How to Play Endgames Brilliantly

View Article

What have I been up to? I read "Improve Your Opening Play" and "Dealing with d4 Deviations" over vacation. Opening study is one of the least rewarding types or study, but now I have a basic idea of what to do when faced with the Trompowsky, Torre, London, or Colle.

This week I've been reviewing pawn endings. I analyzed 632 K+P vs K, 505 K+P vs K+P, and 90 K+P vs K+2P endings as a refresher. Very entertaining. :)

Zugzwang & Steinitz's Rule

Simba writes, Thanks for posting [Karpov-Hort, Budapest 1973] and your commentary which is very informative. Got one question: What does 'spare tempi in zugzwang situations' mean when white moved g5 in move 35?


Zugzwang means it's a disadvantage to have the move.



In the above position, Black to move loses and White to move draws. Such positions are called zugzwang positions and they're very common in the endgame. It's astonishing the first time you see such a position, because in the opening it's an advantage to have the move!



Steinitz's Rule: Keep some pawns on their starting squares, because there they often provide reserve tempi in zugzwang positions.


With a-pawns onboard, the situation's more complicated.

In the first diagram, Black to move draws with 1...a6! a4 a5 and now it's White to move in our original zugzwang position. But he loses with 1...a5?? 2.a4 as now it's Black to move.

In the second diagram, Black to move draws with 1...a5! and now it's White to move in our original zugzwang position. But he loses with 1...a6?? 2.a5 as now it's Black to move.

In the third diagram, Black loses either way. 1...a5 2.a4! and it's Black to move in the original zugzwang position. 1...a6 2.a3! a5 3.a4 and again it's Black to move.



Not every endgame reaches a zugzwang position, but many do, and now you know why it's a good idea to leave some pawns on their starting squares.   :)

King & Pawn vs King

King vs King & Pawn is a simple ending, so why am I covering it? Some fellow bloggers are just learning it. Also, some players rely completely on maxims such as:

"In K vs. K+P endings move the defending King straight back and straight in front of the pawn no matter what. If the opposing King steps up to the left or right of the pawn, the defending King moves right in front of the opposing King."


White to move. Azmaiparashvili-Lechtynsky, Albena 1984. Quick! Can you find a move that allows White to draw this position?


White to move. Salom-Knoblauch, Vitoria 2003. Quick! Can you find White's best attempt at a draw in this position?


White to move. Madl-Titz, Eger 1988. Quick! Is this position won or drawn?

For the record, my favorite rule for this ending is taught by Karsten Mueller. He teaches that if you happen to have any two of the following, you win:

    1. King in front of pawn
    2. The opposition
    3. King on the sixth rank

He also analyzes several instructive positions. Now, we return to today's problems:

1. Azmaiparashvili-Lechtynsky, Albena 1984. 1.Kg1! Kg4 2.Kg2= (1...Kf4 2.Kf2=). All other moves lose because they allow Black's king to get in front of his pawn with the opposition.

2. Salom-Knoblauch, Vitoria 2003. White drew with 61. Kh2! g3+? 62. Kh1 Kg4 63. Kg2 1/2-1/2. Note, 61.Kh2 doesn't guarantee a draw, but it gives White chances to go wrong.

3. Madl-Titz, Eger 1988. It's won. The game continued 62.g6! Ke8 63.Ke6 1-0. The continuation would have been 63...Kf8 64.Kf6 Kg8 65.g7! 1-0.

loomis' pawn ending

 

White to move. Loomis posted this practical, interesting position from one of his endgames. I encourage you to try to solve it, then read his commentary.

Dos Hermanas, 2004

Dos Hermanas is a small city in Andalusia, Spain. It's named after two sisters who, after the conquest of Seville, discovered an image of Saint Anne.

In 2004, Dos Hermanas held a chess tournament. That brings us to our game of interest, between Emre Mecit and Juan Carlos Angel Acosta. We briefly discussed it yesterday, but today I want to master this endgame.



Mecit-Acosta, Dos Hermanas 2004

Which plan should White adopt?
    A. Block the a-pawn with his pawn and the f-pawn with his king
    B. Counterattack by capturing the a-pawn.



White fares very badly if he counterattacks. That's because White promotes in 11 moves, but Black promotes in only 7 moves! However, do check this plan in your games, because it works if the defender's pawn is far advanced.



