Ethereal & Schooner seems to be "Big Monsters"

[Frank Quisinsky]

Hi there,

To Winter ...
At the moment a very high draw quote, 8 won games.

Engine is playing not in eye-catching manner.
Normely computer chess for the strength in Elo!

I need more rounds / games to find out a bit about Winter ... for the moment after 80 of 2000 games (round 02/50) not possible to give a good answere from my site.

Maybe the very nice Winter games will come with the next rounds.

Best
Frank

Code: Select all

  FCP Tourney-2020
  ----------------

  Date           : June 02nd, 2020 (11:25)
  Games          : 1.640
  Version        : 2.1, after R02
  Elo            : GullChess 3.0 BMI2 x64 = 3035
  Conditions     : http://www.amateurschach.de/main/_wasp-beta-test.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 11 BMI2 x64           :  3332.23     80    83.1    54    25     1    66.5   31.3   69.76  3025.18  51.01     80.7
  02. Komodo 14.0 BMI2 x64            :  3294.24     80    80.0    48    32     0    64.0   40.0   61.60  3026.13  51.22     79.6
  03. Houdini 6.03 Pro x64            :  3266.56     80    77.5    44    36     0    62.0   45.0   61.79  3026.82  51.21     85.7
  04. Ethereal 12.25 PEXT x64         :  3198.86     80    70.6    36    41     3    56.5   51.3   54.45  3028.51  51.40     89.7
  05. Fire 7.1 POP x64                :  3187.53     80    69.4    35    41     4    55.5   51.3   55.15  3028.80  51.38     87.7
  06. SlowChess BC 2.2 x64            :  3170.94     80    67.5    32    44     4    54.0   55.0   54.55  3029.21  51.39     82.2
  06. rofChade 2.3 BMI x64            :  3170.94     80    67.5    32    44     4    54.0   55.0   52.68  3029.21  51.44     86.3
  08. Laser 1.7 BMI2 x64              :  3144.25     80    64.4    30    43     7    51.5   53.8   50.28  3029.88  51.50     86.2
  09. Xiphos 0.6 BMI2 x64             :  3133.85     80    63.1    31    39    10    50.5   48.8   50.13  3030.14  51.51     87.0
  10. RubiChess 1.7.3 x64             :  3123.58     80    61.9    27    45     8    49.0   56.3   51.30  3030.40  51.48     90.1
  -------------------------------------------------------------------------------------------------------------------------------
  11. Fritz 17 (Ginkgo) x64           :  3113.43     80    60.6    25    47     8    48.5   58.8   50.20  3030.65  51.50     85.7
  12. Booot 6.4 POP x64               :  3108.40     80    60.0    24    48     8    48.0   60.0   50.50  3030.78  51.50     75.9
  12. Defenchess 2.2 POP x64          :  3108.40     80    60.0    26    44    10    48.0   55.0   49.11  3030.78  51.53     84.0
  14. Shredder 13 x64                 :  3098.40     80    58.8    24    46    10    47.0   57.5   51.20  3031.03  51.48     86.7
  15. Schooner 2.2 SSE x64            :  3083.57     80    56.9    22    47    11    45.5   58.8   49.66  3031.40  51.52     88.4
  16. Andscacs 0.95 BMI2 x64          :  3078.66     80    56.3    24    42    14    45.0   52.5   48.48  3031.52  51.55     87.2
  17. Fizbo 2.0 BMI2 x64              :  3039.82     80    51.3    21    40    19    41.0   50.0   46.13  3032.49  51.61     87.0
  18. Arasan 22.0 BMI2 x64            :  3035.00     80    50.6    13    55    12    40.5   68.8   48.40  3032.61  51.55     89.9
  18. GullChess 3.0 BMI2 x64          :  3035.00     80    50.6    18    45    17    40.5   56.3   47.80  3032.61  51.56     96.2
  20. Nirvanachess 2.4 POP x64        :  2986.77     80    44.4    11    49    20    35.5   61.3   48.33  3033.82  51.55     86.1
  -------------------------------------------------------------------------------------------------------------------------------
  21. Nemorino 5.00 BMI2 x64          :  2977.04     80    43.1    19    31    30    34.5   38.8   48.07  3034.06  51.56     89.6
  22. Wasp 4.00 Modern x64            :  2972.16     80    42.5    16    36    28    34.0   45.0   49.39  3034.18  51.52     91.7
  22. Vajolet2 2.8 BMI2 x64           :  2972.16     80    42.5    11    46    23    34.0   57.5   49.87  3034.18  51.51     92.3
  24. Winter 0.8 x64                  :  2967.27     80    41.9     8    51    21    33.5   63.8   48.30  3034.30  51.55     85.1
  24. Equinox 3.30 x64                :  2967.27     80    41.9    15    37    28    33.5   46.3   48.55  3034.30  51.55     84.5
  26. Chiron 4 x64                    :  2962.36     80    41.3    16    34    30    33.0   42.5   51.01  3034.43  51.48     92.2
  27. Fritz 16 (Rybka) x64            :  2957.43     80    40.6    14    37    29    32.5   46.3   48.03  3034.55  51.56     86.6
  27. Demolito 2020-05-14 PEXT x64    :  2957.43     80    40.6    10    45    25    32.5   56.3   49.81  3034.55  51.51     98.1
  29. iCE 4.0 v853 Modern x64         :  2952.49     80    40.0    12    40    28    32.0   50.0   47.35  3034.67  51.58     85.5
  29. Pedone 2.0 BMI2 x64             :  2952.49     80    40.0    11    42    27    32.0   52.5   48.95  3034.67  51.53     86.9
  -------------------------------------------------------------------------------------------------------------------------------
  31. Igel 2.5.0 BMI2 x64             :  2937.52     80    38.1     8    45    27    30.5   56.3   49.93  3035.05  51.51     75.4
  31. Hannibal 1.7 x64                :  2937.52     80    38.1    11    39    30    30.5   48.8   50.10  3035.05  51.51     96.7
  31. Critter 1.6a x64                :  2937.52     80    38.1    12    37    31    30.5   46.3   51.65  3035.05  51.47     87.3
  31. Senpai 2.0 BMI2 x64             :  2937.52     80    38.1    12    37    31    30.5   46.3   49.02  3035.05  51.53     84.5
  31. Texel 1.07 BMI2 x64             :  2937.52     80    38.1    11    39    30    30.5   48.8   50.91  3035.05  51.49     89.0
  31. Protector 1.9.0 x64             :  2937.52     80    38.1     8    45    27    30.5   56.3   50.73  3035.05  51.49     86.4
  37. Combusken 1.2.0 x64             :  2896.35     80    33.1     6    41    33    26.5   51.3   50.24  3036.08  51.50     85.8
  37. Minic 2.33 x64                  :  2896.35     80    33.1     8    37    35    26.5   46.3   51.21  3036.08  51.48     87.9
  39. Monolith 2 PEXT x64             :  2880.28     80    31.3     5    40    35    25.0   50.0   53.37  3036.48  51.42     82.2
  40. SmarThink 1.98 AVX2 x64         :  2863.76     80    29.4     5    37    38    23.5   46.3   53.60  3036.89  51.42     90.6
  -------------------------------------------------------------------------------------------------------------------------------
  41. Rodent IV 0.22 POP x64          :  2829.04     80    25.6     3    35    42    20.5   43.8   58.76  3037.76  51.29     83.3
  -------------------------------------------------------------------------------------------------------------------------------

  MoveAvg (average)           : 174.89 / 86.365
  White advantage             : 50.27  +/- 6.15
  Draw rate (equal opponents) : 66.08% +/- 1.61

  Games        :   1.640 (finished)
  White Wins   :     496 (30.2%)
  Black Wins   :     302 (18.4%)
  Draws        :     842 (51.3%)
  Unfinished   :       0

  White Perf.  : 55.9%
  Black Perf.  : 44.1%

  ECO A00-A99  :     222 games (13.5%)
  ECO B00-B99  :     412 games (25.1%)
  ECO C00-C99  :     336 games (20.5%)
  ECO D00-D99  :     381 games (23.2%)
  ECO E00-E99  :     289 games (17.6%)

[Frank Quisinsky]

Hi there,

all FEOBOS lines have a ranking.
We developed a ranking system for best opening lines.
So we can created to each of the ECO codes a test-set with best lines and so one.

We are speaking about this one ...

Position = 04222 of 33309 (database FEOBOS 20.1 Contempt 3).
Position = 04222 = Ranking 04222

4222 Pos 32971 B82
What 10 engines like to play: g4 a3 g4 a3 a3 a3 Qg3 a3 Qg3 Qg3

Or better ...
A jpg from our Excel with all the details about Pos 32971, ranking postion 4222



All this was a lot of work and need 4 years development time.
But now the perfect bug free book was born for testing engines and the topic "opening book" are over (for myself).
Again, if you have interest to learn more ... the complete files and docu are fee for all.

Best
Frank

[Frank Quisinsky]

B82 is one of the popular opening systems.
135 B82 lines inlcuded in FEOBOS 20.1 Contempt 3, 3 moves after main line ended
= 135 lines are equal ...
In FEOBOS are equal lines available only (if not an engine goes with advantage in a match).

