Great surprise: balanced openings can still be won

[Kanizsa]

Can the strongest engines still win by playing each other from super balanced opening lines?

In the main engine competitions, like CCC and TCEC, we are getting used to the idea that in order to strike blood on the chessboard, the initial imbalance should be at least of the order of a pawn or so

(* 0,70 according to Larry Kaufman, at least 1.00 according to Jeroen Noomen).

To our surprise, this incredible victory of Leela against SF comes from TCEC.

Just a 3-move opening line,

1. e4 e5 2. Nf3 Nc6 3. Bb5 Bc5

with both engines evaluating it only +0,26;

[Event "TCEC Season 24 - Kibitzer Bonus"]
[Site "https://tcec-chess.com"]
[Date "2023.05.04"]
[Round "12.1"]
[White "LCZero dag-master-pr1821-1842e13-T1-3905000"]
[Black "Stockfish master-72d542f"]
[Result "1-0"]
[Annotator "archive"]

[pgn]
1. e4 e5 2. Nf3 Nc6 3. Bb5 Bc5 4. c3 Nf6 5. O-O O-O 6. d4 Bb6 7. a4 a6 8. Bxc6 dxc6 9. Nxe5 Nxe4 10. Re1 Nd6 11. a5 Ba7 12. Be3 Be6 13. Nf3 f6 14. b4 Nb5 15. Qc2 Qd7 16. Nbd2 Rfe8 17. Rad1 Rad8 18. c4 Bf5 19. Qb3 Be6 20. Kh1 Bf7 21. Qb2 Nd6 22. Rc1 Bh5 23. h3 h6 24. Qc2 Qf7 25. Kh2 g5 26. c5 Bg6 27. Qb2 Nf5 28. Nf1 Ne7 29. Bd2 Nd5 30. Ng3 Qd7 31. Ng1 h5 32. f3 h4 33. Ne4 Qf5 34. Ne2 Bh5 35. Rf1 Bf7 36. N2c3 Be6 37. Rce1 Qg6 38. Nd1 Bf5 39. Ne3 Bxe4 40. fxe4 Rxe4 41. Qb1 Kh7 42. Nf5 Ree8 43. Re4 Rxe4 44. Qxe4 Re8 45. Qg4 Rg8 46. Rf3 Rh8 47. Be1 Kg8 48. Nxh4 Qf7 49. Nf5 Qh5 50. Qe4 Kf8 51. Bd2 Qe8 52. Qd3 Qg6 53. Rf1 Rh7 54. g4 Bb8 55. Qe2 Qe8 56. Qf3 Qg6 57. h4 Ke8 58. h5 b6 59. Ng3 Kd7 60. Ne4 Rh6 61. Kg1 Qg8 62. cxb6 cxb6 63. axb6 Bf4 64. Nc5+ Kd8 65. b7 Kc7 66. Qa3 Bxd2 67. Nxa6+ Kd6 68. b8=Q+ Qxb8 69. Nxb8 Rh7 70. b5+ Ke6 71. Nxc6 Rd7 72. Qf8 Nf4 73. Rd1 Be3+ 74. Kf1 Nd5 75. Ra1 Nb6 76. Ke2 Bxd4 77. Rd1 Bc5 78. Qg8+ Rf7 79. Nd8+ Ke5 80. Kf3 Bd4 81. Qxf7 Kd6 82. Nb7+ Ke5 83. Qe8+ Kd5 84. Qc6+ Ke5 85. Qd6#
*[/pgn]

[irawan]

this fact is amazing.
maybe there is some bug introduced by SF in case of this very narrow opening line.

[abgursu]

I don't think it's some sort of bug. It seems like SF overused its time early and suffered later.

[Jouni]

Lc0 is running with 4x RTX4090. Good budget solution .

[Jouni]

Addition: SF plays with 5995WX! This CPU still hasn't bench in ipman page.

[Kanizsa]

Leela has about a double of speed with 4x4090 instead of previous 2xA100.

And what about Stockfish ?

5995x is better or worse than before ?

[Uri Blass]

Can the strongest engines still win by playing each other from super balanced opening lines?

In the main engine competitions, like CCC and TCEC, we are getting used to the idea that in order to strike blood on the chessboard, the initial imbalance should be at least of the order of a pawn or so

(* 0,70 according to Larry Kaufman, at least 1.00 according to Jeroen Noomen).

