mehmet1921's books tests.

denochss · Post by **denochss** » 17/08/2021, 21:02

Reviewing the games I have noticed that Carlsen's used book does not play exactly as Carlsen would play.

For example Game 99

Here Carlsen would play
9 ... Be7

Move 9 ... b5 is from a game played in Black by Kotronias, Carlsen plays in White

mehmet1921 · Post by **mehmet1921** » 18/08/2021, 3:52

I think Carlsen.bin played 9... f5 move because only 1 game remained in its game data (Carlsen-Kotronias) after 9.f4 move.
9.f4 is a rare move in this opening

denochss · Post by **denochss** » 18/08/2021, 10:01

mehmet1921 wrote: ↑18/08/2021, 3:52 I think Carlsen.bin played 9... f5 move because only 1 game remained in its game data (Carlsen-Kotronias) after 9.f4 move.
9.f4 is a rare move in this opening

Hello mehmet1921, the Carlsen book that you use should not play other players' plays, that is, the Carlsen book that you use does not play as Carlsen would play, as it makes plays of its rivals.

Here is the game that Carlsen wins playing in black 9 ... Be7

Peter Grayson · Post by **Peter Grayson** » 18/08/2021, 12:14

denochss wrote: ↑17/08/2021, 21:02

Here Carlsen would play
9 ... Be7

Move 9 ... b5 is from a game played in Black by Kotronias, Carlsen plays in White

That is where the Chessbase .ctg books have a big advantage when the moves for either side can be prioritised green and only those moves will be selected making it easy to produce a themed book based on the required criteria. Create separate White and Black game databases to reflect the particular player games and then just read them in as prioritised moves for the appropriate colour.

The moves that remain black are moves not played by the player and will not be selected by the GUI when Tournament = checked. For example from 1265 of his games against 2300+ players in my database ...

https://www.mediafire.com/file/uty51mxakmjivxf/Carlsen_Move_Marker.zip/file

If requred create a supporting .ctg book first and then use the import book feature to import the prioritised Carlsen games book and all of Carlsen's moves will be marked green. With Tournament=checked only those moves wil be selected until the line runs out and only then will the supporting book moves be used.

Peter

mehmet1921 · Post by **mehmet1921** » 21/08/2021, 6:26

NimaTiv-3.ctg vs DONBFupdate1.ctg:

Program Elo + - Games Score Av.Op. Draws

1 Stockfish 200821 x64 bmi2 D : 2404 10 5 206 51.2 % 2396 97.6 %
2 Stockfish 200821 x64 bmi2 N : 2396 5 10 206 48.8 % 2404 97.6 %

Individual statistics:

1 Stockfish 200821 x64 bmi2 D: 2404 206 (+ 5,=201,- 0), 51.2 %

Stockfish 200821 x64 bmi2 N : 206 (+ 5,=201,- 0), 51.2 %

2 Stockfish 200821 x64 bmi2 N: 2396 206 (+ 0,=201,- 5), 48.8 %

Stockfish 200821 x64 bmi2 D : 206 (+ 0,=201,- 5), 48.8 %

Game Conditions: Fritz 17 Gui, 4 Cores (Core-i7 9750h), 3 min TC, 1024 Mb Hash, Ponder Off
Book Settings: Both books played with author's recommended settings.
NimaTiv-3 Settings:(Up to move:100, Variety Of Play:10, Influence Of Learn Value:100, Learning Strenght:0),
DONBFupdate1 Setings(Up to move:80, Variety Of Play:50, Influence Of Learn Value:100, Learning Strenght:500)
Stockfish 200821 x64 bmi2 N (NimaTiv-3:Ian Hurnavich/Stockfish 200821 x64 bmi2 N ( DONBFupdate1:Vasid Chouhan)
https://www.mediafire.com/file/ivh66p3m7h3dttq/r102.pgn/file

I did not include 4 games that ended with time loss in the calculation. DONBFupdate1:3 wins, NimaTiv-3:1 win , all this 4 games are dead draw. Unlike Fritz Gui, I haven't seen time loss match in Cutechess Gui for a long time with Stockfish.

There were only 2 opening variants (A28 and C65) in all games played. I will test NimaTiv-3 against DONBFupdate1 with new book settings as it would not be right to decide on a new book based on a match with only 2 opening variants.

Post by **Hurnavich** » 21/08/2021, 8:02

mehmet1921 wrote: ↑21/08/2021, 6:26 NimaTiv-3.ctg vs DONBFupdate1.ctg:

Program Elo + - Games Score Av.Op. Draws

1 Stockfish 200821 x64 bmi2 D : 2404 10 5 206 51.2 % 2396 97.6 %
2 Stockfish 200821 x64 bmi2 N : 2396 5 10 206 48.8 % 2404 97.6 %

Individual statistics:

1 Stockfish 200821 x64 bmi2 D: 2404 206 (+ 5,=201,- 0), 51.2 %

Stockfish 200821 x64 bmi2 N : 206 (+ 5,=201,- 0), 51.2 %

2 Stockfish 200821 x64 bmi2 N: 2396 206 (+ 0,=201,- 5), 48.8 %

Stockfish 200821 x64 bmi2 D : 206 (+ 0,=201,- 5), 48.8 %

Game Conditions: Fritz 17 Gui, 4 Cores (Core-i7 9750h), 3 min TC, 1024 Mb Hash, Ponder Off
Book Settings: Both books played with author's recommended settings.
NimaTiv-3 Settings:(Up to move:100, Variety Of Play:10, Influence Of Learn Value:100, Learning Strenght:0),
DONBFupdate1 Setings(Up to move:80, Variety Of Play:50, Influence Of Learn Value:100, Learning Strenght:500)
Stockfish 200821 x64 bmi2 N (NimaTiv-3:Ian Hurnavich/Stockfish 200821 x64 bmi2 N ( DONBFupdate1:Vasid Chouhan)
https://www.mediafire.com/file/ivh66p3m7h3dttq/r102.pgn/file

I did not include 4 games that ended with time loss in the calculation. DONBFupdate1:3 wins, NimaTiv-3:1 win , all this 4 games are dead draw. Unlike Fritz Gui, I haven't seen time loss match in Cutechess Gui for a long time with Stockfish.

There were only 2 opening variants (A28 and C65) in all games played. I will test NimaTiv-3 against DONBFupdate1 with new book settings as it would not be right to decide on a new book based on a match with only 2 opening variants.

hi, book on book match should of been 140 variety 10 is for playchess

mehmet1921 · Post by **mehmet1921** » 21/08/2021, 8:17

Hurnavich wrote: ↑21/08/2021, 8:02 hi, book on book match should of been 140 variety 10 is for playchess

Ok. 1 hour ago I started a new match against DONBFupdate with variety of play :140 (Multi Book Match Setting).

Post by **Hurnavich** » 21/08/2021, 8:26

mehmet1921 wrote: ↑21/08/2021, 8:17
Hurnavich wrote: ↑21/08/2021, 8:02 hi, book on book match should of been 140 variety 10 is for playchess
Ok. 1 hour ago I started a new match against DONBFupdate with variety of play :140 (Multi Book Match Setting).

Im releasing a fix for the line that lost 3 games and the other losses thanks for the match it has helped a quick fix.

updated book in 10 min release

mehmet1921 · Post by **mehmet1921** » 21/08/2021, 9:20

I stopped the match after 50 games. The score was 26-24 (+2-0 =48) in favor of DONBFupdate1 and the opening variant of 2 wins was also A28.
Under these circumstances and with NimaTiv-3.1 released, there is no point in continuing this match.
Because of this I'm starting a new match NimaTiv-3.1 vs DONBFupdate1 with recommended settings.

mehmet1921 · Post by **mehmet1921** » 21/08/2021, 19:11

NimaTiv-3.1 .ctg vs DONBFupdate1.ctg:

Program Elo + - Games Score Av.Op. Draws

1 Stockfish 200821 x64 bmi2 D : 2403 10 5 200 51.0 % 2397 98.0 %
2 Stockfish 200821 x64 bmi2 N : 2397 5 10 200 49.0 % 2403 98.0 %

Individual statistics:

1 Stockfish 200821 x64 bmi2 D: 2403 200 (+ 4,=196,- 0), 51.0 %

Stockfish 200821 x64 bmi2 N : 200 (+ 4,=196,- 0), 51.0 %

2 Stockfish 200821 x64 bmi2 N: 2397 200 (+ 0,=196,- 4), 49.0 %

Stockfish 200821 x64 bmi2 D : 200 (+ 0,=196,- 4), 49.0 %

Game Conditions: Fritz 17 Gui, 4 Cores (Core-i7 9750h), 3 min TC, 1024 Mb Hash, Ponder Off
Book Settings: Both books played with author's recommended settings.
NimaTiv-3.1 Settings:(Up to move:100, Variety Of Play:140, Influence Of Learn Value:100, Learning Strenght:0),
DONBFupdate1 Settings(Up to move:80, Variety Of Play:50, Influence Of Learn Value:100, Learning Strenght:500)
Stockfish 200821 x64 bmi2 N (NimaTiv-3.1:Ian Hurnavich/Stockfish 200821 x64 bmi2 D( DONBFupdate1:Vasid Chouhan)
https://www.mediafire.com/file/lmm9337seuvh8id/r103.pgn/file

I did not include 3 games that ended with time loss in the calculation. DONBFupdate1:2 wins ( In one of this game NimaTiv-3.1 was very close to win, the other game was dead draw.), NimaTiv-3.1 :1 win (Dead draw)
The opening variants of 4 wins of DONBFupdate1 was A28 like the other games DONBFupdate1 won.