So White's plan is to simply block the f-pawn by moving back and forth between e3, f3, g3, and f2. If you've studied King & Pawn vs King, you know the attacker won't be able to force the f-pawn's promotion.



Sciurus made the interesting observation that Black's king can wander just about anywhere he likes without changing the result. For example, 1... Kh5 2. Kf4 Kg6 3. Kf3 Kg5 4. Kg3 and Black can still win easily.

That actually makes sense. The White king is already ideally positioned to stop the f-pawn, and counter-attacking is doomed even with an extra tempo.

Which plan should Black adopt?
    A. Promote with the f-pawn.
    B. Use the f-pawn as a decoy and promote the a-pawn.

We already know the first plan fails, so Black must try to promote his a-pawn.


Black can wander around aimlessly as long as he likes, but as soon as he sacrifices his f-pawn, every tempo counts. He should choose his starting position for the upcoming race very carefully.

1...Kf6 2.Kf3 Ke5 3.Ke3 and Black will win the race. Not 2...Ke6? 3.Kf4!, when Black's only positioned to draw the race. This may seem tricky but it's not. Simply remember to shoulder your opponent whenever possible.

Even if you goof, there's no need to worry because the race hasn't started yet. Keep maneuvering your king until you obtain a favorable position.



And the race is on! Each player needs only six moves to reach the key squares (b2, c1), but Black moves first. Simple counting tells you Black will win!


But as you'll recall, Black didn't know Bahr's Rule, and played f4??. It happens to be the only move that draws. Why's that the case??

It's quite simple. After capturing the f-pawn, the defending king needs to reach the c1 square to draw. And f4 is closer to c1 than g4 is.



Hopefully now you have a deep understanding of this endgame, and you will be able to play it well whenever it comes up on the board.   :)

I reject the notion that you need Bahr's rule to play such positions accurately. However, Bahr's rule lets you evaluate and play such positions instantly and without error. And having that in your arsenal's a huge advantage.

Rook Pawns & Outside Pawn II

Yesterday I covered eight Rook Pawns & Outside Pawn endings. You may wish to familiarize yourself with that blog entry before proceeding.


Speelman-Chandler, Hastings 1988. White to move. Quick! Who wins?

The techniques we discussed in our last lesson don't help. White's rook-pawn hasn't crossed the middle of the board, and counting isn't easy because the paths of the two kings cross.


What can we do? We can use a new technique called Bahr's Rule. Draw a diagonal line from the defender's pawn towards his first rank. Then, at the bishop file, bend the line back towards the attacker's first rank.

If the attacker's passed pawn is on or behind the line, he wins. If the attacker's pawn is in front of the line, he draws. So in this case White draws.


Mecit-Acosta, Dos Hermanas 2004. Black to move. Quick! Who wins?

Again, let's draw the Bahr's Rule lines:



The attacker's pawn is on the line, so he can win with accurate play! In the actual game, Acosta played the inaccurate 56...f4+? and drew. The right plan for Black was to capture White's a-pawn and then promote his own.

Bahr's Rule is very useful, but only use it when the attacking king is beside his passed pawn and the defending king is in front of it.



Sutovsky-Gulko, Montreal 2006. White to move. Quick! Who wins?

I told you Bahr's rule only works if the attacking king is beside his passed pawn and the defending king in front of it. Clearly, that's not the case here.



But Bahr's Rule works with a slight modification. If the attacking king is in front of his pawn, and the pawn is not a rook's pawn, pretend that the pawn is one square further back before applying Bahr's rule. This position is actually a win for White.

Rook Pawns & Outside Pawn

In pawn endings, an extra pawn is often enough to convert the win. But rook pawns lead to exceptions, which you must learn to prevail. Today we'll cover 8 positions from real games.

Tip: You may want to read my K&P vs K with a rook pawn lesson before this one.



Viengold-Fieandt, Jyvaskyla 1996. Black to move. Quick! Who wins?


If you've read my posts before, you know I'm a huge fan of counting. Black will capture the f-pawn and reach c8 in five moves. White will capture the a-pawn and reach b7 in five moves. So Black to move draws.



Capablanca-Edison, New York 1931. White to move. Quick! Who wins?


This is very similar to the first example. Black will capture the h-pawn and reach c8 in six moves. While will capture the a-pawn and reach b7 in six moves. So White to move wins.

