mehmet1921's books tests.

[mehmet1921]

Predator.bin vs M12.9 .bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 P : 2408 15 13 200 52.2 % 2392 91.5 %
2 Cfish 130721 M : 2392 13 15 200 47.8 % 2408 91.5 %

Individual statistics:

1 Cfish 130721 P : 2408 200 (+ 13,=183,- 4), 52.2 %

Cfish 130721 M : 200 (+ 13,=183,- 4), 52.2 %

2 Cfish 130721 M : 2392 200 (+ 4,=183,- 13), 47.8 %

Cfish 130721 P : 200 (+ 4,=183,- 13), 47.8 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 P (Predator: Eduard Nemeth), Cfish 130721 M (M12.9:Mustafa Canseven)
[External Link Removed for Guests]

This is the heaviest defeat of M12.x so far in my tests.

[mehmet1921]

Predator.bin vs MANTIS-XO.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 P : 2401 8 8 300 50.3 % 2399 96.0 %
2 Cfish 130721 M : 2399 8 8 300 49.7 % 2401 96.0 %

Individual statistics:

1 Cfish 130721 P : 2401 300 (+ 7,=288,- 5), 50.3 %

Cfish 130721 M : 300 (+ 7,=288,- 5), 50.3 %

2 Cfish 130721 M : 2399 300 (+ 5,=288,- 7), 49.7 %

Cfish 130721 P : 300 (+ 5,=288,- 7), 49.7 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 P (Predator: Eduard Nemeth), Cfish 130721 M ( MANTIS-XO: Roger Constantinesco)
[External Link Removed for Guests]


Predator managed to win its second match against its strong opponents.

[mehmet1921]

Predator.bin vs DONBFupdate1.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 P : 2402 7 7 320 50.5 % 2398 96.6 %
2 Cfish 130721 D : 2398 7 7 320 49.5 % 2402 96.6 %

Individual statistics:

1 Cfish 130721 P : 2402 320 (+ 7,=309,- 4), 50.5 %

Cfish 130721 D : 320 (+ 7,=309,- 4), 50.5 %

2 Cfish 130721 D : 2398 320 (+ 4,=309,- 7), 49.5 %

Cfish 130721 P : 320 (+ 4,=309,- 7), 49.5 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 P (Predator: Eduard Nemeth), Cfish 130721 D ( DONBFupdate1:Vasid Chouhan)
[External Link Removed for Guests]

Although Predator performed very well in the first 2 matches, I was guessing that Predator would be defeated against DONBFupdate1 in this match. But Predator managed to get through this match.


For other book -book matches:
[Local Link Removed for Guests]

[kramnik]

Welcome to the NNUE era...

Goi X bin vs Sf14-010921.bin:

Individual statistics:

1 Cfish 130721 G : 2401 300 (+ 2,=297,- 1), 50.2 %

Cfish 130721 S : 300 (+ 2,=297,- 1), 50.2 %

2 Cfish 130721 S : 2399 300 (+ 1,=297,- 2), 49.8 %

Cfish 130721 G : 300 (+ 1,=297,- 2), 49.8 %

Thanks for the test! It's a bit funny, that a book (sf), which has been made from Stefan's and Oliprog's engine vs engine tests (not book vs book), can compete with private - tuned, analyzed lines etc book.

[Eduard]

Thanks for all the matches. I will download all the games (also of the other matches) and update the predator book. As I mentioned, I play with it on Lichess with very good results. Maybe I can still make the book a little better. I will also expand the book with other good variants from online games.
By the way: The book is made for a maximum depth of 40 moves. But it also has few longer variants.