DON · Post by **DON** » 22/08/2021, 7:41

Time loss can be avoid by increasing Move Overhead to 100

always thanks for your test Mehmet

mehmet1921 · Post by **mehmet1921** » 22/08/2021, 8:15

DON wrote: ↑22/08/2021, 7:41 Time loss can be avoid by increasing Move Overhead to 100

In Engine - Engine tests, I always prefer to use engines in their default settings. A good engine should not lose game because of time loss. But in opening book test, the situation is a little different, since the chess engine loses the game due to time loss, the opening book also loses the game.
From now on I'm thinking of making some changes in the chess engine settings for Book-Book matches in Fritz Gui if needed

mehmet1921 · Post by **mehmet1921** » 22/08/2021, 9:13

SF-SE-1c.bin vs DONBFupdate1.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 Don : 2401 7 7 250 50.2 % 2399 97.2 %
2 Cfish 130721 Sfs : 2399 7 7 250 49.8 % 2401 97.2 %

Individual statistics:

1 Cfish 130721 Don : 2401 250 (+ 4,=243,- 3), 50.2 %

Cfish 130721 Sfs : 250 (+ 4,=243,- 3), 50.2 %

2 Cfish 130721 Sfs : 2399 250 (+ 3,=243,- 4), 49.8 %

Cfish 130721 Don : 250 (+ 3,=243,- 4), 49.8 %

Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 D ( DONBFupdate1:Vasid Chouhan), Cfish 130721 S (SF-SE-1c:"Skynet")
https://www.mediafire.com/file/ibb5txpk3n3otx5/r104.pgn/file

mehmet1921 · Post by **mehmet1921** » 24/08/2021, 18:16

LittleCat-Xtreme 0.3 .bin vs M12.7 .bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 M12 : 2416 17 13 200 54.5 % 2384 90.0 %
2 Cfish 130721 Lcx : 2384 13 17 200 45.5 % 2416 90.0 %

Individual statistics:

1 Cfish 130721 M12 : 2416 200 (+ 19,=180,- 1), 54.5 %

Cfish 130721 Lcx : 200 (+ 19,=180,- 1), 54.5 %

2 Cfish 130721 Lcx : 2384 200 (+ 1,=180,- 19), 45.5 %

Cfish 130721 M12 : 200 (+ 1,=180,- 19), 45.5 %

Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 Lcx ( LittleCat-Xtreme 0.3: "Amanda97"), Cfish 130721 M12 (M12.7:Mustafa Canseven)
https://www.mediafire.com/file/frq0ca07vtl9n7q/r107.pgn/file

The performance of LittleCat-Xtreme 0.3 against M12.7 isn't good. LC-X 8.7H is another book of "Amanda97" and it's perfomance was much better against M12.7 (55- 55 (+ 0,=110,- 0))
http://outskirts.altervista.org/forum/viewtopic.php?f=29&t=2477

mehmet1921 · Post by **mehmet1921** » 25/08/2021, 16:00

NimaTiv-3.3.ctg vs DONBFupdate1.ctg:

Program Elo + - Games Score Av.Op. Draws

1 SugaR AI 2.30 bmi2 D : 2400 5 5 360 50.0 % 2400 97.8 %
2 SugaR AI 2.30 bmi2 N : 2400 5 5 360 50.0 % 2400 97.8 %

Individual statistics:

1 SugaR AI 2.30 bmi2 D : 2400 360 (+ 4,=352,- 4), 50.0 %

SugaR AI 2.30 bmi2 N : 360 (+ 4,=352,- 4), 50.0 %

2 SugaR AI 2.30 bmi2 N : 2400 360 (+ 4,=352,- 4), 50.0 %

SugaR AI 2.30 bmi2 D : 360 (+ 4,=352,- 4), 50.0 %

Game Conditions: Fritz 17 Gui, 6 Cores (Core-i7 9750h), 2 min TC, 1024 Mb Hash, Ponder Off,Experience File Off
Book Settings: Both books played with author's recommended settings.
NimaTiv-3.3 Settings:(Up to move:100, Variety Of Play:140, Influence Of Learn Value:100, Learning Strenght:0),
DONBFupdate1 Settings(Up to move:80, Variety Of Play:50, Influence Of Learn Value:100, Learning Strenght:500)
SugaR AI 2.30 bmi2 N(NimaTiv-3.3:Ian Hurnavich/SugaR AI 2.30 bmi2 D ( DONBFupdate1:Vasid Chouhan)
https://www.mediafire.com/file/3s06wx740yj67c5/r108.pgn/file

Outskirts CheSS ForuM

Outskirts CheSS ForuM

mehmet1921's books tests.

Carlsen vs Hubble 1.3:

Carlsen vs Hubble 1.3:

Carlsen vs Hubble 1.3:

Carlsen vs Hubble 1.3:

NimaTiv-3 vs DONBFupdate1:

NimaTiv-3 vs DONBFupdate1:

NimaTiv-3 vs DONBFupdate1:

NimaTiv-3 vs DONBFupdate1:

NimaTiv-3 vs DONBFupdate1:

NimaTiv 3.1 vs DONBFupdate1

NimaTiv 3.1 vs DONBFupdate1

NimaTiv 3.1 vs DONBFupdate1

SF-SE-1c vs DONBFupdate1:

LittleCat-Xtreme 0.3 vs M12.7

NimaTiv-3.3 vs DONBFupdate1: