TCEC Season 17 CPU League 1 Game 146 – Fire-RubiChess
|Game||Season 17, CPU League 1, Game 146|
|Links|| TCEC archive|
|Position after 10... Re8|
1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. f3 O-O 6. Bg5 c5 7. d5 h6 8. Be3 e6
Book exit into King's Indian Defense, Steiner Attack. This position has seen a healthy number of games at the top grandmaster level. The opening database indicates good chances for both sides with a small advantage for white.
9. Qd2 exd5 10. cxd5 Re8 (diagram)
This is an interesting point in the opening. The main book move in the Lichess masters database is 11. Nge2 (145 games), but this is not the move played in the game. Both Stockfish and Leela suggest 11. a4/Bd3/Bb5 (16/1/9 games) over the main book move with some margin for error. The point is likely to prefer more active development, and to liberate the light square bishop first, and to keep the knight's future flexible. And indeed, white developed the knight to h3 in the game. Perhaps this position could do with some reevaluation in the opening theory. (See also Editor's notes below.)
|Position after 21... f5|
11. a4 a6 12. a5
Note that 12. Bxh6?! here would not win a pawn, but would instead compromise white's position. For example: 12... Nxe4 13. Nxe4 Qh4+ 14. g3.
12... Nh7 13. Be2 h5 14. Ra3 Qh4+
Asks white to either weaken the kingside with g3, or misplace the bishop.
15. Bf2(N) 15... Qd8
Now, if white were to repeat with 16. Be3, black could go for a three-fold draw as the position evaluates slightly positive for white. To avoid this, white needs to make a deviation either now or later.
16. Bg3 Qe7
A question may be asked here: Why did black not play 15... Qe7 directly to win a tempo? The answer is that 15... Qd8 will slow down white's queenside push by a bit by attacking a5 and eyeing on b6, should white go with the Na4 and Rb3 plan. It is rather that 16. Bg3 allowed the black queen to enter the e-file to consider a future f-pawn push.
Compare these two lines:
- 15... Qe7 16. Rb3 Nd7 and white would be in time to play 16. Nh3 and then start pressuring the queenside. Now the black queen on e7 is slightly misplaced.
- 15... Qd8 16. Bg3. Now after 17. Rb3 black can gain a tempo 17... h4 18. Bf2 Nd7 with follow-up ideas such as Ne5 and h3. A bit surprisingly, by moving the queen to Qd8, black actually gains a tempo.
White plays Na4 anyways and black will later gain the tempo (move 20.)
17. Na4 Nd7 18. Nh3 Rb8 19. O-O g5 20. Kh1
The king sidesteps to avoid potential complications related to the open diagonal d4-g1. Consider 20. Rb3 h4 21. Be1 f5 22. Nf2 Bd4 where black gets some play with the centralized bishop.
20... h4 21. Be1 f5 (diagram)
Black increases pressure to the e-file and intends to break the center open. However, the f-pawn push has a long-reaching impact, as the black king will become somewhat open. Perhaps a more solid active option was to push the b-pawn, instead. But now white will break the queenside open.
Of course, white cannot take: 22. exf5?? Qxe2
|Position after 32. Re3|
22. b4 fxe4 23. bxc5 exf3 24. Bxf3 Rf8 25. Bf2 Ne5 26. cxd6
The queen cannot yet take due to the skewer 26... Qxd6 27. Bc5 winning the exchange.
26... Qf6 27. Be2
The bishop now protects against then Nc4 threat of forking the queen and the rook, but it also allows black to recapture the pawn on d6. The bishop needed to move anyways due to the threat of g4 forking the bishop and the knight.
The recapture is now possible, since the skewer can be dealt with an intermezzo check: 28. Bc5 Rxf1+ 29. Bxf1 Qf6.
27... Qxd6 28. Rb3 Rd8 31. Nb6 Nf6 32. Re3
Black cannot yet capture the d5-pawn: 32... Nxd5 33. Rxe5 Qxe5 34. Bc4 and the pin will win a full piece. For example: 34... h3 35. Bxd5+ Rxd5 36. Qxd5+ Qxd5 37. Nxd5. Trying to keep the material deficit in rook against two pieces was not better: 35... Kh7 36. Qd3+ Rf5 37. Qxh3+ Bh6 and white's rook and bishop pair will dominate on the open board against black's pair of rooks.
To enable capturing the d5-pawn, black sacrifices the h-pawn.