The line for the Ethereal match are in our ranking system on place 17 of 135 of B82 lines.
17 best E82 line!

Best E82 line in database A00-E99 is on place 127 of 33.309

So I can give to each of the openings information in detail.
To cut B82 before move 12 isn't right because main line started with move 9 ...

All that and much more stats, the ranking system, the enigne analyszes can be found in the excel with 16 millions of formulas, most in macros or the Excel will never ended to load.

Best
Frank

We have ECO codes started with move 13 or 14 I believe.
If engines should play such lines and if we like to have more as one variant for such opening systems, we need around 3 moves after ECO codes ended. In FEOBOS are to all of the 500 ECO codes best lines included. 12 engines in teamwork created this project, so the book is bugfree and perfect for engines testing. For FEOBOS very important that engines can play in engine tournaments or test-runs all of the 500 ECO codes if the positions are equal. If not the 12 engines killed such moves in teamwork. Again, have a look in FEOBOS project. I worte a 60 sites documentation in German langauage about the ideas.

[Dr.Wael Deeb]

Hi, Dr. D.



In the last four years I am using to many times with the older chess computers and have much fun with the older chess computers.
Also using time for FEOBOS or in testing Wasp but I lost a bit the view for other engine developments.

It seems that you looking in detail!
You do that all the time if I remember

Magical is the right word by yourself I think.

Now I will see more of such things in the next 48 rounds of my tournament.
But sure here, Ethereal will produced much more of such material.

11 of 80 ... just amazing!

Best
Frank

Hi Frank,

Good to see you active and testing as I was away from the computer chess scene for a while now ....

Yes,I like to look at things in detail you remember right

Ethereal is amazing indeed and the other chess programmers should be afraid from the next release ....

Cheers

[Dr.Wael Deeb]

Hi Dr. D.,

if an engine have an average strenght in competition to Komodo or Stockfish for endgames but comes with a clearly more aggressive style in middlegames ... all is possible. So the Ethereal programmer is on the right way. But I am looking never for endgame strength. Thats is really boring for humans because all of TOP-200 engines have the Grandmaster level here, not in middlegames and openings.

If a programmer goes here an other own way he will find a lot of fans. But a lot of knowledge about statistics and fantasy must a programmer have. That can't be knowledge about programming only. So Ethereal have an own and interesting face.

Looking in the times Houdini are on first position a long time.
Robert have own ideas to the right points humans like, can be see in the style Houdini produced.
Later Stockfish have the same playing level as Houdini. It was nice to see that Houdini plays more aggressive as Stockfish with the same level. So Robert make differents things really better as the others.

Again, the secret are the pawn structures ... to 95% sure here!
Sure, the Ethereal programmer find out a lot about it.

Best
Frank

PS: Many of us like the LCO playing style. I replay a lot of LCO games. In endgames such engines are not more interesting as Komodo or Stockfish. The secret here is the middlegame. Humans like the middlegame and for such reason the style LCO produced is for humans interesting. In most of cases the great games by LCO are produced with aggressive pawns. Now we are in the year 2020. With lesser pieces on board engines are not to beat. The advantage from Stockfish from versions 8 to 9 to 10 to 11 are in the transposition into endgame, not in earlier middlegames. The other way is more interesting ... how I can make an engine stronger in earlier middlegames. Thats for humans much more interesting as the perfect strength in transposition into the endgame. If we find out a new engine comes with a nice middlegames we all will like that.

To the topic pawn structures it seems we can go two ways:

Way 1: We make the pawns more aggressive but we lost endgame strengths
Way 2: We try to hold the pawns on the right fields without to produced speculative attacking moves with pawns and the endgame strengths goes higher. The way Komodo goes after all I can see.

A combination from Way 1 and 2 ... not easy to find out in programming. Maybe the Ethereal programmer is here on the right way.

The same in the years Spark, Aristarch, Phalanx are available later Glaurung. We like such things a lot. Engines that can produced fast won games in middlegames are allways interesting. Thinking on TheKing in chess computer times.

Hi Frank,

I am a big fan of the pawn structures and the pawn storms that a player or a chess engine can start in the middle game ....

But as mentioned before,it's not easy to push your pawns in the early middle game because one minor mistake and you're done,bye-bye thank you for coming

Dr.D

P.S I kinda try a minor variations of the pawn storms when playing the Larsen opening but not so aggressively ....

[Dr.Wael Deeb]

Thanks for posting the games, Frank - and great to see you around and active again.

This should put to rest any talk of Ethereal being a boring, style-less engine.

Try to take a look at Winter, too, including previous versions. The programmer is a master chess player.

I've known about Schooner's pleasant style since it was 'just' a 2500-rated engine, and having a lot more strength hasn't really detracted from its style.

