Dvoretsky's Endgame Manual Read online

  1 Rd5! Rd2+!

  Now this rook endgame, after 1…ed 2 R×a5, is harmless for White: 2…d2 3 Rd5, or 2…Ke6 3 Re5+ Kd6 4 Kf2 d2 (4…Rg1 5 Re3) 5 Ke2 Rg1 6 K×d2 R×g3 7 Re3. But, thanks to the pin of the knight on the d-file, Black has an in-between rook check, securing him a transition into a favorable pawn endgame.

  2 Kf1 R×d3 3 R×d3

  A forced exchange – otherwise, the g3-pawn is lost.

  3…ed 4 Kf2

  Unfortunately, here too it is not clear whether White can be saved. Nonetheless, pawn endings are generally of a more forcing nature than rook endings are. Here, as a rule, it is possible to obtain an accurate appraisal of the position, if you can take a variation to its conclusion. So it makes sense to concentrate our efforts on the calculation of this pawn endgame.

  Black has two plans of action: bringing the king to the center, in the hope of putting his opponent in zugzwang, and the kingside break with g6-g5.

  We can easily establish that the first plan is harmless: 4…Ke6 5 Ke3 Kd6 (5…Kd5 6 K×d3 h6 7 g4) 6 K×d3 Kd5 7 g4 h6 (on 7…fg 8 hg h5, there is 9 f5!, although 9 gh gh 10 Ke3 also does not lose) 8 g5! hg (8…h5 9 h4) 9 fg f4 10 h4 Ke5 11 Ke2, with equality.

  4…g5! 4 fg+!

  5 Ke3? would be a mistake: 5…gf+ 6 gf Ke6 7 h4 Kd5 8 K×d3 h5.

  5…K×g5 6 Ke3 h5 7 K×d3 h4

  We have already seen the situation occurring after 7…f4 8 gf+ K×f4 in the previous example. White’s king cannot attack the a-pawn – but this is not necessary: it is enough to squeeze the enemy king onto the h-file. For example: 9 h4 (or 9 Kd4) 9…Kg4 10 Ke4 K×h4 11 Kf4, with a draw.

  8 gh+ K×h4

  Which of the two natural moves – 9 Ke3 or 9 Ke2 – should White make? Let’s go back to the rule laid down above. After Black wins the h-pawn, our “arithmetic” shows that he will have one tempo, since the f-pawn is one square above the crucial c1-h6 diagonal. White can only save himself if he can force that pawn forward to f4.

  So clearly, 9 Ke2? loses: 9…Kg3 10 h4 (10 Kf1 K×h3 11 Kf2 Kg4, and Black now has two tempi) 10…K×h4 11 Kf3 Kg5 12 Kg3 Kf6 13 Kf4 Ke6 14 Kf3 Kd5, etc.

  9 Ke3! Kg3 10 Ke2

  Now it is Black who is in zugzwang: he must advance his pawn to f4, since 10…Kg2 11 Ke3 Kg3 12 Ke2 is a useless “pendulum.”

  10…f4 11 Kf1 K×h3 12 Kf2 Kg4 13 Kg2

  And we have the “normal” drawn position.

  Matanovic was unable to calculate the pawn ending accurately, and so preferred to keep rooks on. Unfortunately for him, the rook ending turned out to be lost.

  Let’s return to Diagram 1-130.

  1 Rd6+? Ke7 2 Ra6 ed! 3 R×a5 Kd6 4 Ra8 Kc7 (Black repeats moves to gain time to think) 5 Ra5 Kc6 6 Ra8 Kc5 7 Kf2

  7…Ra1?

  7…Kd4! 8 a5 Ra1 would have won. For example: 9 a6 (9 Rd8+ Kc3 10 Rc8+ Kd2 would lead to approximately the same situation as in the game) 9…d2 10 a7 d1N+! (the only way; now wherever the king retreats, it will be checked by the knight, allowing it to get to d5) 11 Kf3 Ra3+ 12 Kg2 (12 Ke2 Nc3+) 12…Ne3+ 13 Kf2 Ra2+ 14 Kf3 Nd5 15 g4 Nc7–+. After the text, White manages to save himself by sacrificing several pawns: 8 Ke3! Rg1 9 g4! fg 10 hg Rg3+ (10…R×g4 11 Rd8!?) 11 Kd2 R×g4 12 f5! gf 13 K×d3=.

