BoCC -- Beauty of Computer Chess

[Rebel]

Nowadays millions of computer chess games are played each day either to create better datasets for neural networks or testing engine changes. And actually none of those millions of games is ever looked at because only the result matters, does the engine play stronger or not. And in those millions must be tens of thousands most interesting and beautiful games that will never see the daylight. I like to introduce a tool that filters large PGN collections and store the most interesting games in much smaller separate PGN's. Some examples :

....

Call for engines
​​If you know an engine with an emphasis on playing style please drop me a note and I might include it in my testing.

LINK

[Damir]

Hi Ed

Where is the tool?

[pohl4711]

Great news. I always said, it would be great for all computerchess-fans to have not just 1 EAS-tool, but instead 2 of these tools, both using a complete different internal approach (EAS-Tool: 100% eval-free, just looking at the moves, number of pieces and game-results, BoCC using engine-evals (and game-results, I presume) instead (+ an internal HCE for the king-attack algorithm (?))

Some thoughts and hints by me about your claims on your new website:

1)
"Our BoCC (Beauty of Computer Chess) tool takes a different approach, first of all the emphasis is not so much on the rating list but on the PGN output with scores and depths"
"However experience has learned me that playing the same 15.000 games on a faster computer or at a somewhat slower time control the first 3 places on BoCC can easily swap places, likewise for the 5 first places on EAS since its scores are so close.
Therefore what really matters (to me, that is) are the output PGN's for each engine, included in the download and enjoy the games. These are the kind of games I fell in love with chess as a teenager, the Morphy style, attack......!"

This is a little confusing to me, because my EAS-tool sorts out all sac-games and short-wins and put them in different small pgn-files (one file for each sac-category and one for the short wins), too. Additionally, I offer my IWS-Tool, which does this job without any math (much faster than the EAS-Tool) an builds just one output-file with great games, sorted by the sacs (highest sac on top of the list). And the IWS-Tool comes with 2 versions. One of them keeps the game-comments of the gamebase-file (so, if there are evals and depths of any engine stored in the comments, they still can be found in the games in the pgn-output-file).
https://www.sp-cc.de/files/interesting_ ... ch_tool.7z

2) I strongly recommend to add some percent-numbers to the output of your tool. Using 4-digit percent-numbers (12.34%) instead of huge score-numbers (with different length), makes the output way better to read for humans.

3) I very strongly recommend, to write an explanation text-file, where to explain, how the different parts of the BoCC-Tool work. Right now, the whole tool, especially the King-attack-algorithm, is a black-box. Just I did in the ReadMe of my EAS-Tool, where I explain exactly, what the tool does for each category (sacs, shorts, bad draws) and how the score is calculated out of the found games.
As a user, I want to know, how an EAS- or BoCC- score is build, otherwise I can not fully judge the results. Especially for engine-developers (Patricia author told me, how important it was for him, to fully understand, what the EAS-tool does exactly) this is very important.

4) I complete miss any sac-search in your tool? I find this quite strange for 2 reasons
a) sacs are the most important thing in exciting enginegames
b) because you have an internal pgn-parser (I dont!), it would be very easy for you, not only to find the sacs very fast, but also mark them in the pgn-comments, just directly behind the sac move. This would be very helpful for the users.

5) My IWS-Tool adds this to each found interesting game (this is very helpful for the user, I got several very positive feedbacks, so I recommend, you doing something similar):
Each game gets a new Annotator-Tag, so it is clear, which category the game belongs to. One of these 8 tags is added to each game-notation:
[Annotator "IWS-Tool: Queen Sacrifice found in this game"]
[Annotator "IWS-Tool: 5+ PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 4 PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 3 PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 2 PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 1 PawnUnit Sacrifice found in this game"]
[Annotator "IWS-Tool: Game ended before endgame (material)"]
[Annotator "IWS-Tool: Material imbalance found in this game"]

A game in the output-file looks like this (from my testrun of latest Obsidian 250706 dev), when I use the IWS-tool version, called:
Interesting_Wins_Search_Tool_V4.1_keep_comments.bat (mention the Annotator-Tag and the evals/depth-comments, still in the game)

