Game Of Missed Chances

PLAYED last night in the Battersea Club Championship.
Paul Stokes (158) - Spanton (168)
QGD 4...c5
1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Bg5 c5!?
This little-known line has been tried by many strong players since Steinitz played it in his 1886 match against Zukertort.
5.cxd5 cxd4
Steinitz preferred 5...exd5 against Zukertort. Another alternative is 5...Qb6!?, known as the Dutch-Peruvian Gambit.
6.Qa4+!?
The main line runs 6.Qxd4 Be7 7.e4 Nc6 8.Qd2 Nxd5!? 9.exd5 Bxg5 10.f4 Bh4+ 11.g3 exd5 12.gxh4 Qxh4+ 13.Qf2 Qe7+ 14.Qe2 Be6 15.Nf3 d4, as seen in John M Quinn (187) - Spanton (161), Metropolitan 1993 (0-1, 36 moves), and several other games. Black has two pawns and an initiative for a knight. The analysis engines Stockfish10 and Komodo10 prefer White after the continuation 16.f5, which does not appear in ChessBase's 2019 database.
The text, which may have been first played by Ludek Pachman in 1949, is still sometimes chosen by strong players.
6...Qd7
Black has insufficient compensation for a pawn after 6...Bd7?! 7.Qxd4 exd5 8.Bxf6 (not 8.Nxd5?? Qa5+) followed by capturing on d5.
7.Qxd4 Nc6 8.Qd2 Nxd5 9.Nxd5 Qxd5
Jonny Hector (2485) preferred the engines' choice 9...exd5 in a 1989 loss to Juan Bellon Lopez (2470).
10.Qxd5 exd5 11.0-0-0?!
The white king is vulnerable on the queenside. Safer was Pachman's 11.Nf3, or, as played by Vadim Malakhatko (2594) in 2006, 11.a3.

Black to make his 11th move

*****
*****
*****
*****
*****
11...Be6
A routine move, but I had analysis from 1999 showing the strength of 11...Bf5! The point is that 12.Rxd5 is good for Black after 12...Rc8 (my note from 1999) or the engines' 12...Be6/12...Be4. Indeed the engines reckon White's best is 12.e4!?, when Black regains his pawn and still has threats against the white king.
12.Nf3 Rc8 13.Kb1 Bc5 14.e3 f6 15.Bf4 0-0 16.Be2 Nb4?
This would have been a good reply to 16.Bd3?, but here it is positively harmful as White gets to make a useful defensive move with tempo.
17.a3 Nc6 18.Rc1 Be7 19.Rhd1 Rfd8 20.h3 a6?!
Komodo10's second-choice for a while, but there was no need to guard the b5 square. A queenside attack with ...b5 is unlikely to ever be good and Black now has to be alert to a possible Bxa6.
21.g4
Stockfish10 wants to sac the exchange with 21.Rxc6!? when both 21...Rxc6 and 21...bxc6 are met by 22.Nd4.
21...g6 22.Bg3
The exchange sac is still possible but the engines reckon the text is even stronger.
22...Kg7 23.Bf1
Preparing to bring pressure to d5 by putting this bishop on the long light-square diagonal. Again the engines reckon Rxc6!? is a reasonable alternative.
23...f5?
The second-choice of both engines for a while, but I regretted the move almost as soon as I played it.
24.Ne5?
Exchanging pieces usually helps the player who does not have the isolani, but here Black's king is well-placed for an ending, and anyway 24.g5 makes Black's light-square bishop look silly.
24...Nxe5 25.Bxe5+ Bf6 26.Bxf6+ Kxf6 27.f3?!
This appears reasonable, but the engines do not like it, reckoning White is equal after 27.Be2.
27...Rxc1+ 28.Rxc1?!
Black is just a little better after 28.Kxc1 fxg4 29.fxg4 (or 29.hxg4 h5) Kxg5.
28...Rc8?
I rejected 28...d4! because of the obvious 29.Rd1, but the engines reckon Black is well on top after 29...Kg5. Also good for Black is the immediate 28...Kg5.
29.Rxc8 Bxc8 30.Kc2
PS offered a draw.
30...fxg4 31.hxg4!?
This lets Black create a passed h pawn, but it is not clear that 31.fxg4 is better. The engines reckon both moves give an equal position.
31...h5 32.gxh5 gxh5 33.Kd3 h4

White to make his 34th move

*****
*****
*****
*****
*****
34.Kd4?
White holds the balance with 34.Ke2.
34...Bd7?
34...h3 wins, eg 35.Kxd5 h2 36.Bg2 Kg5 etc.
35.Kxd5 Bc6+
The engines reckon 35...h3!? is a draw after, for example, the fascinating line 36.e4 h2 37.Bg2 Kg5 38.e5 Kh4 39.e6 Bc6+ 40.Kd4 Kg3 41.Bh1 Kf2 42.e7 Kg1 43.Ke3 Kxh1 44.Kf2.
36.Kd4 Bxf3 37.Bh3?
37.e4 holds.
37...Kg5 38.Ke5 Bg4 39.Bg2 Bc8?
39...h3 wins, eg 40.Bxb7 h2 41.Bh1 Be2 42.Kd6 Kg4 43.e4 Bf3 etc.
40.Kd6 Kg4 41.e4 Kg3 42.Bh1 Kh2 43.Bf3 Kg3 ½–½