.33  3042.16  23.54   38.9
  21 Pedone 2.0 BMI2 x64             :  2988.66    570    42.5  135   215   220   242.5   37.7  21.84  3056.18  23.74   38.5
  22 Fritz 16 (Rybka) x64            :  2986.64    550    42.2  137   190   223   232.0   34.5  22.75  3053.85  23.62   38.8
  23 Chiron 4 x64                    :  2982.34    562    44.1  132   232   198   248.0   41.3  21.87  3042.64  23.57   38.8
  24 Vajolet2 2.8 BMI2 x64           :  2955.05    590    40.4  131   215   244   238.5   36.4  21.98  3041.05  23.59   38.4
  25 Equinox 3.30 x64                :  2952.15    566    39.4  105   236   225   223.0   41.7  22.83  3046.72  23.52   39.1
  26 Igel 2.5.0 BMI2 x64             :  2951.17    611    39.6  116   252   243   242.0   41.2  21.16  3047.70  23.70   39.0
  27 Critter 1.6a x64                :  2949.82    565    38.5  116   203   246   217.5   35.9  22.27  3057.26  23.75   38.6
  28 Winter 0.8 x64                  :  2934.98    650    37.6  128   233   289   244.5   35.8  20.80  3047.69  23.64   38.8
  29 Demolito 2020-05-14 PEXT x64    :  2929.41    549    35.4  117   155   277   194.5   28.2  23.89  3059.83  23.70   38.5
  30 Nemorino 5.00 BMI2 x64          :  2921.64    579    35.8  117   181   281   207.5   31.3  23.21  3050.51  23.57   38.8
  31 iCE 4.0 v853 Modern x64         :  2909.56    642    34.3  114   213   315   220.5   33.2  21.50  3047.81  23.58   39.1
  32 Nirvanachess 2.4 POP x64        :  2906.42    577    33.3   77   230   270   192.0   39.9  22.94  3058.52  23.55   38.6
  33 Texel 1.07 BMI2 x64             :  2892.19    594    31.1   92   185   317   184.5   31.1  23.75  3068.90  23.73   38.2
  34 Protector 1.9.0 x64             :  2889.68    609    32.6   97   203   309   198.5   33.3  23.43  3048.23  23.65   39.0
  35 Hannibal 1.7 x64                :  2877.13    587    30.4   84   189   314   178.5   32.2  22.86  3054.80  23.61   38.3
  36 Minic 2.33 x64                  :  2873.61    644    30.4  103   186   355   196.0   28.9  22.80  3050.83  23.70   38.7
  37 Senpai 2.0 BMI2 x64             :  2853.61    592    28.0   75   181   336   165.5   30.6  24.22  3058.39  23.70   38.9
  38 Combusken 1.2.0 x64             :  2819.91    592    24.7   74   145   373   146.5   24.5  25.28  3055.56  23.67   38.7
  39 Rodent IV 0.22 POP x64          :  2811.51    618    22.7   50   181   387   140.5   29.3  24.40  3070.64  23.68   38.5
  40 Monolith 2 PEXT x64             :  2781.74    702    19.8   34   210   458   139.0   29.9  24.65  3068.56  23.56   38.6
  41 SmarThink 1.98 AVX2 x64         :  2772.31    641    19.7   57   138   446   126.0   21.5  24.95  3065.84  23.55   38.8