  8 Rd8? Kc4 9 Ke3 Re1+ (9…Rg1? 10 Rd4+) 10 Kf2 Re2+ 11 Kf3 Re6! 12 a5 Kc3 13 Rc8+ Kd2! 14 h4

  According to Botvinnik’s analysis, 14 Rc7 would not have saved White either: 14…h5 (14…Re1? 15 a6 Ra1 16 a7) 15 Kf2 Kd1 16 Kf3 d2 17 Kf2 Re2+! 18 Kf1 Re3 19 a6 (19 Kf2 Ra3, followed by …Ra1-c1) 19…R×g3 20 a7 Ra3 21 Kf2 h4 22 Kf1 Ra4 23 Kg2 Ke2 24 Re7+ Kd3 25 Rd7+ Ke3.

  14…Re1! 15 a6 Ra1

  Now on 16 Rc6 Ke1 is decisive: 17 Re6+ Kf1 18 Rd6 (18 Ke3 Re1+) 18…d2 19 R×d2 Ra3+, and White gets mated! The same thing happens after 16 Ra8 Ke1 17 a7 d2 18 Re8+ Kf1 19 Rd8 Ra3+.

  16 Rc7 Ke1 17 Kg2 R×a6 18 Re7+ Kd1 19 R×h7 Ra2+ 20 Kf1 d2 21 Rc7 Ra1 22 Kf2 Rc1 White resigned.

  Tragicomedies

  Colle – Grünfeld

  Karlsbad 1929

  Grünfeld resigned, not realizing that, by squeezing the white king onto the h-file, he had an easy draw.

  1…Kd3! 2 Kg5 Ke4 3 K×h5 Kf5=, etc.

  Winants – L. Hansen

  Wijk aan Zee 1994

  The position is drawn. Black tries one last chance:

  1…f3!? 2 gf+??

  The correct 2 g3! f4 3 gf K×f4 leads to the “normal,” i.e., drawn, position.

  1…Kh3!

  White resigned, since after 3 Ke3 Kg3 4 f4 Kg4 5 Ke2 K×f4 6 Kf2, Black has two extra tempi (even one would have been enough), since the pawn is above the c1-h6 diagonal and the king is in front of the pawn.

  Exercises

  The Protected Passed Pawn

  The protected passed pawn, like the outside passed pawn, is usually a most definite positional advantage. The enemy king cannot leave its square, and cannot capture it, whereas our king has full freedom of movement.

  Two Pawns to One

  These positions are generally won.

  1 Kd3 Kd5 2 Ke3 Ke5 3 Kf3 Kd5 4 Kf4 Kd6 5 Ke4 Ke6 6 Kd4 Kd6 7 Kc4 (Black must give up the opposition) 7…Kc7 8 Kd5!

  8 Kc5?! is inaccurate: 8…Kb7, and White cannot continue 9 b6? because of 9…Ka6! 10 Kc6 stalemate.

  8…Kb6 9 Kd6 Kb7 10 Kc5 Kc7 11 b6+ Kb7 12 Kb5+–.

  Now let’s look at the two most important drawn positions. The first is an elementary one, but it comes up rather regularly. The second is less likely to occur, but it is very instructive.

  Black plays 1…Kb8 2 Kc6 Kc8=.

  Move the whole position one file to the right, and White wins easily by sacrificing the pawn and then winning the enemy’s last pawn.

  F. Dedrle, 1921

  The key squares are c4, d4, and e4. Black can protect them, if he can control the opposition when the enemy king approaches.

  Let’s determine the corresponding squares. With White’s king on d3, f3 or h3, Black’s king must occupy d5; the e3- and g3-squares correspond to e5. When the king advances farther, Black must keep the lateral opposition, maneuvering along the d- and e-files.

  When White’s king is on the second rank, Black’s king must stay next to the d5- and e5-squares – specifically, on d6 or e6. So the first move – as well as all the play that follows – now becomes clear:

  1…Kd6! (but not 1…Kc6? 2 Kg3! Kd6 3 Kf4! Kd5 4 Kf5+–; or 2…Kc5 3 Kg4! Kd4 4 Kf4+–) 2 Kh3 (2 Kf2 Ke6! 3 Ke2 Kd6!) 2…Kd5! 3 Kg3 Ke5! 4 Kh4 Kd4! 5 Kh5 Kd5! 6 Kg6 Ke6!, etc.

  Multi-Pawn Endgames

  The next example features a typical plan for exploiting the advantage.

  I. Bottlik, 1952

  Black may have an extra pawn, but his position is difficult. How does he meet the threatened invasion of the white king?

  1 Kd4 f4 2 Ke5 a5! 3 a3 a4

  Black would prefer to exchange a pair of pawns; but after 3…ab 4 ab, he is in zugzwang.

  4 Kf5 d4! 5 Ke4 d3 6 K×d3 Kd5 7 g3!

  A necessary undermining of the enemy kingside pawns (undermining, by the way, is the theme of our next section). It is only a draw after 7 Ke2? g4 8 Kd3 Ke5 9 c6 Kd6 10 Ke4 K×c6 11 K×f4 Kd5 12 K×g4 Kc4, because both pawns will queen.

  7…fg

  No better is 7…Ke5 8 gf+ gf 9 Ke2 Kd5 10 Kf3 Ke5 11 c6 Kd6 12 K×f4 K×c6 13 Ke5+–. This position would be a draw if Black had exchanged pawns on his 3rd move; but if he had, alas, he would not have gotten this position.

  8 Ke2 (or 8 Ke3 g4 9 Ke2) 8…Ke5 (8…g4 9 Kf1 changes nothing) 9 Kf3 Ke6 10 K×g3 Kf5

  Here is a typical position with a protected passed pawn versus an outside passed pawn. The mined squares are g4 and f6. Most often (as here), the stronger side is unable to place his opponent in zugzwang. The only thing to be done then is to advance one’s own passed pawn, and exchange it for the other side’s passed pawn. Sometimes this wins, sometimes not. In the similar situation that occurred as one of the variations of the game Averbakh-Bebchuk (Diagram 1-112), we were able to lose a move to Black, making use of our reserve tempo (in that case, h4-h5).

  11 Kf
3 Ke5 12 Kg4 Kf6 13 c6 Ke6 14 K×g5 Kd6 15 Kf5 K×c6 16 Ke6+–.

  Tragicomedies

  Here we shall include examples of overestimating the power of the protected passed pawn.

  Shirov – Timman

  Wijk aan Zee 1996

  1 Kf3! looks good. Black responds 1…Re6, and if White defends the pawn by 2 Rc3, then Black will have decent drawing chances, in view of the passive position of White’s rook. White could trade off the rooks by playing 2 Rd8+ Kf7 3 Rd7+ Re7 4 R×e7+ K×e7 – this leads by force to a queen endgame, where White has an extra g-pawn: 5 Kg4 Kd6 6 Kg5 K×c6 7 Kh6 Kb7 8 K×h7 c5 9 K×g6 c4 10 f5 c3 11 f6 c2 12 f7 c1Q 13 f8Q Qc2+ 14 Qf5 Q×h2. Objectively speaking, this position is won (see Chapter 12); however, converting this advantage is not easy, and would take another several dozen moves.

  None of this appealed to Shirov. The grandmaster discovered what seemed to him like a more forcing means to the desired end.

  1 g4? Re6 2 Rd8+?! (2 Rc3) 2…Kf7 3 Rd7+ Re7 4 R×e7+ K×e7 5 g5!

  And in this position, Timman resigned. Both sides believed that 5…Kd6 was refuted by the pawn breakthrough 6 h4 K×c6 7 f5 gf 8 h5 Kd7 9 g6 hg 10 h6+–.

  But Black does not have to take on f5! He could draw with the continuation 7…Kd7 8 f6 Ke6(e8). This is the same position as in our previous example (a protected passed pawn versus an outside passed pawn). The mined squares are c6 and d8. With kings at d5 and d7, Black plays 1…Ke8!, after which neither 2 Ke6 Kf8 3 f7 c5, nor 2 Kc6 Kd8 3 f7 Ke7 4 K×c7 K×f7 will do better than draw. If Black wants, he can even leave his pawn at c6, instead of c7.

  Aronin – Smyslov

  USSR ch, Moscow 1951

  Black’s position is quite hopeless. The simplest solution is 1 R×c6, or 1 Rg8 Kh7 2 Re8! 3 Re7. However, the game was adjourned here, and Aronin chose, after home analysis, to cash in his advantage by entering a pawn ending.

  1 Rg8 Kh7 2 R×g7+?? R×g7 3 N×g7 K×g7 4 g4

  Before marching his king over to the queenside, White wishes to close up the kingside, to prevent Black’s potential counterplay by …f6-f5 and …g5-g4. Aronin examined the lengthy variation 4…Kf7 5 Ke2 Ke6 6 Kd3 Kd6 7 Kc4 a5 (7…c5 8 Kb5) 8 f3 Kd7 9 Kc5 Kc7 10 c3 bc 11 bc Kb7 12 Kd6 Kb6 13 c4 Kb7 14 c5+–.

  He did not think the exchange on g3 was playable, since White then gets the possibility of creating an outside passed h-pawn. However, Smyslov found an elegant defense: he offered his opponent, not an outside passed pawn, but a protected passed pawn!

  4…hg! 5 fg g4!! 6 h4 c5 7 Ke2 Kh7 8 Kd3 Kh6

  It turns out that the king can go no further: 9 Kc4? f5! 10 ef (10 Kd3 f4 11 gf ef 12 Ke2 Kh5 13 e5 Kg6, and 14…Kf5–+) 10…e4–+.

  9 c3 a5 10 cb ab! Draw.

  Exercises

  Undermining

  Sometimes the pawns are too strong to be successfully attacked by the king. In such cases, undermining can be used successfully – the exchange of a pair or two of pawns, with the aim of weakening the pawn chain.

  Keres – Alekhine

  Dresden 1936

  Grigoriev demonstrated the simplest winning method, involving an undermining on the queenside.

  1…Ke5! 2 Ke2 Kd6 3 Ke3 Kc7 4 Ke2 Kb7 5 Ke3 a6 (5…a5) 6 ba+ K×a6 7 Ke2 Kb7 (7…b5?? 8 d6 Kb6 9 cb=) 8 Ke3 Kc7 9 Ke2 Kd6 10 Ke3 b5 11 cb K×d5–+

  On the other hand, Alekhine’s plan of going into a queen endgame was also quite strong.

  1…Kg4!? 2 d6 g2 3 Kf2 Kh3 4 d7 e3+! 5 Kf3 g1Q 6 d8Q Qf2+ 7 Ke4 e2 8 Qd7+ Kg2 9 Qg4+ Kf1 White resigned.

  Tragicomedies

  Golberg – Zhuk

  USSR 1934

  1…a4 (1…g6?? 2 h6 g5 3 f5 wins; 1…g5? 2 f5=)

  2 g5+ Kf5??

  The only winning plan was undermining with …f7-f6. But first, Black had to bring his king to h7. As Grigoriev pointed out, the right way was 2…Ke7! 3 Kc3 Kf8 4 Kb2 Kg8 5 Ka3 Kh7 6 Kb2 f6! 7 Ka3 fg 8 fg Kg8 9 Kb2 Kf7 10 Ka3 Ke6 11 Kb2 Kf5.

  3 Kc3 Ke6?

  Having let the win slip, Black now lets slip the draw, which he could have had by playing 3…f6! 4 g6 Ke6.

  4 h6 gh 5 gh Kf6 6 f5

  Black resigned. We have already seen the final position in the chapter devoted to the rule of the square.

  Sulipa – Gritsak

  Lvov 1995

  1…g5??

  Far from improving Black’s position, this move actually degrades it significantly by giving his opponent the possibility of exchanging a pair of kingside pawns, and creating a passed h-pawn.

  White had no answer to the undermining plan with 1…Kd7! 2 Kf3 Kc7 (2…Ke7 is also good: 3 Ke4 Ke6 4 Kf3 Kd5 5 c6 Kd6 6 Ke4 a6–+, or 4 h4 Kd7! 5 Kf3 Ke7 6 Ke4 Ke6, triangulating again and again with the king, until the opponent runs out of pawn moves) 3 h4 Kc8 (not 3…Kb7 4 Ke4 a6? at once, in view of 5 ba+ K×a6 6 c6! Kb6 7 K×e5 f3 8 Kd6 f2 9 c7=) 4 Ke4 Kb7 5 h3 Kc8 6 Kf3 Kc7 7 Ke4 Kb7 8 Kf3 a6!–+

  2 Kf3??

  The wrong order of moves. After 2 h4! gh 3 Kf3 Kd5 4 c6 Kd6 5 Kg4 a6 6 ba K×c6 7 K×h4 Kb6 8 Kg4 K×a6 9 h4, it is now White who wins.

  2…Kd5 3 c6 Kd6??

  3…e4+! was necessary: 4 Kg4 Kd6–+.

  4 Ke4??

  For the fourth time, the appraisal of the position is reversed. White wins with 4 h4! gh 5 Kg4.

  4…a6 5 ba K×c6 6 Kf3 Kb6 7 h4 (too late!) 7…gh 8 Kg4 K×a6 9 K×h4 Kb6 10 Kg4 Kc6 11 h4 Kd6 White resigned.

  Two Connected Passed Pawns

  B. Horwitz, J. Kling, 1851

  Here we have a typical situation with two connected passed pawns. The draw would appear to be inescapable, since the white king is tied to the square of the protected passed pawn at c4. But in fact, in such cases White can sometimes leave the square to help his pawns queen or mate his opponent.

  White’s plan usually consists of the following elements:

  The farthest possible advance of the pawns;

  The optimum placement of the pawns – “ready to roll”;

  Choosing the best time for the king’s decisive advance.

  Let’s watch this plan in action. In the first stage the king, without leaving the square of the c4-pawn (which ends at f4), aids in the advance of its pawns.

  1 Kd4 Kg4 2 h4 Kh5 3 Ke3 Kg4 4 Ke4 Kh5 5 Kf4 Kh6 6 g4 Kg6 7 h5+ Kh6 8 Kf3 Kg5 9 Ke4 Kh6 10 Kf4

  Triangulation is White’s most important weapon in this ending.

  10…Kh7 11 g5 Kg7

  12 g6!

  The ideal pawn array! The erroneous 12 h6+? would throw away the win.

  12…Kf6 13 Ke4 Kg7 14 Kf3 Kf6 15 Kf4 Kg7

  Now that White has strengthened his position to its utmost, it is time for the decisive advance!

  16 Kg5! c3 17 h6+ Kg8 18 Kf6 c2 19 h7+ Kh8 20 g7+ (or 20 Kf7 c1Q 21 g7+) 20…K×h7 21 Kf7 c1Q 22 g8Q+ Kh6 23 Qg6#.

  Tragicomedies

  Potter – Zukertort

  London m (5) 1875

  The position is in fact the same as in our preceding example, which was published a quartercentury before this game. White, unacquainted with endgame theory, agreed to a draw here.

  The win is elementary:

  1 b5 Ka7 2 b6+ Ka6 3 Kb4 Kb7 4 Kb5! d3 5 a6+ Kb8 6 Kc6 d2 7 a7+ Ka8 8 b7+ K×a7 9 Kc7 d1Q 10 b8Q+ Ka6 11 Qb6#.

  A century later, chessplayers, alas, continue to make the very same mistakes.

  Bouaziz – Pomar

  Siegen ol 1970

  The proper array of the pawns would be g6/h7. So the win is: 1 h6! Kf6 2 h7 Kg7 3 Kg4 (the immediate 3 Kf5 e3 4 Ke6 e2 5 h8Q+ K×h8 6 Kf7 was also possible) 3…Kh8 4 Kf5 e3 5 Kf6 e2 6 g7+ K×h7 7 Kf7.

  White chose 1 g7?? Kf7 2 h6 Kg8. Draw, because 3 Kf5(e5) is met by 3…Kf7!.

  Exercises

  Stalemate

  The Stalemate Refuge

  When there are only a very few pieces left on the board, stalemate becomes one of the most important defensive resources – remember the “king and pawn vs. king” ending, if nothing else.

  Out of the many possible stalemate situations, it is worth noting the following:

  The loss of the c5-pawn appears inevitable; however, Black can still save
himself.

  1…Kb6! 2 Kd5 a6! 3 Kd6 Ka5!, and the pawn is untouchable, because of the stalemate.

  Transposition of moves by 1…a6?? would be a grievous error – White would reply 2 a5!, eliminating the king’s stalemate refuge.

  In the following endgame, we shall see, besides stalemate, other techniques we saw earlier.

  Nikolaevsky – Taimanov

  USSR ch, Tbilisi 1966

  How should this game end?

  1 d4! (breakthrough) 1…Ng6!

  Of course not 1…cd? 2 c5, and a pawn queens.

  2 dc dc 3 B×g6

  3 d6?! Ne5, and now it is White who must work for the draw.

  3…K×g6 4 Kf2

  Here is a position we know: the protected versus the outside passed pawn. White cannot get a zugzwang position – kings at h4 and g6, with Black to move; therefore, he will have to trade his d-pawn for Black’s h-pawn. This exchange would have led to victory, if Black’s pawn were at a5 (instead of a6), or White’s pawn at a4 (instead of a2). As it is, the upshot is stalemate.

  4…Kf6 5 Kg3 Kg5 6 Kh3 Kf5 (6…h4) 7 Kh4 Kg6 8 d6 Kf6 9 K×h5 Ke6 10 Kg5 K×d6 11 Kf5 Kc6 12 Ke5 Kb6 13 Kd5 Ka5! 14 K×c5 stalemate.

  Tragicomedies

  Chigorin – Tarrasch

  Ostende 1905