[kramnik]

What is so funny for you?....

hahaha goi vs sf 14 lool

[Japhet3719]

Next match hopefully

Pavel vs. Goi X

Predator vs. Goi X

That's a good match

[Eduard]

Predator.bin vs DONBFupdate1.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 P : 2402 7 7 320 50.5 % 2398 96.6 %
2 Cfish 130721 D : 2398 7 7 320 49.5 % 2402 96.6 %

Individual statistics:

1 Cfish 130721 P : 2402 320 (+ 7,=309,- 4), 50.5 %

Cfish 130721 D : 320 (+ 7,=309,- 4), 50.5 %

2 Cfish 130721 D : 2398 320 (+ 4,=309,- 7), 49.5 %

Cfish 130721 P : 320 (+ 4,=309,- 7), 49.5 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 P (Predator: Eduard Nemeth), Cfish 130721 D ( DONBFupdate1:Vasid Chouhan)
[External Link Removed for Guests]

Although Predator performed very well in the first 2 matches, I was guessing that Predator would be defeated against DONBFupdate1 in this match. But Predator managed to get through this match.


For other book -book matches:
[Local Link Removed for Guests]

I have improved some variants in the Predator book (but I'm still waiting for the update because I want to import more games). However, it is not always easy. The engines also make gross mistakes, especially in blitz. It's not always the books. The mistakes I saw were made by the engines, and most of all, gross mistakes.

Example:

[SetUp "1"] [FEN "1Q6/2p5/p1k3r1/1p2p1r1/1q2Pp2/1N1P1P1p/2P1R1pK/6R1 b - - 0 1"] { 1Q6/2p5/p1k3r1/1p2p1r1/1q2Pp2/1N1P1P1p/2P1R1pK/6R1 b - - 0 1 } *

Black played here 47...Qc3?

48.Rd2 Kd7 49.d4 exd4 50.Rxd4+ Rd6 51.Rxd6+ cxd6 52.Qa7+ Ke8 53.Nd4 Qc5 54.Qa8+ Kf7 55.Qd8 a5 56.Qd7+ Kg6 57.Qe8+ Kh7 58.Qf7+ Kh8 59.Qf8+ Kh7 60.Qxf4 Qe5 61.Qxe5 dxe5 62.Nxb5 Rh5 63.Nc3 Rh6 64.Ne2 a4 65.Ra1 Rc6 66.Rxa4
White has a decisive advantage: +- (3.46) Depth: 24/46 00:00:03 66037kN, tb=18709

Lose instantly.

The position is a draw, however, the books haven't done anything wrong:

Analysis by Hypnos legacy 020921:

47...a5 48.d4 a4 49.Qa8+ Kd7 50.Nc5+ Ke7 51.Qc8 Qxd4 52.Qxc7+ Ke8 53.Ree1 Qf2 54.Qd7+ Kf8 55.Ne6+ Rxe6 56.Qd8+ Kf7 57.Qxg5 Rh6 58.c4 bxc4 59.Qf5+ Kg8 60.Qc8+ Kg7 61.Qc7+ Kg6 62.Qd6+ Kh7 63.Qd7+ Kh8 64.Qe8+ Kh7 65.Qf7+ Kh8 66.Qf8+ Kh7 67.Qe7+
The position is equal: = (0.00) Depth: 54/40 00:00:05 101MN, tb=42553

and here:

[SetUp "1"] [FEN "4Rb1k/p2n3P/1p6/3P3r/r4B2/2P1P1P1/P3K3/3R4 b - - 0 1"] { 4Rb1k/p2n3P/1p6/3P3r/r4B2/2P1P1P1/P3K3/3R4 b - - 0 1 } *

Analysis by Hypnos legacy 020921:

33...Rxh7 34.e4 Kg7 35.Kf3 Rc4 36.e5 Rxc3+ 37.Kg4 Rc4 38.Re1 Rc5 39.d6 Kf7 40.Rd8 Ke6 41.Re2 Rg7+ 42.Kf3 Rc3+ 43.Kg2 Rh7 44.Re8+ Kf7 45.Ra8 Ke6 46.Rf2 Rc4 47.Re8+ Kf7 48.Rd8 Ke6 49.Ra8 Rc5 50.a4 Ra5 51.Re8+ Kf7 52.Rd8 Ke6 53.Re8+
The position is equal: = (0.00) Depth: 36/44 00:00:17 332MN, tb=198877

The position is a draw, but Black played here Rxa2?! and White got an advantage.