White advantage = 69.52 +/- 2.69
Draw rate (equal opponents) = 51.13 % +/- 0.62

Games ended move 80-99

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 11 BMI2 x64           :  3457.80    478    88.7  371   106     1   424.0   22.2  38.40  3045.24  26.70   38.5
   2 Houdini 6.03 Pro x64            :  3357.08    461    83.0  320   125    16   382.5   27.1  34.35  3035.12  26.77   38.4
   3 Komodo 14.0 BMI2 x64            :  3344.57    484    82.5  338   123    23   399.5   25.4  32.50  3031.57  26.67   38.2
   4 Ethereal 12.25 PEXT x64         :  3255.17    469    75.1  272   160    37   352.0   34.1  29.74  3029.43  27.02   38.8
   5 Fire 7.1 POP x64                :  3226.91    497    72.8  285   154    58   362.0   31.0  28.37  3022.63  26.79   38.6
   6 SlowChess BC 2.2 x64            :  3202.11    459    69.6  246   147    66   319.5   32.0  28.92  3032.77  26.96   38.3
   7 Booot 6.4 POP x64               :  3195.52    478    67.6  232   182    64   323.0   38.1  27.48  3039.05  26.96   38.6
   8 rofChade 2.3 BMI x64            :  3188.65    543    69.7  284   189    70   378.5   34.8  26.03  3018.24  26.83   38.6
   9 Xiphos 0.6 BMI2 x64             :  3167.53    498    66.2  235   189    74   329.5   38.0  26.98  3030.71  26.88   38.8
  10 Laser 1.7 BMI2 x64              :  3155.76    497    64.8  235   174    88   322.0   35.0  26.87  3036.48  27.00   38.7
  11 Fritz 17 (Ginkgo) x64           :  3129.99    474    62.1  211   167    96   294.5   35.2  25.96  3031.44  26.76   38.3
  12 RubiChess 1.7.3 x64             :  3127.53    517    61.1  232   168   117   316.0   32.5  24.52  3032.81  27.09   38.9
  13 Shredder 13 x64                 :  3125.21    535    62.0  245   173   117   331.5   32.3  24.45  3032.28  27.03   38.8
  14 Defenchess 2.2 POP x64          :  3114.01    479    59.3  194   180   105   284.0   37.6  26.41  3037.04  26.93   38.9
  15 Schooner 2.2 SSE x64            :  3108.95    466    57.4  168   199    99   267.5   42.7  26.48  3049.38  27.04   38.6
  16 Fizbo 2.0 BMI2 x64              :  3063.15    484    52.7  176   158   150   255.0   32.6  25.84  3045.37  26.98   38.8
  17 Andscacs 0.95 BMI2 x64          :  3052.24    491    52.7  173   172   146   259.0   35.0  25.61  3034.30  26.97   38.8
  18 GullChess 3.0 BMI2 x64          :  3046.45    474    54.6  178   162   134   259.0   34.2  26.47  3010.80  26.81   37.9
  19 Arasan 22.0 BMI2 x64            :  3020.14    508    47.6  163   158   187   242.0   31.1  25.06  3040.86  26.97   39.2
  20 Fritz 16 (Rybka) x64            :  3010.96    460    51.3  161   150   149   236.0   32.6  26.39  3012.31  26.79   38.5
  21 Winter 0.8 x64                  :  2995.07    427    47.9  138   133   156   204.5   31.1  26.73  3020.29  26.76   38.2
  22 Critter 1.6a x64                :  2966.46    524    42.5  144   157   223   222.5   30.0  24.63  3032.01  26.94   38.9
  23 Pedone 2.0 BMI2 x64             :  2965.96    536    41.8  121   206   209   224.0   38.4  25.13  3040.35  27.01   38.8
  24 Equinox 3.30 x64                :  2959.17    470    42.8  119   164   187   201.0   34.9  25.70  3027.50  26.94   38.9
  25 Chiron 4 x64                    :  2951.95    507    39.0  118   159   230   197.5   31.4  25.64  3050.04  27.16   38.6
  26 Igel 2.5.0 BMI2 x64             :  2950.88    485    41.0  119   160   206   199.0   33.0  25.02  3028.97  26.65   38.4
  27 Vajolet2 2.8 BMI2 x64           :  2944.28    514    37.9  106   178   230   195.0   34.6  24.99  3051.47  27.02   38.9
  28 Nirvanachess 2.4 POP x64        :  2942.72    516    39.9  125   162   229   206.0   31.4  25.41  3030.46  26.81   38.7
  29 Wasp 4.00 Modern x64            :  2923.44    523    34.0   93   170   260   178.0   32.5  25.52  3059.29  27.15   38.6
  30 iCE 4.0 v853 Modern x64         :  2917.19    516    36.8  104   172   240   190.0   33.3  24.08  3033.40  26.96   39.1
  31 Minic 2.33 x64                  :  2916.14    482    38.3   98   173   211   184.5   35.9  26.22  3023.28  26.75   38.4
  32 Hannibal 1.7 x64                :  2912.13    459    37.0   97   146   216   170.0   31.8  25.84  3029.88  26.82   38.5
  33 Demolito 2020-05-14 PEXT x64    :  2908.16    526    34.4  109   144   273   181.0   27.4  24.91  3043.32  26.84   38.7
  34 Protector 1.9.0 x64             :  2901.84    510    34.5   95   162   253   176.0   31.8  24.48  3032.88  26.79   38.8
  35 Nemorino 5.00 BMI2 x64          :  2901.70    492    35.0  100   144   248   172.0   29.3  26.06  3032.49  26.91   38.6
  36 Texel 1.07 BMI2 x64             :  2881.57    476    32.0   80   145   251   152.5   30.5  28.10  3040.57  26.87   38.8
  37 Combusken 1.2.0 x64             :  2879.35    466    32.2   69   162   235   150.0   34.8  27.31  3038.51  26.78   39.0
  38 Senpai 2.0 BMI2 x64             :  2878.31    475    31.7   81   139   255   150.5   29.3  26.03  3039.70  26.79   38.8
  39 SmarThink 1.98 AVX2 x64         :  2874.34    434    35.5   86   136   212   154.0   31.3  27.10  3008.11  26.65   38.2
  40 Monolith 2 PEXT x64             :  2798.62    476    24.5   55   123   298   116.5   25.8  29.01  3030.44  26.87   38.6
  41 Rodent IV 0.22 POP x64          :  2716.00    532    16.4   36   103   393    87.5   19.4  29.60  3037.81  26.65   38.3