[Event "UHO Ratinglist"]
[Site "?"]
[Date "2025.07.07"]
[Round "30"]
[White "Obsidian 250706 a512"]
[Black "RubiChess 250606 a512"]
[Result "1-0"]
[ECO "A45"]
[Opening "Queen's pawn game"]
[TimeControl "180+1"]
[Annotator "IWS-Tool: 5+ PawnUnits Sacrifice found in this game"]
[Termination "adjudication"]
[PlyCount "99"]
[GameDuration "00"]
[GameEndTime "2025-07-07T11.124 Mitteleuropäische Sommerzeit"]
[GameStartTime "2025-07-07T11.688 Mitteleuropäische Sommerzeit"]

1. d4 { book } 1... Nf6 { book } 2. Bf4 { book } 2... g6 { book } 3. Nc3 {
book } 3... Bg7 { book } 4. e4 { book } 4... d6 { book } 5. Qd2 { book }
5... Nc6 { book } 6. O-O-O { book } 6... O-O { book } 7. f3 { +1.12/22 13s
} 7... a6 { -1.22/24 34s } 8. Nge2 { +1.31/20 4.1s } 8... h5 { -0.96/22
4.7s } 9. h4 { +1.30/22 8.2s } 9... e6 { -1.38/23 23s } 10. Bg5 { +1.46/22
4.9s } 10... b5 { -1.27/21 2.7s } 11. g4 { +1.74/22 4.2s } 11... b4 {
-1.06/20 2.8s } 12. Na4 { +1.81/23 7.5s } 12... hxg4 { -1.52/20 4.3s } 13.
h5 { +1.94/20 2.8s } 13... Qe8 { -2.01/23 15s } 14. e5 { +2.41/20 2.9s }
14... dxe5 { -1.72/21 2.4s } 15. Bg2 { +2.77/20 3.4s } 15... exd4 {
-2.39/22 15s } 16. h6 { +4.26/20 2.9s } 16... Bh8 { -3.51/20 4.9s } 17. h7+
{ +4.35/20 4.5s } 17... Nxh7 { -3.85/18 1.3s } 18. Rxh7 { +4.66/19 3.2s }
18... Kxh7 { -4.56/17 1.8s } 19. Rh1+ { +4.64/22 4.9s } 19... Kg7 {
-4.39/19 2.8s } 20. fxg4 { +4.77/22 5.0s } 20... e5 { -4.57/20 3.7s } 21.
Bd5 { +5.01/20 3.6s } 21... Be6 { -4.91/21 7.7s } 22. Be7 { +5.05/19 3.8s }
22... f5 { -4.91/20 2.0s } 23. Bxf8+ { +5.08/21 5.3s } 23... Qxf8 {
-5.50/19 2.9s } 24. Bxc6 { +5.20/20 3.8s } 24... Rd8 { -5.35/22 8.5s } 25.
gxf5 { +5.20/20 3.9s } 25... Bxf5 { -5.36/18 1.7s } 26. Ng3 { +5.24/19 3.3s
} 26... d3 { -5.65/21 8.1s } 27. Nxf5+ { +5.22/19 2.9s } 27... Qxf5 {
-5.74/19 2.2s } 28. Nc5 { +5.25/19 2.4s } 28... Kf7 { -5.72/20 5.3s } 29.
cxd3 { +5.18/22 7.8s } 29... Qf4 { -5.50/20 1.9s } 30. Ba4 { +5.27/24 9.4s
} 30... g5 { -5.46/22 7.0s } 31. Rh6 { +5.42/21 3.0s } 31... Ke7 { -5.49/22
13s } 32. Bb3 { +5.37/21 3.2s } 32... Qxd2+ { -5.28/19 3.7s } 33. Kxd2 {
+5.44/19 2.1s } 33... e4 { -5.26/20 3.1s } 34. Re6+ { +5.51/17 3.6s } 34...
Kf8 { -5.31/20 2.7s } 35. Nxe4 { +5.50/22 7.1s } 35... a5 { -5.35/21 5.4s }
36. Nxg5 { +5.63/18 2.3s } 36... Bxb2 { -5.37/19 5.4s } 37. Kc2 { +5.88/21
6.9s } 37... Bc3 { -5.21/20 3.4s } 38. Rc6 { +5.83/21 3.3s } 38... Re8 {
-5.22/19 2.1s } 39. Ne4 { +5.90/21 2.3s } 39... Bd4 { -5.07/19 1.4s } 40.
Rxc7 { +6.62/21 1.8s } 40... a4 { -5.34/19 1.5s } 41. Bxa4 { +6.62/19 2.6s
} 41... b3+ { -5.85/19 2.9s } 42. Bxb3 { +7.00/18 1.6s } 42... Re7 {
-6.21/25 2.1s } 43. Rxe7 { +7.41/22 1.6s } 43... Kxe7 { -6.08/25 1.8s } 44.
a4 { +7.51/27 1.8s } 44... Kd7 { -6.90/24 3.8s } 45. a5 { +7.95/26 1.7s }
45... Kc6 { -6.66/20 1.2s } 46. Bc4 { +8.19/22 1.6s } 46... Bg1 { -7.12/22
1.7s } 47. Nc3 { +8.50/23 1.7s } 47... Ba7 { -7.13/19 1.00s } 48. a6 {
+9.13/24 4.1s } 48... Kb6 { -7.16/22 1.0s } 49. Nb5 { +8.05/29 3.1s } 49...
Kxa6 { -4.36/20 1.0s } 50. Nxa7+ { +7.90/27 2.5s, White wins by
adjudication: SyzygyTB } 1-0