In the following position, White can easily make a draw:

[SetUp "1"] [FEN "1k1r3r/pppq4/2p2b2/4p2p/4Ppp1/2BP1P2/PPPNKQ2/R6R w - - 0 1"] { 1k1r3r/pppq4/2p2b2/4p2p/4Ppp1/2BP1P2/PPPNKQ2/R6R w - - 0 1 } *

Analysis by Hypnos legacy 020921:

21.fxg4 hxg4 22.Rag1 b6 23.Kd1 Qe6 24.b3 Kb7 25.Qg2 Rxh1 26.Rxh1 Rg8 27.Rh5 Bg5 28.Qh2 Qe7 29.Rh7 Qe6
The position is equal: = (0.00) Depth: 36/49 00:00:23 447MN, tb=178

If you want to play safely, then White moves 21. fxg4 and after that the position is pretty much a draw. However, practice with 64 cores has shown that White can take risks very well by playing Nc4!?. In practice, with strong hardware (from 32 cores), White always won. If you only have a few cores, you should better play fxg4 here.

I have both moves in the book. In the event of a defeat, the other move is played.

[kramnik]

Errors by engines may happen for various reasons:

- incorrect hashtable size for that time of play

- weak cpu

- too hot cpu (yes, it can occur)

- few threads used (in short times especially)

When testing books, always add the time increase per any move (i.e. 3+2) and set the hashtable size so that it's average use it's around 40% (you can watch it in the engine output window).

I have improved some variants in the Predator book (but I'm still waiting for the update because I want to import more games). However, it is not always easy. The engines also make gross mistakes, especially in blitz. It's not always the books. The mistakes I saw were made by the engines, and most of all, gross mistakes.

[mehmet1921]

NimaTiv 4.1.ctg vs Goi X.bin

Program Elo + - Games Score Av.Op. Draws

1 SugaR AI 2.30 bmi2 N : 2400 0 0 200 50.0 % 2400 100.0 %
2 SugaR AI 2.30 bmi2 G : 2400 0 0 200 50.0 % 2400 100.0 %

Individual statistics:

1 SugaR AI 2.30 bmi2 N : 2400 200 (+ 0,=200,- 0), 50.0 %

SugaR AI 2.30 bmi2 G : 200 (+ 0,=200,- 0), 50.0 %

2 SugaR AI 2.30 bmi2 G : 2400 200 (+ 0,=200,- 0), 50.0 %

SugaR AI 2.30 bmi2 N : 200 (+ 0,=200,- 0), 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.
SugaR AI 2.30 bmi2 N (NimaTiv-4.1:Ian Hurnavich/Cfish 130721 G (Goi X: Massimiliano Goi )
[External Link Removed for Guests]


According to my public and private tests top 3 most powerful opening books are Nimativ 4.1 ( Ian Hurnavich) , Goi X (Massimiliano Goi) and 1337chess Pro X (Michael Wagner). Nimativ is the only free opening book of these three opening books. It should also be noted that Vasid Chouhan (Don) and Angel Morana (Chucaro) haven't published their new opening books yet.

[mehmet1921]

MANTISX1.bin vs Cute.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 M : 2414 26 21 100 54.0 % 2386 88.0 %
2 Cfish 130721 C : 2386 21 26 100 46.0 % 2414 88.0 %

Individual statistics:

1 Cfish 130721 M : 2414 100 (+ 10,= 88,- 2), 54.0 %

Cfish 130721 C : 100 (+ 10,= 88,- 2), 54.0 %

2 Cfish 130721 C : 2386 100 (+ 2,= 88,- 10), 46.0 %

Cfish 130721 M : 100 (+ 2,= 88,- 10), 46.0 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 M ( MANTISX1: Roger Constantinesco), Cfish 130721 C (Cute:Cevdet Sarı)
[External Link Removed for Guests]


I had to keep the test short because MANTISX1 immediately gained a serious advantage over his opponent.