White advantage = 62.90 +/- 3.06
Draw rate (equal opponents) = 42.12 % +/- 0.60

Games ended move 100-299

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Komodo 14.0 BMI2 x64            :  3188.89    344    68.2  133   203     8   234.5   59.0  21.42  3049.12  15.71   35.9
   2 Stockfish 11 BMI2 x64           :  3168.71    386    65.5  122   262     2   253.0   67.9  20.87  3047.68  15.83   35.5
   3 Houdini 6.03 Pro x64            :  3166.94    407    66.2  143   253    11   269.5   62.2  18.74  3044.37  15.84   37.1
   4 Ethereal 12.25 PEXT x64         :  3128.71    715    61.5  184   512    19   440.0   71.6  14.09  3042.56  16.10   38.4
   5 Fire 7.1 POP x64                :  3112.41    541    60.1  142   366    33   325.0   67.7  16.55  3038.65  15.77   37.5
   6 Shredder 13 x64                 :  3105.36    527    59.7  154   321    52   314.5   60.9  16.84  3033.84  15.61   38.0
   7 rofChade 2.3 BMI x64            :  3102.82    640    59.3  161   437    42   379.5   68.3  14.30  3034.92  15.84   38.9
   8 SlowChess BC 2.2 x64            :  3091.55    487    56.2  102   343    42   273.5   70.4  16.23  3044.70  15.76   37.0
   9 Xiphos 0.6 BMI2 x64             :  3085.71    475    56.0  103   326    46   266.0   68.6  16.92  3039.86  15.76   38.0
  10 Laser 1.7 BMI2 x64              :  3084.39    549    56.3  127   364    58   309.0   66.3  15.81  3037.93  15.72   38.8
  11 RubiChess 1.7.3 x64             :  3081.27    581    56.9  129   403    49   330.5   69.4  15.62  3031.99  15.64   38.4
  12 Booot 6.4 POP x64               :  3068.72    332    52.3   60   227    45   173.5   68.4  20.12  3049.08  15.63   34.8
  13 Defenchess 2.2 POP x64          :  3054.58    698    53.0  112   516    70   370.0   73.9  14.18  3032.82  15.72   38.6
  14 Fritz 17 (Ginkgo) x64           :  3051.68    531    51.7   90   369    72   274.5   69.5  15.46  3038.75  15.74   38.5
  15 Arasan 22.0 BMI2 x64            :  3051.37    551    53.0  105   374    72   292.0   67.9  15.37  3030.35  15.62   38.2
  16 Fizbo 2.0 BMI2 x64              :  3049.70    587    52.6   96   425    66   308.5   72.4  14.70  3031.66  15.76   38.7
  17 Winter 0.8 x64                  :  3044.96    530    52.5  100   357    73   278.5   67.4  15.27  3027.50  15.68   38.3
  18 Andscacs 0.95 BMI2 x64          :  3044.16    486    51.3   87   325    74   249.5   66.9  16.67  3035.88  15.54   37.5
  19 GullChess 3.0 BMI2 x64          :  3030.16    619    50.2   87   448    84   311.0   72.4  15.00  3029.84  15.64   38.5
  20 Schooner 2.2 SSE x64            :  3023.43    544    47.5   65   387    92   258.5   71.1  15.34  3039.29  15.75   38.0
  21 Nirvanachess 2.4 POP x64        :  3022.75    517    49.8   82   351    84   257.5   67.9  16.39  3024.90  15.45   37.3
  22 Fritz 16 (Rybka) x64            :  3018.41    602    48.7   81   424    97   293.0   70.4  14.67  3028.76  15.64   38.5
  23 Vajolet2 2.8 BMI2 x64           :  3015.58    578    48.2   77   403    98   278.5   69.7  14.93  3029.12  15.58   38.1
  24 Minic 2.33 x64                  :  2999.54    486    46.7   57   340    89   227.0   70.0  15.86  3024.72  15.57   37.5
  25 SmarThink 1.98 AVX2 x64         :  2998.30    555    47.2   76   372   107   262.0   67.0  15.64  3020.95  15.56   37.3
  26 Nemorino 5.00 BMI2 x64          :  2995.41    616    45.9   81   403   132   282.5   65.4  14.34  3026.16  15.65   38.1
  27 Equinox 3.30 x64                :  2993.48    638    45.1   66   444   128   288.0   69.6  14.39  3029.22  15.68   38.7
  28 Critter 1.6a x64                :  2989.25    600    44.8   67   404   129   269.0   67.3  14.98  3027.49  15.59   38.0
  29 Pedone 2.0 BMI2 x64             :  2987.27    559    44.3   63   369   127   247.5   66.0  15.60  3029.22  15.56   37.8
  30 Chiron 4 x64                    :  2986.47    606    43.7   65   400   141   265.0   66.0  14.75  3031.31  15.68   38.7
  31 Igel 2.5.0 BMI2 x64             :  2985.32    492    44.2   55   325   112   217.5   66.1  16.57  3027.57  15.58   37.6
  32 iCE 4.0 v853 Modern x64         :  2983.27    499    44.3   55   332   112   221.0   66.5  16.24  3026.33  15.60   37.9
  33 Protector 1.9.0 x64             :  2982.69    563    43.7   55   382   126   246.0   67.9  15.02  3030.04  15.67   38.8
  34 Combusken 1.2.0 x64             :  2982.52    546    45.6   53   392   101   249.0   71.8  15.34  3016.82  15.53   36.5
  35 Wasp 4.00 Modern x64            :  2981.13    592    43.5   56   403   133   257.5   68.1  15.36  3029.65  15.60   37.9
  36 Senpai 2.0 BMI2 x64             :  2978.97    585    44.3   53   412   120   259.0   70.4  14.97  3021.36  15.55   37.3
  37 Hannibal 1.7 x64                :  2975.69    668    43.3   69   440   159   289.0   65.9  14.25  3025.79  15.63   37.8
  38 Demolito 2020-05-14 PEXT x64    :  2971.84    722    42.6   63   489   170   307.5   67.7  13.75  3027.38  15.69   38.7
  39 Texel 1.07 BMI2 x64             :  2970.49    615    42.8   67   393   155   263.5   63.9  15.04  3021.85  15.46   37.0
  40 Monolith 2 PEXT x64             :  2954.60    447    41.3   29   311   107   184.5   69.6  17.34  3019.81  15.42   36.3
  41 Rodent IV 0.22 POP x64          :  2926.52    532    37.3   32   333   167   198.5   62.6  17.23  3022.15  15.46   37.3

