When to Sacrifice

It’s important to learn about the idea of sacrificing a piece for a pawn or two in front of the castled enemy king. There are many books which will show you examples of this, and all teachers enjoy showing these games to their students.

But if you read these books and follow these lessons you may well get the idea that such sacrifices always work. In real life, of course, they don’t, but it’s much more entertaining to demonstrate a short sacrificial win than a long game where the defender refutes the sacrifice and wins the ending with his extra piece.

Sometimes you can calculate a forced win. Sometimes it will be obvious to an experienced player (but not necessarily to someone less experienced) that the sacrifice won’t work. There’s also a grey area in the middle where you clearly have compensation. It’s partly a question of style, temperament and personality whether or not you will choose to make the sacrifice. You might also like to remember the rule (something I read years ago, I think in a book by Julian Hodgson) that you should play the sacrifice if, after giving up the piece, you have more attackers than your opponent has defenders.

The game below was posted on Facebook by my colleague at CSC, Ferris Lindsay. He asked why he lost this game. Let’s have a look.

1. e4 e6
2. d4 d5
3. exd5 exd5

White chooses the Exchange Variation in reply to his opponent’s French Defence. This is often a sign of peaceful intent, but Ferris has an aggressive plan in mind.

4. Nc3 Nf6
5. Bg5 c6
6. Nf3 Be7
7. Qd2 O-O
8. O-O-O

Natural moves, I suppose, but this position has only been seen in a few fairly low level amateur games. Superficially it looks a bit scary for Black, who has to reckon with a possible sacrifice if he ever tries to play h6. So we might want to consider why stronger players don’t start the game in this way.

It’s worth considering what would happen if Black decided to put the question to the bishop at this point. After 8… h6 White can consider two thematic sacrifices. Do either of them work. You might like to go away and analyse this position yourself before reading on, or, if you’re a chess teacher, set it as an exercise for your students.

So: 9. Bxh6 gxh6 10. Qxh6 when Stockfish finds several viable defences for Black, which may not, however, be so easy to find over the board at amateur level. One such defence is 10… Ne4 with the idea of Qd6 as well as a possible fork on f2. Then 11. Nxe4 dxe4 and the knight on f3 can’t move because of the nasty Bg5 pinning and winning the queen. Now Black’s not threatening exf3 because gxf3 will open the g-file for a white rook, but he’ll be able to play Qd6 next move and defend comfortably.

Or White could try 9. h4 hxg5 10. hxg5 hoping for an attack down the h-file. Now Black has only one way to gain an advantage: 10… Ne4 (against other knight moves 11. Bd3 will give White at least a perpetual check) 11. Nxe4 dxe4. Again the knight can’t move because of Bxg5 and this time, because of the pawn on g5, Black will be able to capture on f3 next move. After 12. Bc4 exf3 13. Qd3 Bxg5+ 14. Kb1 Re8 15. Qh7+ Kf8 White is two pieces down but has a strong attack. Not quite strong enough, according to Stockfish.

8… h6 is fine if you have a silicon brain, but a human would have to be pretty brave and pretty good at defending to play it.

Instead Black played…

8… Nbd7
9. h4

A useful move as White might have a rook lift via h3 in some lines.

9… Bb4

Black decides to take action on the queen side.

10. Bh6

This is it. Ferris sacrifices a bishop even though he’s only going to get one pawn in return. My first thought as a more experienced player is that this really isn’t going to work. Before we see what happened in the game, we’ll consider a more measured approach for White.

White’s most natural move seems to be 10. Bd3 when Black could continue with Qa5, a logical follow-up to his previous move. Given the subject of this article, though, we might also consider 10… h6 and decide whether 11. Bxh6 works. Again you might want to step back and decide on the blue pill or the red pill before going any further.

This time the sacrifice seems to work for White because he’s already played h4. One possible variation runs 10. Bd3 h6 11. Bxh6 gxh6 12. Qxh6 Re8 13. Ng5 (threatening mate in 3 starting with 14. Bh7+) 13… Nf8 14. Rh3 Bxh3 15. gxh3 Qd6 16. Rg1 and Black has to play 16… Re1+ to gain time to stave off White’s attack.

But back to the game…

10… gxh6
11. Qxh6 Kh8

A typical panic reaction in the face of White’s sacrifice. This sort of move is often seen in games at this level. Black moves his king because he’s scared of a possible check. He should have preferred 11… Ng4, driving the queen back, when he should be able to defend.

12. Bd3

Developing and attacking, but 12. Ng5, preventing an immediate Rg8, would have given Ferris an extra move for his attack and put him back in the game. 12. Ng5 Qe8 13. Bd3 Rg8 14. Nxh7 Ng4 15. Qd2 when White has won a second pawn for the piece and exposed the black king.

12… Bxc3
13. Ng5

White bravely continues with his attack rather than recapturing the piece. 13. bxc3 would have given Black time to defend with 13… Rg8

13… Bxb2+

Black seems determined to lose his bishop. Instead the natural move 13.. Bxd4 was correct, giving him a vital extra defender. Then 14. Bxh7 Ng4 15. Qh5 Ndf6 rather amusingly traps the white queen, while after 14. Nxh7 Black just has time to defend: 14… Ng8 15. Qh5 Bg7 16. Nxf8+ Bh6+ 17. Kb1 Qxf8.