Winter 0.8 is nice chess engine and is now rated in my rating list 3170 Elo ....

I am not familiar with Schooner chess engine but will get to it in my testing ....

Cheers,
Dr.D

[Dr.Wael Deeb]

Hi there,

FEOBOS Book Lines ... all FEOBOS lines = 3 moves after ECO code ended.
If we used opening book up to 12 moves, different of the 500 codes are not included.
So we go an other way because all of the 500 lines engines should play.

All FEOBOS opening lines end positions (for Contempt 3 = over 33.000) checked by 12 engines with 1 minute analyses on 6 cores. The complete FEOBOS analyses by 12 engines, the Excel file about it (over 16 millions of formulas) can be found on my site.

Book is completly bugfree.
Documention, PGN, EPD and all other files around this project are available on my site.
Project time = 4 years:
Downloads at the moment: 17.322

For more information please download the files or have a look on FEOBOS detail page.

Best
Frank

I am aware of the FEOBOS project Frank and it's a fantastic project ....

Cheers,
Dr.D

[Frank Quisinsky]

I am a big fan of the pawn structures and the pawn storms that a player or a chess engine can start in the middle game ....

But as mentioned before,it's not easy to push your pawns in the early middle game because one minor mistake and you're done,bye-bye thank you for coming

That's the point!
To aggressive pawns and the endgame is lost.

Best example is Spark:
Can produced fantastic fast won games with aggressive pawns, great king safty but the endgame ... around 150-200 Elo weaker as other engines with the same strength. Hakkapelitta comes with around the same stats with the different to Spark that the king safty in middlegames are in additional the second problem (endgame the first).

With a good basic strength in endgames, maybe an engine lost perhaps only 100 Elo to Stockfish and Komodo and 150 Elo in transposition into endgames. But with the aggressive pawns much more nice games will be produced in middlegames. The 100 Elo lesser in endgames / 150 lesser in transposition into endgames are from my point of view (human view) not interesting, but the aggressive style in middlegames is very interesting. So, I have much more fun with the engines because most of humans I know with main interest computer chess ... like to see the sensation with many pieces on board.

Furthermore, engines stronger and more aggressive in middlegames produced better statistics to opening systems. One of the main points for human chess.

Most are looking on Elo only and this one I never will be understand.
Most boring point in times Engines are stronger as 2.850 Elo (best humans) are Elo from computer chess engines. Most interesting point is the style an engine can produced.

So, its time for new stats for the current group of TOP-40.
I have interest to learn more about engines in the year 2020 (again my point of view).

Best
Frank

[Dr.Wael Deeb]

I am a big fan of the pawn structures and the pawn storms that a player or a chess engine can start in the middle game ....

But as mentioned before,it's not easy to push your pawns in the early middle game because one minor mistake and you're done,bye-bye thank you for coming

That's the point!
To aggressive pawns and the endgame is lost.

Best example is Spark:
Can produced fantastic fast won games with aggressive pawns, great king safty but the endgame ... around 150-200 Elo weaker as other engines with the same strength. Hakkapelitta comes with around the same stats with the different to Spark that the king safty in middlegames are in additional the second problem (endgame the first).

With a good basic strength in endgames, maybe an engine lost perhaps only 100 Elo to Stockfish and Komodo and 150 Elo in transposition into endgames. But with the aggressive pawns much more nice games will be produced in middlegames. The 100 Elo lesser in endgames / 150 lesser in transposition into endgames are from my point of view (human view) not interesting, but the aggressive style in middlegames is very interesting. So, I have much more fun with the engines because most of humans I know with main interest computer chess ... like to see the sensation with many pieces on board.

Furthermore, engines stronger and more aggressive in middlegames produced better statistics to opening systems. One of the main points for human chess.

Most are looking on Elo only and this one I never will be understand.
Most boring point in times Engines are stronger as 2.850 Elo (best humans) are Elo from computer chess engines. Most interesting point is the style an engine can produced.

So, its time for new stats for the current group of TOP-40.
I have interest to learn more about engines in the year 2020 (again my point of view).

Best
Frank

So you are chasing the style not the chess engines brute force ....
Excellent approach Frank

Dr.D

[carldaman]

Hi there,

To Winter ...
At the moment a very high draw quote, 8 won games.

Engine is playing not in eye-catching manner.
Normely computer chess for the strength in Elo!

I need more rounds / games to find out a bit about Winter ... for the moment after 80 of 2000 games (round 02/50) not possible to give a good answere from my site.

Maybe the very nice Winter games will come with the next rounds.

Best
Frank

Winter v0.8 is a new hybrid that uses a neural net, and is less aggressive, unfortunately, than previous versions such as 0.7 or earlier.