32... h3 33. Rxh3 g4 34. Re3 (diagram) Nxd5
Now the capture was possible, as the previous trick no longer works: 35. Rxe5 Qxe5 36. Bc4 Kh8 37. Bxd5. Now as the h-file has opened and the g-pawn pushed, black can play a strong move 37... Qh5 to begin adding pressure to the pinned bishop on d5. This allows black to equalize the material with a follow-up such as: 38. Kg1 Rf5 39. Qc2 Rfxd5 40. Nxd5 Qxd5. With so few pawns and pieces on board, the position would be practically a draw.
|Position after 38... Qxa5|
35. Nxd5 Qxd5 36. Qe1 Rc8 37. Bg1 Rxf1 38. Qxf1 Qxa5 (diagram)
After the exchanges, white is a pawn down, but the bishop pair is strong against an open king.
|Position after 49. Bc2|
39. Bd1 Rf8 40. Bb3+ Kh8 41. Qd1 Bf6 42. Qb1 Qc7 43. Qf5 Kg7 44. Re4 b6 45. Be3
White is improving the pieces to point towards the black king. The lack of a challenger for white's light square bishop is clear, as all black pieces have been driven to dark squares.
It is also remarkable how seemingly without a care white is handling the back rank with the bishop pair controlling the entry squares (c1/d1) and active mating threats requiring black's constant attention. For example, if black spent a tempo to create a back rank mate threat, it would be disastrous. For example: 45... Qc3 46. Rxg4+ Nxg4 47. Qxg4+ Kh8 48. Qh5+ Kg7 49. Qh6#
Therefore, black plays Rh8 instead to further protect against this particular mate threat idea.
45... Rh8 46. Rf4 Qd6 47. Bd4 a5 48. Qe4 Rh5 49. Bc2 (diagram)
Black is out of defensive moves. This is not quite a zugzwang, as given another move, white could play Qb7+ anyways transposing to game continuation after 51. Qb7+. To illustrate black's problems, trying to increase coordination would not have helped: 49... Rh6 50. Rxf6 Kxf6 51. Qf4+ Ke7 52. Bxe5 Qb4 53. Qe3.
|Position after 73... Kd6|
49... Kf8 50. Qa8+ Kg7 51. Qb7+ Qe7 52. Rxg4+
The rook cannot be captured as the bishop would become pinned: 52... Nxg4?? 53. Qxe7+ winning the queen.
52... Kf8 53. Qc8+ Qd8 54. Qxd8+ Bxd8 55. Rf4+ Ke7 56. Bd1
The final part of the puzzle is to drive the black rook away from protecting the knight on e5. Ideas on saving the knight would not work:
- Attempting to stay on the fifth rank is of no use: 56... Rg5 57. h4 Rg6 58. Bxe5. In fact, this would only give white a pawn advance for free.
- Attempting to create a counter-threat to capture a rook back fails to an intermediate check: 56... Nd3 57. Re4+ and black would lose the rook instead of the knight.
So, black gives up the knight. After this, white is a clean piece up and the rest is simple technique by engine terms.
56... Rh6 57. Bxe5 Re6 58. Rf5 Bc7 59. Bc3 Rh6 60. h3 Rd6 61. Bf3 Rh6 62. Bd2 Rh7 63. Kg1 Kd7 64. Kf2 a4 65. g4 Bd8 66. g5 Rxh3 67. g6 Bh4+ 68. Ke2 Rh2+ 69. Kd1 a3 70. g7 a2 71. Bc3 a1=R+ 72. Bxa1 Rf2 73. Rf7+ Kd6 1-0. (diagram)
White wins by the TCEC win rule.
There is no stopping from white queening on the next move. An example continuation: 74. g8=Q Rf1+ 75. Ke2 Rf2+ 76. Ke3 Rxf3+ 77. Kxf3 Bd8 78. Qxd8+ Kc5 79. Qc7+ Kb4 80. Qxb6+ Kc4 81. Rc7+ Kd5 82. Qc6#
Analysis results after 10... Re8:
- Stockfish 11-dev (10ead8a7); depth 37 with 16.1G nodes: 11. Bb5 +0.57; 11. Bd3 +0.55; 11. a4 +0.53; 11. h4 +0.13.
- Lc0 v0.24.0-dev (78d9cc1) + test 42810; with 7.1M nodes: 11. a4 +0.14; 11. Bd3 +0.11; 11. Bb5 +0.10; 11. Nge2 +0.05