The Cochrane Gambit at Linares '99

Position after:

(45) Topalov,V (2700) - Kramnik,V (2751) [C42]
XVI Ciudad de Linares 99 Linares ESP (8), 02.03.1999
[Lopez "Chess Kamikazes"]

1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nxf7!! The Cochrane Gambit. This is *tremendously* unusual at the GM level and doubly so at a tournament as important as Linares! White sacrifices a Knight in exchange for two pawns and an exposed Black King. The opening is (somwhat unjustly) considered very unsound. It's actually a good choice for correspondence players, as well as over the board players with a good tactical eye. Top level players are typically aware of how to meet the Cochrane, which is why it's rare in grandmaster practice. 4...Kxf7 5.Nc3 Montavon scored a quick 21-move win against Pichl last year at the Leipzig Open with this move, which is what may have motivated Topalov to play the Cochrane in this game. However, this is considered to be a fairly weak continuation. [5.d4 The Bronstein Attack, generally considered to be the main line of the Cochrane. W. John Lutes devotes considerable space to it in his book on the Petroff and Cochrane.] 5...c5 This is a "novelty". Kramnik inhibits White from building the center with d4-d5 (and possibly transposing into other Cochrane variations), [5...Qe7 This is the most popular reply. The Queen and f6-Knight gang up on the e4-pawn. In a famous theoretical in New in Chess 19, Osnos and Kalinichenko consider this move to be the refutation of 5.Nc3. Kramnik was either unaware of this article or thought he had a better idea.; 5...Be7 This is how Pichl met it in Leipzig. His idea is to develop his Kingside Rook and follow it up with Kg8 (essentially "castling by hand").] 6.Bc4+ Be6 7.Bxe6+ Kxe6 8.d4 White threatens to win a pawn with 9.dxc5 (the reply 9...dxc5 is, of course, impossible since the Black Queen is undefended). 8...Kf7 Kramnik misses it. He expected 9.d5+ instead of 9.dxc5. White now wins a pawn and full compensation for the sacrificed Knight. 9.dxc5 Nc6 10.Qe2 Qd7 Preparing to meet Qc4+ with ...Qe6. 11.Be3 White prepares to castle Queenside (to get his Rook to the same file as the Black Queen)... 11...dxc5 ...but Black nips this in the bud (White can't castle through check). 12.f4 [12.Qc4+ Qe6 13.Qxe6+ Kxe6 14.0-0-0=] 12...Re8 13.e5 Ng4 14.Rd1 Qf5 15.0-0 h5 The point is to expand the h8-Rook's control of the file. Black anticipates an eventual Qh5+ from White. By advancing the pawn, Black allows the h8-Rook to control h5 and prevents the White Queen from going to that square to start a King hunt. 16.Bc1 Black's logical plan was to trade the Knight for the dark-squared Bishop. Topalov decides to preserve his Bishop. 16...Nd4 Unfortunately, the withdrawal of the White Bishop allows the invasion of the Knight (the Bishop no longer covers d4). 17.Qc4+ Kg6 [17...Qe6 18.Qxe6+ Rxe6 19.h3 Nh6 Black has a space advantage, but his problems aren't over yet -- his King is still exposed.] 18.h3 Nh6 19.Nb5 a6 [19...Nxc2 20.Nxa7 Be7 Connecting the Rooks seems to be a good idea. The position is far from clear, though.] 20.Nxd4 cxd4 21.Qxd4 [21.Qd5 b5 22.Qc6+ Re6 23.Qc7=] 21...Rc8 [21...Qxc2 22.Rf2 Qc6] 22.Qb6+ Kh7 23.Qxb7 Rxc2 24.Be3 Preventing 24...Bc5+ 25.Kh1 Qg6 with some potential ugliness on g2 should the White Queen move from the long diagonal.. 24...Qg6 25.Rc1 Rxc1 [25...Qd3 26.Rxc2 Qxc2 27.Rc1 Qd3 28.Bf2 Nf5 White has a slight advantage (more actively placed pieces and better pawns).] 26.Rxc1 Nf5 Kramnik begins trying to break through on the Kingside. 27.Bf2 h4 28.Rc7 A nice idea, but it unfortunately allows Black too much counterplay. [28.Rc6 Driving the Queen from the g-file (the same file as the White King). This would have been White's best winning chance. 28...Qh5 29.Qxa6 White wins a pawn and the White Queen has rapid access to the Kingside along the a6-f1 diagonal. 29...Qd1+ (29...Ng3 This doesn't work now. 30.Bxg3 hxg3 31.Qd3+ Kg8 32.Qxg3) 30.Qf1] 28...Ng3 29.Kh2 [29.Bxg3 hxg3 With the possibility of a bank-rank mate: ...Qb1+ followed by Rc1 Qxc1#.] 29...Nf1+ 30.Kg1 Qb1 Looking for another mate in two: ...Ng3+ followed by Kh2 Qh1#. 31.Bxh4 Preventing the mate in two (...Ng3+ doesn't work now -- see the final variation). 31...Bc5+ [31...Bc5+ 32.Rxc5 Ng3+ 33.Kh2 Nf1+ (33...Qh1+ 34.Kxg3) 34.Kg1 Ng3+ with a draw by threefold repetition on the way.] 1/2-1/2

Generated with ChessBase 7.0