Some books give the impression exchange sacrifices are available in almost every game.
But at club level they are rare birds indeed - the exchange is often lost, but seldom sacrificed.
Depending on whom you take as your authority, the exchange is worth somewhere between 1.5 and 1.75 pawns, with the latter figure being more commonly favoured.
One of the problems at club level is that most players still use the traditional piece-value table: P-1, N-3, B-3, R-5, Q-9.
This creates several anomalies, eg giving up bishop and knight for rook and pawn on f7 would seem to be an even trade - 6pts each - and so should favour the player giving up the bishop and knight as the opponent loses castling rights. In practice, this is nearly always a bad trade.
Another anomaly is that a minor piece is no stronger than three pawns - again this is rarely the case in the opening or the middlegame (although in an ending it all depends on the placement of the remaining material).
The traditional table also gives the impression that two pawns are needed to compensate for the exchange.
I prefer to use a table discovered - the right word, I believe - by international master (now grandmaster) Larry Kaufman.
He used a computer to analyse almost 300,000 games, coming up with the following average piece values: P-1, N-3.25, B-3.25, R-5, Q-9.75 (his full article, which is a fascinating read, can be seen here: https://www.chess.com/article/view/the-evaluation-of-material-imbalances-by-im-larry-kaufman).
The values are not hugely different from the traditional ones, and it is nice to see confirmation of the amazing coincidence that knights and bishops, despite being very different in their actions, have the same average strength. Think how difficult it would be to devise a game from scratch with such a feature.
But while knight and bishop are on average equal, two bishops are better than knight and bishop or two knights. Kaufman's computer analysis showed that on average the bishop-pair is worth half a pawn.
This new table immediately clears up the anomalies mentioned above. Giving up bishop and knight for rook and pawn is now losing 6.5pts for 6pts. Actually, taking into account that the bishop-pair is lost, it is actually giving up 7pts for 6pts - rarely will positional factors compensate for that.
Similarly, three pawns are no longer worth a bishop (as with all material evaluations, the caveat "on average" is implied).
Getting back to sacrificing the exchange, it can be seen that the sacrificer is giving up 1.75pts. So getting a pawn for the exchange loses 0.75pts. And if the winner of the exchange has had to give up the bishop-pair, the material deficit is just 0.25pts.
In other words, giving up the exchange when you have a pawn and the bishop-pair as compensation is almost an even trade - the placement of the remaining material on the board is likely to determine whether the sacrifice is sound.
All of this is a preamble to my game last night in round three of the Pimlico Summer open tournament.
 |
| Black to make his 17th move in Richard Black (188) - Spanton (171) |
*****
*****
*****
*****
My main analysis engines want Black to play around the knight, eg Stockfish10 gives 17...Qc6 18.b4 Be6 19.Qd3 with a slight edge to White. But I felt I had to get rid of the monster.
17...Bf8!? 18.Bg5 Bxd6
18...Be7 19.Bxe7 seemed positionally hopeless to me.
19.Bxf8
White takes the exchange. I was actually more worried about 19.cxd6!?
19...Bxc5
For the exchange, Black has a pawn and the bishop-pair - as discussed above, this is almost full material compensation. Positionally, at first glance, Black might even seem to be better as White has doubled and isolated b pawns. However, Black has problems smoothly completing development, and this gives White the initiative.
20.Bf6?!
This natural-looking move is not liked by the engines, and their reasoning will soon become apparent.
20...Bd4 21.Qd2 Qf5 22.Bg5
Intending to get rid of the bishop-pair by playing Be3, but the manoeuvre Bd8-f6-g5 has cost White a tempo.
22...Be6
Stockfish10 and Komodo9 slightly prefer 22...Bd7!? The point after 23.Be3 is to continue 23...Bc6 24.Bxe3 Rd8, when Black has lost the bishop-pair but gained active pieces.
23.Be3 Rd8 24.f4?!
This looks weakening. The engines prefer 24.Rfe1, eg 24...Bxb3 25.Qb4 Bd5 26.Bxd4 Qg5 27.g4 exd4 28.Qxd4, when only White can reasonably hope to win, but Black seems to have decent drawing chances.
24...Bxb3
I rejected 24...Qe4 because of 25.Rae1 exf4 26.Rxf4??, missing that Black can simply take the rook on f4 because White's bishop is pinned.
25.fxe5 Qxe5 26.Bxd4 Qxd4+ 27.Qxd4 Rxd4 28.Rac1 Be6
Could I have played 28...a5 and tried to activate my queenside majority? Stockfish10 is at first quite keen on the idea, but its evaluation soon changes to =. Komodo9 also quite likes the move, but reckons the position remains even. To me, 28...a5 looked rather loosening.
29.Rfd1 Rb4 30.Rd2 h5 31.Rc3 Kg7 32.Kf2 Kf6 33.Ke3 Ke5 ½–½
No comments:
Post a Comment