14. Kd2

Likewise White insists on not capturing the bishop. This move should lose, but after 14. Kxb2 Qb6+ 15. Kc1 Qxd4 16. f3 Qb4 17. Bxh7 Qa3+ is one possible variation, leading to a position where both players have a perpetual check.

14… Qa5+

Moving the queen offside (14… Bxd4 was easier) but should still be good enough.

15. Ke2 Re8+

A tempting check, but it’s dangerous to undefend f7 in this way. 15… Bxd4 was still the simpler option.

16. Kf1 Re4

A very complicated position has arisen. Black tries block the attack on h7 by returning material, but, as it happens, the move is a mistake.

Yet again 16… Bd4 was correct. Then after 17. Nxh7 Black should gain time by returning one of his extra pieces: 17… Ng4 18. Qh5 Nh6 19. Qxh6 Bg7 20. Qh5 Kg8 – but I guess these moves would be very difficult to find over the board.

17. Nxf7+

What could be more natural than capturing an undefended pawn with check? But, unfortunately for Ferris, it loses.

Taking the rook also loses: 17. Bxe4 dxe4 18. Rh3 Qb5+ 19. Kg1 Qe2 20. Rf1 (an amusing position, given that White castled queen-side 12 moves earlier) 20… Qh5 and the black queen has managed to get back to defend her king.

But, amazingly, Stockfish reveals that there is a win here. The improbable winning move is 17. Rh3, with the slow but unanswerable threat of Rg3, forcing mate. The best Black can do is give up his queen: 17… Qc7 18. Rg3 Qxg3 19. fxg3 Kg8 when Black has rook (about to be captured), bishop and knight for queen. Paradoxically, White needs his knight and bishop for the attack, so he should use his rook, not a minor piece, to trade off the intruder on e4. Stockfish continues: 20. Re1 b6 21. Rxe4 dxe4 22. Bc4 when, because of the threats on f7, Black has nothing better than Ba6, shedding a piece.

Of course working all this out with your clock ticking is far too hard for most humans.

17… Kg8
18. Ng5 b6

Black is two pieces up but still needs to defend accurately. This move lets White back into the game. There were preferable defences such as Re7.

19. f3

Ferris misses his last chance. Again he could have considered the rook lift: 19. Rh3 (threatening both 20. Rg3 and 20. Nxe4 dxe4 21. Bc4+) when Black’s only route to an advantage is 19… Re7 when Stockfish analyses two variations: 20. Bxh7+ Nxh7 21. Re3 Ba6+ 22. Kg1 Qb4 23. Rxe7 Qxe7 24. Qg6+ Kh8 25. Nf7+ Qxf7 26. Qxf7 Ndf6 when White has queen and two pawns for four minor pieces in a position which looks totally unclear to me or 20. Nxh7 Bxd4 21. Rg3+ Kf7 22. c3 Ba6 23. Qg6+ Ke6 24. cxd4 Bxd3+ 25. Rgxd3 Rg8 and Black’s king just about survives.

19… Ba6
20. fxe4 dxe4

White has regained most of his material but now it’s Black who has the winning attack. Another insufficient try here is 21. Ne6 Bxd3+ 22. cxd3 Qf5+ 23. Nf4 Rf8.

21. Kg1 Bxd4+ 22. Kh2 Qe5+ 23. g3 exd3 24. Rde1 Qc5 25. cxd3 Qc2+ 26. Kh3 Qxd3 27. Re7 Qg6 28. Qxg6+ hxg6
29. Re6 Re8 30. Rxc6 Re3 31. Rc7 Nc5 32. Rxa7 Bc8+ 33. Kg2 Re2+ 34. Kf1 Rf2+ 35. Ke1 Nd3+ 36. Kd1 Bg4+
37. Nf3 Bxf3#

So was Ferris right to make the sacrifice? In theory, no, because the sacrifice could have been refuted simply by 11… Ng4. In practice, no, because he eventually lost the game. But, you know what? I think he was indeed right to sacrifice. He will have learnt a lot about when sacrifices against the castled king are or aren’t likely to work, and will have learnt also about how to calculate and what sort of moves to look for in that sort of position. The beauty of online games is that the result doesn’t really matter. There’s no prize at stake. There are no teammates to complain if you lose. So you can afford to try things out and treat your games as a learning experience.

Richard James

This entry was posted in Annotated Games, Articles, Improver (950-1400), Intermediate (1350-1750), Richard James on by .

About Richard James

Richard James is a professional chess teacher and writer living in Twickenham, and working mostly with younger children and beginners. He was the co-founder of Richmond Junior Chess Club in 1975 and its director until 2005. He is the webmaster of chessKIDS academy (www.chesskids.org.uk or www.chesskids.me.uk) and, most recently, the author of Chess for Kids and The Right Way to Teach Chess to Kids, both published by Right Way Books. Richard is currently the Curriculum Consultant for Chess in Schools and Communities (www.chessinschools.co.uk) as well as teaching chess in local schools and doing private tuition. He has been a member of Richmond & Twickenham Chess Club since 1966 and currently has an ECF grade of 177.