White advantage = 19.00 +/- 1.83
Draw rate (equal opponents) = 73.58 % +/- 0.45

Please do not look at Elo.
Have a look at the ranking and for an example: Rybka or Wasp.
Wasp is very strong in the first phase of the game, because WAsp has a lot of quick wins, a great move average.
And you can see the problem Wasp has later in the games ...

Again, stats with the number of pieces on the board are OK, but often not good enough.

2-6 pieces on board

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 11 BMI2 x64           :  3354.47    866    83.6  585   278     3   724.0   32.1  22.31  3036.06  16.34   39.3
   2 Komodo 14.0 BMI2 x64            :  3278.48    834    76.9  481   320    33   641.0   38.4  20.13  3040.48  16.37   39.0
   3 Houdini 6.03 Pro x64            :  3249.86    898    74.3  460   415    23   667.5   46.2  18.83  3038.52  16.44   39.3
   4 SlowChess BC 2.2 x64            :  3202.14    912    71.0  469   357    86   647.5   39.1  17.75  3026.76  16.35   39.4
   5 Ethereal 12.25 PEXT x64         :  3200.48   1007    69.6  448   506    53   701.0   50.2  16.87  3037.09  16.46   39.4
   6 Fire 7.1 POP x64                :  3196.13    973    70.7  486   403    84   687.5   41.4  17.34  3025.98  16.34   39.4
   7 rofChade 2.3 BMI x64            :  3172.99   1102    68.8  506   505    91   758.5   45.8  16.06  3018.87  16.33   39.4
   8 Booot 6.4 POP x64               :  3168.61    857    65.9  382   366   109   565.0   42.7  18.62  3036.12  16.37   38.9
   9 Xiphos 0.6 BMI2 x64             :  3159.95   1010    66.0  432   470   108   667.0   46.5  16.19  3028.64  16.34   39.5
  10 Shredder 13 x64                 :  3133.73   1035    64.1  454   418   163   663.0   40.4  16.01  3023.51  16.29   39.5
  11 Laser 1.7 BMI2 x64              :  3122.82   1013    61.5  395   455   163   622.5   44.9  15.73  3034.90  16.44   39.6
  12 Fritz 17 (Ginkgo) x64           :  3101.18    956    57.8  330   446   180   553.0   46.7  16.19  3041.04  16.48   39.5
  13 Defenchess 2.2 POP x64          :  3098.12   1019    58.5  339   515   165   596.5   50.5  16.24  3031.78  16.31   39.3
  14 RubiChess 1.7.3 x64             :  3096.16    994    58.4  364   432   198   580.0   43.5  16.09  3028.02  16.40   39.5
  15 Schooner 2.2 SSE x64            :  3077.96   1008    55.1  305   500   203   555.0   49.6  15.59  3038.41  16.45   39.3
  16 Fizbo 2.0 BMI2 x64              :  3048.04    895    51.1  269   376   250   457.0   42.0  16.31  3040.79  16.42   39.6
  17 GullChess 3.0 BMI2 x64          :  3047.65    975    53.2  290   458   227   519.0   47.0  15.90  3025.16  16.36   39.2
  18 Andscacs 0.95 BMI2 x64          :  3040.56    905    50.7  255   408   242   459.0   45.1  16.50  3036.71  16.45   39.6
  19 Arasan 22.0 BMI2 x64            :  3033.71    998    50.1  286   428   284   500.0   42.9  15.57  3034.64  16.37   39.6
  20 Fritz 16 (Rybka) x64            :  3024.25   1018    51.6  278   494   246   525.0   48.5  15.24  3018.83  16.36   39.0
  21 Winter 0.8 x64                  :  2993.04    893    45.6  214   387   292   407.5   43.3  17.09  3031.50  16.36   39.6
  22 Equinox 3.30 x64                :  2990.16    953    45.0  214   430   309   429.0   45.1  15.49  3030.09  16.36   39.6
  23 Chiron 4 x64                    :  2976.23   1052    43.2  222   465   365   454.5   44.2  15.10  3032.48  16.40   39.7
  24 Critter 1.6a x64                :  2975.73    999    43.3  221   424   354   433.0   42.4  15.28  3032.56  16.37   39.5
  25 Pedone 2.0 BMI2 x64             :  2974.27   1019    43.1  216   446   357   439.0   43.8  14.96  3030.38  16.35   39.3
  26 Igel 2.5.0 BMI2 x64             :  2968.13    923    43.6  199   407   317   402.5   44.1  16.65  3023.38  16.29   39.0
  27 Vajolet2 2.8 BMI2 x64           :  2964.62   1031    40.8  193   456   382   421.0   44.2  16.10  3037.34  16.40   39.5
  28 Nemorino 5.00 BMI2 x64          :  2962.75    998    42.6  227   396   375   425.0   39.7  15.79  3024.92  16.33   39.3
  29 Hannibal 1.7 x64                :  2949.84   1057    41.6  199   481   377   439.5   45.5  14.68  3022.63  16.35   39.3
  30 Nirvanachess 2.4 POP x64        :  2949.66   1010    40.2  187   438   385   406.0   43.4  15.79  3029.18  16.39   39.5
  31 Minic 2.33 x64                  :  2936.86    989    39.4  170   439   380   389.5   44.4  16.00  3024.34  16.31   39.3
  32 Wasp 4.00 Modern x64            :  2934.16   1043    37.0  166   440   437   386.0   42.2  16.09  3040.32  16.38   39.5
  33 Demolito 2020-05-14 PEXT x64    :  2929.15   1171    37.7  185   513   473   441.5   43.8  14.70  3031.88  16.39   39.7
  34 Protector 1.9.0 x64             :  2927.02   1050    37.6  169   452   429   395.0   43.0  16.00  3029.05  16.31   39.4
  35 SmarThink 1.98 AVX2 x64         :  2926.32    886    38.9  165   359   362   344.5   40.5  16.70  3018.02  16.22   38.9
  36 Senpai 2.0 BMI2 x64             :  2918.87   1013    36.4  148   442   423   369.0   43.6  16.13  3031.62  16.32   39.4
  37 Texel 1.07 BMI2 x64             :  2913.10   1084    35.4  148   471   465   383.5   43.5  15.29  3030.41  16.29   39.4
  38 iCE 4.0 v853 Modern x64         :  2900.59   1042    33.4  154   389   499   348.5   37.3  15.75  3035.59  16.40   39.6
  39 Combusken 1.2.0 x64             :  2890.41   1031    34.3  136   436   459   354.0   42.3  15.49  3024.26  16.31   39.1
  40 Monolith 2 PEXT x64             :  2843.18   1003    27.7   76   404   523   278.0   40.3  17.27  3029.30  16.27   39.2
  41 Rodent IV 0.22 POP x64          :  2803.62   1046    23.8   78   341   627   248.5   32.6  17.81  3032.03  16.29   39.3