To our surprise, this incredible victory of Leela against SF comes from TCEC.

Just a 3-move opening line,

1. e4 e5 2. Nf3 Nc6 3. Bb5 Bc5

with both engines evaluating it only +0,26;

[Event "TCEC Season 24 - Kibitzer Bonus"]
[Site "https://tcec-chess.com"]
[Date "2023.05.04"]
[Round "12.1"]
[White "LCZero dag-master-pr1821-1842e13-T1-3905000"]
[Black "Stockfish master-72d542f"]
[Result "1-0"]
[Annotator "archive"]

[pgn]
1. e4 e5 2. Nf3 Nc6 3. Bb5 Bc5 4. c3 Nf6 5. O-O O-O 6. d4 Bb6 7. a4 a6 8. Bxc6 dxc6 9. Nxe5 Nxe4 10. Re1 Nd6 11. a5 Ba7 12. Be3 Be6 13. Nf3 f6 14. b4 Nb5 15. Qc2 Qd7 16. Nbd2 Rfe8 17. Rad1 Rad8 18. c4 Bf5 19. Qb3 Be6 20. Kh1 Bf7 21. Qb2 Nd6 22. Rc1 Bh5 23. h3 h6 24. Qc2 Qf7 25. Kh2 g5 26. c5 Bg6 27. Qb2 Nf5 28. Nf1 Ne7 29. Bd2 Nd5 30. Ng3 Qd7 31. Ng1 h5 32. f3 h4 33. Ne4 Qf5 34. Ne2 Bh5 35. Rf1 Bf7 36. N2c3 Be6 37. Rce1 Qg6 38. Nd1 Bf5 39. Ne3 Bxe4 40. fxe4 Rxe4 41. Qb1 Kh7 42. Nf5 Ree8 43. Re4 Rxe4 44. Qxe4 Re8 45. Qg4 Rg8 46. Rf3 Rh8 47. Be1 Kg8 48. Nxh4 Qf7 49. Nf5 Qh5 50. Qe4 Kf8 51. Bd2 Qe8 52. Qd3 Qg6 53. Rf1 Rh7 54. g4 Bb8 55. Qe2 Qe8 56. Qf3 Qg6 57. h4 Ke8 58. h5 b6 59. Ng3 Kd7 60. Ne4 Rh6 61. Kg1 Qg8 62. cxb6 cxb6 63. axb6 Bf4 64. Nc5+ Kd8 65. b7 Kc7 66. Qa3 Bxd2 67. Nxa6+ Kd6 68. b8=Q+ Qxb8 69. Nxb8 Rh7 70. b5+ Ke6 71. Nxc6 Rd7 72. Qf8 Nf4 73. Rd1 Be3+ 74. Kf1 Nd5 75. Ra1 Nb6 76. Ke2 Bxd4 77. Rd1 Bc5 78. Qg8+ Rf7 79. Nd8+ Ke5 80. Kf3 Bd4 81. Qxf7 Kd6 82. Nb7+ Ke5 83. Qe8+ Kd5 84. Qc6+ Ke5 85. Qd6#
*[/pgn]

1)Not sure if this is a balanced line.
3...Bc5 may be a mistake
3...a6 or 3...Nf6 are more popular.
2)The stockfish team does not test patches from positions with +0.26 exit so maybe stockfish is not optimized to play well in them.

[Ozymandias]

1)Not sure if this is a balanced line.
3...Bc5 may be a mistake
3...a6 or 3...Nf6 are more popular.

It's not. The score isn't everything.
It is.
Far more popular, even the Smyslov, Cozio and Lucena Defences are better:



This thread is for the birds.

[Kanizsa]

"Cordel variation" in Spanish was Bobby Fischer's favorite.
How can it be bad ?? I remember: +0,26 it is Leela & Stockfish evaluation at book exit after 3....Bc5

P.S. My Elo Fide is almost 2400 points.

[Ozymandias]

It's like the oldest way to meet Bb5. Long ago superseded.

Fisher played many things to avoid theory, he also played f4, Kf2, Kf3. doesn't mean anything.

If SF loses with it, it's bad enough. Elo 4000 or thereabouts.