[mehmet1921]

Bronco.bin vs Sf14-010921.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 B : 2402 10 9 300 50.7 % 2398 94.0 %
2 Cfish 130721 S : 2398 9 10 300 49.3 % 2402 94.0 %

Individual statistics:

1 Cfish 130721 B : 2402 300 (+ 11,=282,- 7), 50.7 %

Cfish 130721 S : 300 (+ 11,=282,- 7), 50.7 %

2 Cfish 130721 S : 2398 300 (+ 7,=282,- 11), 49.3 %

Cfish 130721 B : 300 (+ 7,=282,- 11), 49.3 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 B (Bronco: "Anton101"), Cfish 130721 S (Sf14-010921: "Skynet")
[External Link Removed for Guests]

Bronco is an opening book published 3 months ago and it managed to beat a strong opening book (Sf14-010921). Don' t forget that Sf14-010921 beat Pavel-v2 and performed well against a commercial book (Goi X).

[mehmet1921]

Predator2 vs Goi X.bin:

Program Elo + - Games Score Av.Op. Draws

1 Cfish 130721 G : 2403 10 5 200 51.0 % 2397 98.0 %
2 Cfish 130721 P : 2397 5 10 200 49.0 % 2403 98.0 %

Individual statistics:

1 Cfish 130721 G : 2403 200 (+ 4,=196,- 0), 51.0 %

Cfish 130721 P : 200 (+ 4,=196,- 0), 51.0 %

2 Cfish 130721 P : 2397 200 (+ 0,=196,- 4), 49.0 %

Cfish 130721 G : 200 (+ 0,=196,- 4), 49.0 %


Game Conditions: Cutechess Gui, 1 Core (Core-i7 9750h), 3 min TC, 512 Mb Hash, Ponder Off, BestMove=Off / BookDepth=Max
Cfish 130721 G (Goi X: Massimiliano Goi ), Cfish 130721 P (Predator2: Eduard Nemeth)
[External Link Removed for Guests]

I made two matches. At first match I forgot to add the pgn file to Chess Gui.

The first match :
Score of Cfish 130721 G vs Cfish 130721 P: 3 - 2 - 295 [0.502]
... Cfish 130721 G playing White: 3 - 0 - 147 [0.510] 150
... Cfish 130721 G playing Black: 0 - 2 - 148 [0.493] 150
... White vs Black: 5 - 0 - 295 [0.508] 300
Elo difference: 1.2 +/- 5.1, LOS: 67.3 %, DrawRatio: 98.3 %

The real score at 500 games: Goi X: 252.5 - Predator2:247.5 ( +7-2 =491) % 50.5 +4 elo

It' a good result for Predator2 against a very strong opening book like Goi.

[Eduard]

Thanks, and Congratulations to GOI book.

But I have a problem: I don't play with such slow hardware (slow 1 core). I don't know how to tune a book when moves are played in depth 20 (Stockfish)? I could do that, but it doesn't make much sense for live games on servers. So that you can really improve a book in a meaningful way, it should be played with at least 4 cores.

If I see such positions that are a draw (I play with 6-pieces), how should I improve something? Black didn't do anything wrong. I can't improve that.

Cfish 130721 G - Cfish 130721 P 1-0, My Tournament 2021
[SetUp "1"] [FEN "5n2/5k2/5p2/5K2/6B1/8/8/4B3 b - - 0 1"] { 5n2/5k2/5p2/5K2/6B1/8/8/4B3 b - - 0 1 } *

1. = (0.00): 80...Kg7
2. = (0.00): 80...Nh7
3. +- (6): 80...Ke7
4. +- (6): 80...Nd7
5. +- (3): 80...Ne6
6. +- (2): 80...Ng6
7. +- (1): 80...Ke8
8. +- (1): 80...Kg8

There are only 200 games in the database. Where are the other 300?

[mehmet1921]

There are only 200 games in the database. Where are the other 300?

The answer is in the post "I made two matches. At first match I forgot to add the pgn file to Chess Gui".