White advantage = 54.10 +/- 1.94
Draw rate (equal opponents) = 53.40 % +/- 0.42

7-12 pieces on board

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 11 BMI2 x64           :  3401.53    837    88.6  647   189     1   741.5   22.6  25.37  3009.52  18.50   39.0
   2 Komodo 14.0 BMI2 x64            :  3326.32    861    84.0  592   262     7   723.0   30.4  21.30  3006.92  18.55   38.8
   3 Houdini 6.03 Pro x64            :  3325.23    826    83.9  568   250     8   693.0   30.3  21.72  3006.70  18.57   39.0
   4 Ethereal 12.25 PEXT x64         :  3247.02    711    77.1  414   268    29   548.0   37.7  21.04  3009.28  18.64   38.7
   5 Fire 7.1 POP x64                :  3174.83    697    67.7  302   340    55   472.0   48.8  17.96  3026.43  18.85   38.9
   6 SlowChess BC 2.2 x64            :  3174.81    712    69.0  337   309    66   491.5   43.4  19.45  3018.85  18.79   39.0
   7 Booot 6.4 POP x64               :  3157.90    741    66.8  300   390    51   495.0   52.6  18.49  3021.35  18.77   38.7
   8 Xiphos 0.6 BMI2 x64             :  3155.47    699    65.9  293   335    71   460.5   47.9  19.30  3028.15  18.84   39.1
   9 Schooner 2.2 SSE x64            :  3154.51    678    66.7  276   352    50   452.0   51.9  19.59  3019.11  18.68   38.9
  10 Andscacs 0.95 BMI2 x64          :  3142.65    772    64.1  303   384    85   495.0   49.7  17.74  3029.94  18.82   39.3
  11 Laser 1.7 BMI2 x64              :  3134.73    648    63.4  242   338    68   411.0   52.2  19.87  3023.98  18.70   39.2
  12 Fritz 17 (Ginkgo) x64           :  3127.75    717    63.6  277   358    82   456.0   49.9  18.66  3020.03  18.74   39.3
  13 rofChade 2.3 BMI x64            :  3116.78    624    60.6  235   286   103   378.0   45.8  19.32  3033.33  18.80   38.5
  14 Defenchess 2.2 POP x64          :  3096.81    676    58.9  222   352   102   398.0   52.1  18.40  3027.32  18.98   38.5
  15 RubiChess 1.7.3 x64             :  3095.30    686    58.2  223   353   110   399.5   51.5  17.77  3034.51  18.84   38.8
  16 Fizbo 2.0 BMI2 x64              :  3094.35    832    57.9  276   412   144   482.0   49.5  16.07  3031.84  18.90   39.6
  17 Shredder 13 x64                 :  3084.51    656    55.3  190   346   120   363.0   52.7  17.86  3044.79  18.97   38.7
  18 Arasan 22.0 BMI2 x64            :  3033.33    688    50.6  185   326   177   348.0   47.4  16.90  3032.66  18.97   39.0
  19 Wasp 4.00 Modern x64            :  3024.57    646    49.8  152   339   155   321.5   52.5  18.46  3030.80  18.87   39.0
  20 GullChess 3.0 BMI2 x64          :  3020.57    709    47.0  155   357   197   333.5   50.4  16.94  3047.56  18.95   38.9
  21 Pedone 2.0 BMI2 x64             :  2991.99    710    43.3  134   347   229   307.5   48.9  17.64  3051.74  18.99   38.4
  22 Vajolet2 2.8 BMI2 x64           :  2990.89    659    45.0  138   317   204   296.5   48.1  17.89  3038.61  18.84   39.1
  23 Chiron 4 x64                    :  2977.90    668    41.5  113   328   227   277.0   49.1  18.28  3051.26  19.03   39.0
  24 iCE 4.0 v853 Modern x64         :  2971.93    675    42.4  129   314   232   286.0   46.5  18.03  3041.05  18.88   38.9
  25 Nirvanachess 2.4 POP x64        :  2963.85    689    40.1  107   339   243   276.5   49.2  18.31  3049.95  18.83   38.8
  26 Fritz 16 (Rybka) x64            :  2963.48    662    39.0  116   285   261   258.5   43.1  18.70  3054.85  18.91   38.3
  27 Igel 2.5.0 BMI2 x64             :  2959.18    743    38.5  109   354   280   286.0   47.6  17.25  3054.90  18.91   38.9
  28 Winter 0.8 x64                  :  2954.97    819    39.6  155   339   325   324.5   41.4  16.99  3042.42  18.90   39.5
  29 Critter 1.6a x64                :  2949.44    717    39.1  120   320   277   280.0   44.6  17.55  3043.41  18.92   38.7
  30 Demolito 2020-05-14 PEXT x64    :  2947.19    656    37.6  119   255   282   246.5   38.9  18.34  3051.95  18.88   39.2
  31 Equinox 3.30 x64                :  2930.11    742    37.1   89   372   281   275.0   50.1  17.72  3043.49  18.90   39.3
  32 Texel 1.07 BMI2 x64             :  2928.59    666    35.1  107   254   305   234.0   38.1  18.66  3059.88  18.98   38.6
  33 Protector 1.9.0 x64             :  2919.80    682    33.9   87   288   307   231.0   42.2  18.54  3055.71  18.85   39.0
  34 Nemorino 5.00 BMI2 x64          :  2912.85    715    33.3   85   306   324   238.0   42.8  18.56  3054.84  18.81   39.0
  35 Minic 2.33 x64                  :  2886.37    726    30.5   94   255   377   221.5   35.1  18.39  3055.25  18.92   39.0
  36 Combusken 1.2.0 x64             :  2885.27    630    29.6   63   247   320   186.5   39.2  20.24  3056.93  18.85   38.7
  37 Hannibal 1.7 x64                :  2876.36    721    28.6   57   298   366   206.0   41.3  18.44  3059.23  18.89   38.9
  38 Senpai 2.0 BMI2 x64             :  2864.59    702    28.8   67   270   365   202.0   38.5  19.47  3047.82  18.75   39.1
  39 Rodent IV 0.22 POP x64          :  2847.39    683    26.2   48   262   373   179.0   38.4  20.92  3059.47  18.83   38.6
  40 Monolith 2 PEXT x64             :  2817.01    732    23.3   43   255   434   170.5   34.8  20.92  3059.49  18.82   38.6
  41 SmarThink 1.98 AVX2 x64         :  2806.84    865    22.5   59   271   535   194.5   31.3  18.02  3056.92  18.89   39.1