[Rebel]

Hi Ed

Where is the tool?

A beta version can be found here : https://prodeo.actieforum.com/t1735-eat ... tool#18203

[Rebel]

Great news. I always said, it would be great for all computerchess-fans to have not just 1 EAS-tool, but instead 2 of these tools, both using a complete different internal approach (EAS-Tool: 100% eval-free, just looking at the moves, number of pieces and game-results, BoCC using engine-evals (and game-results, I presume) instead (+ an internal HCE for the king-attack algorithm (?))

Yes, for the king-attack evaluation I use HCE code, else it's impossible to filter meaningful games.

Some thoughts and hints by me about your claims on your new website:

1)"Our BoCC (Beauty of Computer Chess) tool takes a different approach, first of all the emphasis is not so much on the rating list but on the PGN output with scores and depths"

"However experience has learned me that playing the same 15.000 games on a faster computer or at a somewhat slower time control the first 3 places on BoCC can easily swap places, likewise for the 5 first places on EAS since its scores are so close.
Therefore what really matters (to me, that is) are the output PGN's for each engine, included in the download and enjoy the games. These are the kind of games I fell in love with chess as a teenager, the Morphy style, attack......!"

This is a little confusing to me, because my EAS-tool sorts out all sac-games and short-wins and put them in different small pgn-files (one file for each sac-category and one for the short wins), too. Additionally, I offer my IWS-Tool, which does this job without any math (much faster than the EAS-Tool) an builds just one output-file with great games, sorted by the sacs (highest sac on top of the list). And the IWS-Tool comes with 2 versions. One of them keeps the game-comments of the gamebase-file (so, if there are evals and depths of any engine stored in the comments, they still can be found in the games in the pgn-output-file).
https://www.sp-cc.de/files/interesting_ ... ch_tool.7z

I considered to filter sac games as well but noticed the vast majority is already covered by the king-attack and shortie evaluation and there is already a considerable overlap, that a game is rewarded twice. Adding sac games evaluation would give a triple overlap and is over the top.

2) I strongly recommend to add some percent-numbers to the output of your tool. Using 4-digit percent-numbers (12.34%) instead of huge score-numbers (with different length), makes the output way better to read for humans.

3) I very strongly recommend, to write an explanation text-file, where to explain, how the different parts of the BoCC-Tool work. Right now, the whole tool, especially the King-attack-algorithm, is a black-box. Just I did in the ReadMe of my EAS-Tool, where I explain exactly, what the tool does for each category (sacs, shorts, bad draws) and how the score is calculated out of the found games.
As a user, I want to know, how an EAS- or BoCC- score is build, otherwise I can not fully judge the results. Especially for engine-developers (Patricia author told me, how important it was for him, to fully understand, what the EAS-tool does exactly) this is very important.

Note, the 2 work-in-progress icons on the website, but some info is already available on the beta version : https://prodeo.actieforum.com/t1735-eat ... tool#18203

4) I complete miss any sac-search in your tool? I find this quite strange for 2 reasons
a) sacs are the most important thing in exciting enginegames
b) because you have an internal pgn-parser (I dont!), it would be very easy for you, not only to find the sacs very fast, but also mark them in the pgn-comments, just directly behind the sac move. This would be very helpful for the users.

See above.

5) My IWS-Tool adds this to each found interesting game (this is very helpful for the user, I got several very positive feedbacks, so I recommend, you doing something similar):
Each game gets a new Annotator-Tag, so it is clear, which category the game belongs to. One of these 8 tags is added to each game-notation:
[Annotator "IWS-Tool: Queen Sacrifice found in this game"]
[Annotator "IWS-Tool: 5+ PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 4 PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 3 PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 2 PawnUnits Sacrifice found in this game"]
[Annotator "IWS-Tool: 1 PawnUnit Sacrifice found in this game"]
[Annotator "IWS-Tool: Game ended before endgame (material)"]
[Annotator "IWS-Tool: Material imbalance found in this game"]

I will have a look at this tool, thanks.

A game in the output-file looks like this (from my testrun of latest Obsidian 250706 dev), when I use the IWS-tool version, called:
Interesting_Wins_Search_Tool_V4.1_keep_comments.bat (mention the Annotator-Tag and the evals/depth-comments, still in the game)

[Event "UHO Ratinglist"]
[Site "?"]
[Date "2025.07.07"]
[Round "30"]
[White "Obsidian 250706 a512"]
[Black "RubiChess 250606 a512"]
[Result "1-0"]
[ECO "A45"]
[Opening "Queen's pawn game"]
[TimeControl "180+1"]
[Annotator "IWS-Tool: 5+ PawnUnits Sacrifice found in this game"]
[Termination "adjudication"]
[PlyCount "99"]
[GameDuration "00"]
[GameEndTime "2025-07-07T11.124 Mitteleuropäische Sommerzeit"]
[GameStartTime "2025-07-07T11.688 Mitteleuropäische Sommerzeit"]

1. d4 { book } 1... Nf6 { book } 2. Bf4 { book } 2... g6 { book } 3. Nc3 {
book } 3... Bg7 { book } 4. e4 { book } 4... d6 { book } 5. Qd2 { book }
5... Nc6 { book } 6. O-O-O { book } 6... O-O { book } 7. f3 { +1.12/22 13s
} 7... a6 { -1.22/24 34s } 8. Nge2 { +1.31/20 4.1s } 8... h5 { -0.96/22
4.7s } 9. h4 { +1.30/22 8.2s } 9... e6 { -1.38/23 23s } 10. Bg5 { +1.46/22
4.9s } 10... b5 { -1.27/21 2.7s } 11. g4 { +1.74/22 4.2s } 11... b4 {
-1.06/20 2.8s } 12. Na4 { +1.81/23 7.5s } 12... hxg4 { -1.52/20 4.3s } 13.
h5 { +1.94/20 2.8s } 13... Qe8 { -2.01/23 15s } 14. e5 { +2.41/20 2.9s }
14... dxe5 { -1.72/21 2.4s } 15. Bg2 { +2.77/20 3.4s } 15... exd4 {
-2.39/22 15s } 16. h6 { +4.26/20 2.9s } 16... Bh8 { -3.51/20 4.9s } 17. h7+
{ +4.35/20 4.5s } 17... Nxh7 { -3.85/18 1.3s } 18. Rxh7 { +4.66/19 3.2s }
18... Kxh7 { -4.56/17 1.8s } 19. Rh1+ { +4.64/22 4.9s } 19... Kg7 {
-4.39/19 2.8s } 20. fxg4 { +4.77/22 5.0s } 20... e5 { -4.57/20 3.7s } 21.
Bd5 { +5.01/20 3.6s } 21... Be6 { -4.91/21 7.7s } 22. Be7 { +5.05/19 3.8s }
22... f5 { -4.91/20 2.0s } 23. Bxf8+ { +5.08/21 5.3s } 23... Qxf8 {
-5.50/19 2.9s } 24. Bxc6 { +5.20/20 3.8s } 24... Rd8 { -5.35/22 8.5s } 25.
gxf5 { +5.20/20 3.9s } 25... Bxf5 { -5.36/18 1.7s } 26. Ng3 { +5.24/19 3.3s
} 26... d3 { -5.65/21 8.1s } 27. Nxf5+ { +5.22/19 2.9s } 27... Qxf5 {
-5.74/19 2.2s } 28. Nc5 { +5.25/19 2.4s } 28... Kf7 { -5.72/20 5.3s } 29.
cxd3 { +5.18/22 7.8s } 29... Qf4 { -5.50/20 1.9s } 30. Ba4 { +5.27/24 9.4s
} 30... g5 { -5.46/22 7.0s } 31. Rh6 { +5.42/21 3.0s } 31... Ke7 { -5.49/22
13s } 32. Bb3 { +5.37/21 3.2s } 32... Qxd2+ { -5.28/19 3.7s } 33. Kxd2 {
+5.44/19 2.1s } 33... e4 { -5.26/20 3.1s } 34. Re6+ { +5.51/17 3.6s } 34...
Kf8 { -5.31/20 2.7s } 35. Nxe4 { +5.50/22 7.1s } 35... a5 { -5.35/21 5.4s }
36. Nxg5 { +5.63/18 2.3s } 36... Bxb2 { -5.37/19 5.4s } 37. Kc2 { +5.88/21
6.9s } 37... Bc3 { -5.21/20 3.4s } 38. Rc6 { +5.83/21 3.3s } 38... Re8 {
-5.22/19 2.1s } 39. Ne4 { +5.90/21 2.3s } 39... Bd4 { -5.07/19 1.4s } 40.
Rxc7 { +6.62/21 1.8s } 40... a4 { -5.34/19 1.5s } 41. Bxa4 { +6.62/19 2.6s
} 41... b3+ { -5.85/19 2.9s } 42. Bxb3 { +7.00/18 1.6s } 42... Re7 {
-6.21/25 2.1s } 43. Rxe7 { +7.41/22 1.6s } 43... Kxe7 { -6.08/25 1.8s } 44.
a4 { +7.51/27 1.8s } 44... Kd7 { -6.90/24 3.8s } 45. a5 { +7.95/26 1.7s }
45... Kc6 { -6.66/20 1.2s } 46. Bc4 { +8.19/22 1.6s } 46... Bg1 { -7.12/22
1.7s } 47. Nc3 { +8.50/23 1.7s } 47... Ba7 { -7.13/19 1.00s } 48. a6 {
+9.13/24 4.1s } 48... Kb6 { -7.16/22 1.0s } 49. Nb5 { +8.05/29 3.1s } 49...
Kxa6 { -4.36/20 1.0s } 50. Nxa7+ { +7.90/27 2.5s, White wins by
adjudication: SyzygyTB } 1-0

[pohl4711]

I considered to filter sac games as well but noticed the vast majority is already covered by the king-attack and shortie evaluation and there is already a considerable overlap, that a game is rewarded twice. Adding sac games evaluation would give a triple overlap and is over the top.

IMO, a twice (or triple) rewarded game is nothing bad: A king-attack must not be a short game, too. But if it is a short game, too, it should get points for king-attack and for being a shortie.
That does not mean, to put the game two or three times in the gamefile-output. In my IWS-Tool, I do this:

The IWS-Tool knows 8 categories:
1) Queen Sacrifices, followed by
2) 5+ PawnUnit Sacrifices, followed by
3) 4 PawnUnit Sacrifices, followed by
4) 3 PawnUnit Sacrifices, followed by
5) 2 PawnUnit Sacrifices, followed by
6) 1 PawnUnit Sacrifices, followed by
7) Games, ended before endgame (material) was reached, followed by
8) Games with material imbalance (Rook vs. Bishop and 2 pawns for example)

The games in the output-files are sorted in 2 ways:
First: The games are sorted by categories (category 1 is followed by category 2, 3, ... etc.).
Second: In each category, the games are sorted by length (0-19 moves, followed by 20-29 moves, followed by
30-39 moves... and so on, up to 120 moves and beyond). So, in each category, the shortest wins are at the
beginning and followed by the longer wins...
And, there are no double games in one output-file: If a game fits more than one category, it is stored
in the lowest category, all other apperances of this game in higher categories are deleted.
For example: A game contains a 3 PawnUnit-Sacrifice and is won before the endgame material is reached:
This game is stored in category 4 (= 3 PawnUnit Sacrifices) and not in category 7...

[pohl4711]

I considered to filter sac games as well but noticed the vast majority is already covered by the king-attack and shortie evaluation

But you could do a sac-search (as I already mentioned) in games not only for finding sac-games, but for finding the sac-move in the game-notation and make a note in the pgn-comments at this move. This would be a huge benefit over my EAS-Tool and could allow building engine-test-sets, by finding sac-move positions in games, automatically.
So, please reconsider your decision, not doing a sac-search in your tool.

[Rebel]

Made a beta available, link below.

1. Fixed a mean bug which changed the landscape considerable.
2. Added percentages for King-Attack and Shorties on advice from Stefan, looks better indeed.

The complete list then of the 135.000 games, the opponent elo rating is 3578 CEGT elo.

Code: Select all

BoCC Rating-list  Tue Jul 15 10:52:33 2025

PGN   : pgn\bocc-3578.pgn
Games : 135000
Won   : 77849
Tine  : 168 seconds
ELO   : 3578

  BoCC   King          Short     
 Total  Attack   %     Games   %    Engine
159634  106519 23.8%   53115 46.1%  CSTal 2.1 EAS
119848   85742 21.4%   34106 34.5%  Rebel Extreme Dev
102990   71930 20.4%   31060 33.0%  Patricia 3.01
102637   63974 22.4%   38663 36.4%  Rebel Extreme
 94256   79686 26.1%   14570 18.9%  Patricia 4 dev
 92146   78415 26.4%   13731 17.1%  Patricia 4
 73558   50220 19.5%   23338 32.1%  Velvet 8.1 risky
 71720   41183 19.2%   30537 32.2%  Rebel EAS 2.0
 63205   42955 16.5%   20250 27.7%  SF 17.1
 36694   19013 06.7%   17681 28.4%  Titan
 33152   18461 07.2%   14691 23.6%  Obsidian130
 33058   17954 07.1%   15104 23.9%  Clover.8.0.2
 31011   16329 06.0%   14682 24.2%  viridithas 14.0.1
 30780   17130 07.1%   13650 23.2%  berserk 13
 30650   17026 06.4%   13624 22.3%  PlentyChess 2.1.0
 30052   15448 05.5%   14604 24.0%  seer_v2.8
 29776   16467 05.8%   13309 22.7%  caissa 1.20
 29504   15958 06.3%   13546 22.2%  Alexandria 7.0
 28990   17421 06.7%   11569 20.4%  Lizard 11_0
 28987   14984 05.3%   14003 19.8%  Titan 1.1

Now that the bug is fixed BoCC and EAS differ considerable.

Download and double click BoCC-BETA.exe

135.000 games PGN included.

Next and running an 3500 elo pool.

Not sure what to expect, unexplored horizons.

[pohl4711]

When I put my ratinglist gamebase (of my UHO-Top15 Ratinglist) in the tool (120000 games), I got this nonsense output:

BoCC Rating-list Wed Jul 16 06:03:25 2025

PGN : pgn\uho_ratinglist_games.pgn
Games : 120000
Won : 60322
Tine : 225 seconds

BoCC King Short
Total Attack % Games % Engine
1 1 ˆþ 0 è Alexandria
1 1 bmi2 0 bmi2 Reckless
1 1 a512 0 a512 Viridithas
1 1 a512 0 a512 Torch
1 1 a512 0 a512 Stormphrax
1 1 a512 0 a512 Stockfish
1 1 a512 0 a512 Stockfish
1 1 a512 0 a512 RubiChess
1 1 a512 0 a512 PlentyChess
1 1 a512 0 a512 Obsidian
1 1 a512 0 a512 KomodoDragon
1 1 a512 0 a512 Integral
1 1 a512 0 a512 Horsie
1 1 a512 0 a512 Ethereal
1 1 a512 0 a512 Caissa
1 1 a512 0 a512 Berserk

[Rebel]

I noticed myself and fixed it, just copy the new executable over the old.