White advantage = 53.63 +/- 2.15
Draw rate (equal opponents) = 61.03 % +/- 0.55

13-32 pieces on board

Code: Select all

   # Player                          :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 11 BMI2 x64           :  3184.87    297    70.4  121   176     0   209.0   59.3  21.67  3028.11  14.48   36.6
   2 Houdini 6.03 Pro x64            :  3139.87    276    63.8   76   200     0   176.0   72.5  18.87  3037.25  14.32   36.2
   3 Komodo 14.0 BMI2 x64            :  3129.98    305    62.5   77   227     1   190.5   74.4  17.94  3036.55  14.52   37.2
   4 Ethereal 12.25 PEXT x64         :  3098.01    282    58.3   47   235     0   164.5   83.3  16.05  3036.92  14.26   37.4
   5 Booot 6.4 POP x64               :  3077.66    402    56.2   51   350     1   226.0   87.1  12.08  3033.36  14.03   38.0
   6 SlowChess BC 2.2 x64            :  3076.41    376    54.5   37   336     3   205.0   89.4  12.15  3041.58  14.40   37.4
   7 Laser 1.7 BMI2 x64              :  3074.55    339    55.2   35   304     0   187.0   89.7  12.55  3036.24  14.03   37.8
   8 Xiphos 0.6 BMI2 x64             :  3074.26    291    54.8   30   259     2   159.5   89.0  13.65  3038.96  14.03   36.5
   9 Fire 7.1 POP x64                :  3071.61    330    54.4   30   299     1   179.5   90.6  12.43  3038.64  14.07   36.7
  10 rofChade 2.3 BMI x64            :  3066.56    274    52.2   23   240    11   143.0   87.6  13.61  3050.59  14.21   34.1
  11 Defenchess 2.2 POP x64          :  3065.67    305    53.9   25   279     1   164.5   91.5  12.72  3037.64  14.19   37.2
  12 Andscacs 0.95 BMI2 x64          :  3064.23    323    54.3   31   289     3   175.5   89.5  13.08  3031.90  14.26   38.1
  13 Fizbo 2.0 BMI2 x64              :  3060.79    273    54.8   37   225    11   149.5   82.4  14.17  3028.29  14.11   37.9
  14 Schooner 2.2 SSE x64            :  3060.51    314    53.7   25   287     2   168.5   91.4  13.24  3034.05  14.18   37.4
  15 RubiChess 1.7.3 x64             :  3058.90    320    53.1   24   292     4   170.0   91.3  13.01  3037.90  14.18   37.6
  16 Fritz 17 (Ginkgo) x64           :  3054.31    327    52.8   28   289    10   172.5   88.4  12.08  3032.38  14.05   37.7
  17 Wasp 4.00 Modern x64            :  3048.34    311    52.4   24   278     9   163.0   89.4  11.66  3032.29  14.06   38.0
  18 Arasan 22.0 BMI2 x64            :  3044.70    314    51.3   19   284    11   161.0   90.4  11.93  3036.48  13.89   37.3
  19 Shredder 13 x64                 :  3043.12    309    50.5   12   288     9   156.0   93.2  11.15  3037.03  14.05   37.8
  20 Chiron 4 x64                    :  3042.86    280    52.3   23   247    10   146.5   88.2  12.62  3027.92  13.97   37.5
  21 Vajolet2 2.8 BMI2 x64           :  3032.60    310    50.3   14   284    12   156.0   91.6  11.65  3032.07  13.94   37.4
  22 Pedone 2.0 BMI2 x64             :  3030.00    271    50.6   12   250     9   137.0   92.3  12.59  3027.46  14.18   36.7
  23 Winter 0.8 x64                  :  3023.05    288    48.6   22   236    30   140.0   81.9  12.70  3033.69  14.26   37.3
  24 iCE 4.0 v853 Modern x64         :  3018.13    283    48.1    9   254    20   136.0   89.8  13.35  3033.31  14.21   37.3
  25 Nirvanachess 2.4 POP x64        :  3017.62    301    47.7    6   275    20   143.5   91.4  12.81  3035.23  14.30   38.2
  26 Protector 1.9.0 x64             :  3014.62    268    47.4    8   238    22   127.0   88.8  13.42  3035.21  14.33   36.1
  27 GullChess 3.0 BMI2 x64          :  3013.91    316    47.0    6   285    25   148.5   90.2  12.36  3035.56  14.00   37.1
  28 Demolito 2020-05-14 PEXT x64    :  3012.96    173    47.1    7   149    17    81.5   86.1  16.70  3033.56  14.21   36.3
  29 Texel 1.07 BMI2 x64             :  3005.92    250    47.0    8   219    23   117.5   87.6  14.18  3029.76  14.31   36.5
  30 Critter 1.6a x64                :  3003.12    284    45.6   12   235    37   129.5   82.7  13.24  3034.88  14.04   37.3
  31 Nemorino 5.00 BMI2 x64          :  3001.48    287    45.1   11   237    39   129.5   82.6  14.20  3037.26  14.55   37.8
  32 Rodent IV 0.22 POP x64          :  2998.27    271    44.6    3   236    32   121.0   87.1  14.10  3036.67  13.96   37.3
  33 Hannibal 1.7 x64                :  2996.84    222    43.7    4   186    32    97.0   83.8  16.16  3042.30  13.94   36.6
  34 Igel 2.5.0 BMI2 x64             :  2996.23    334    44.3    6   284    44   148.0   85.0  12.42  3037.73  14.31   37.1
  35 Equinox 3.30 x64                :  2989.96    305    43.3    2   260    43   132.0   85.2  14.52  3039.23  13.94   37.4
  36 Monolith 2 PEXT x64             :  2987.54    265    43.8    3   226    36   116.0   85.3  14.85  3035.52  14.09   37.3
  37 Minic 2.33 x64                  :  2978.23    285    41.6    5   227    53   118.5   79.6  15.27  3040.30  14.05   37.1
  38 Senpai 2.0 BMI2 x64             :  2961.78    285    39.1    5   213    67   111.5   74.7  16.58  3042.46  14.14   37.4
  39 Fritz 16 (Rybka) x64            :  2958.21    320    38.0    3   237    80   121.5   74.1  17.64  3043.45  13.84   37.8
  40 SmarThink 1.98 AVX2 x64         :  2939.62    249    35.3    6   164    79    88.0   65.9  19.77  3047.59  13.91   36.4
  41 Combusken 1.2.0 x64             :  2917.67    339    32.6    5   211   123   110.5   62.2  19.28  3046.92  14.10   37.2

White advantage = 18.36 +/- 1.67
Draw rate (equal opponents) = 100.00 % +/- 0.62

Have a look again in the results from Wasp and in the ranking ... please not looking on Elo.

With other words:
You can see a lot for engines higher as 3000 Elo.
But you are right, the higher the Elo the more complicated to create statistics. Because we all are not strong enough in playing chess. We can working with many small helps, what I wrote before to the pawn structures.

Back to the examples:
I am working 3 or 4 days for combinations between quantity of pieces on board and move-average for create better stats. But all in all ... the final results for looking are end of the day not clearly better.

The combination from all ...
Most of available engines today have all the same strength ...
The transposition into endgames.
But only a small group of engines are very strong in the first playing phase.

I am thinking that strongest engines can play the transposition into endgame with around 3800 Elo. But the strongest engines can not play the first phase after opening books with 3800 Elo. The strength is perhaps 3200 Elo. But different engines, like Uralochka on high niveau can play the first playing phase with maybe 3400 Elo.

And the first playing phase is most interesting for humans because most like to look in fast wins.

Back to your posting, your last paragraph:
If such a strong engine like Stockfish have not more fast wins as Uralochka it must be a reason for it.

You are the programmer:
I am thinking often the forward pruning is to high and attacking moves oversearch.
Better ... not find in deapth 30 and not find in deapth 40 is often the final results with super strong engines.

Sure, maybe the engine is stronger with an higher pruning but the analyze results with many pieces on board are often questionable. Not sure I am right or not.

So, if I start an "over-night" analyze with Wasp ... Wasp really find often more as clearly stronger engines. Sure, the programmer can create a stronger Wasp for time-controls we are using for testing ... but to which price.

Best
Frank