You can't use the Counting Method if the path of the two kings touch or cross.



Horowitz-Factor, New York 1936. White to move. In this position, the count is irrelevant, because the paths of the two kings touch on move two. Fortunately, there's a rule you can use to instantly evaluate this position.

“If the stronger side's rook-pawn crosses the middle of the board, he wins.”

So it's won. Simple, huh?

Above we have:
    Flacnik-Ippolito, Las Vegas 2006
    Skrabanja-Pahole, Sentjur 2005
    Spraggett-Oratovsky, Maia 2003
    Nanu-Ducsoara, Miercurea Ciuc 2000

They're all won, because White's pawn crossed the middle of the board.



Novak-Lacina, Prague 2006. Black to move. What should Black play?

In the actual game, Lacina played 58...Kb6?, losing. 58...h5! would have drawn, because that prevents the White pawn from crossing the middle of the board. What a simple position, once you actually understand it! Too bad Lacina didn't study his pawn endgames.

Pawn Races III

1. Ljubojevic-Browne, Amsterdam 1972


Black to move. If the pawn race were allowed to continue without any tricks, White would queen first (because of b5+) and the game would be a draw. So Black knows he has to come up with something better if he wants to win.

I tried 1...f5 planning 2. b4 f4 3. b5+ Kc5 4. b6 Kc6 5. Ka6 f3 6. b7 f2 7. b8=Q
f1=Q+ +/- and 2.Kb4 Kd5 3.Kc3 Ke4 4.Kd2 Kf3 5.Ke1 Kg2 +/-.

Unfortunately, I miscalculated. So did the world-class players in the game! White doesn't waste a tempo on 5.Ke1, and so he queens a turn earlier than Black and draws. If I had calculated this line correctly I would probably have found the winning move, 1...Kd5! White's early avoidance of b5+ also prevents penetration by the opposing king.

The winning line is 1... Kd5 2. Kb4 Kd4 3. Ka3 f5 4. Kb2 f4 5. Kc2 Ke3 6. Kd1 Kf2 7. Kc2 Kg2 8. b4 f3 9. b5 f2 10. b6 f1=Q +/-.

2. Najdorf-Vinuenza, Mar De Plata 1941


Black to move. Black is outside the square of the White pawn, but White is inside the square of the Black pawn. Clearly black must support his pawn. Candidate moves are Kg3 and Kh3, so the king can land on g2 in two moves. Kg3 is bad because then the b-pawn's promotion comes with check, so Kh3.

1...Kh3 2.Ke2 Kg2 3.Ke3 Kg3 4.b5 f4+ 5.Ke4 f3 6.b6 f2 7.b7 f1=Q 8.b8=Q =.

3. Battsetseg-Groberman, Seattle 2003


1. b5 g4 2. f4 h4 3. b6 g3 4. hxg3 hxg3 5. b7 g2 6. b8=Q g1=Q

It's a simple matter for White to queen first with a pawn majority. What's less apparent is how to win the apparently winning Q&P vs Q&P endgame.


That's yet another endgame I need to learn.

I first encountered these positions in Flear's "starting out: pawn endgames". It's worth picking up if you happen to find the book on sale.

Pawn Races II

Nimzowitsch-Tarrash, San Sebastian 1911


Black to move. Black has an outside passed pawn (a-pawn). While the White king rushes to the a-file to stop it, Black eliminates White's g-pawn. Now the race is on--it's Black's f-pawn vs. White's h-pawn. Both sides queen and draw.

But Tarrasch saw something much, much better: 1...a5 2.Ke4 f5+! 0-1.

White has no good moves. If 3.Kxf5, the a-pawn queens. If 3.Kd3 f4! and White's kingside pawns aren't allowed to connect and will be easy prey for the Black king. What a brilliant tactic--it's called "pawn dislocation".

Pelling-Miles, Ilford 1974


Black to move. Without any pawn moves, it's a draw. Black moves Ke4, Kf3, Kg2, Kxh2, Kxh3. White moves Ke7, Kxf7, Kxg6, Kxh6, Kg5. (Notice I look at Black's and White's moves independently--this is a fast and accurate way to calculate when the kings are moving independently of one another).

But consider 1...g5. If 2.fxg5 hxg5 and Black can begin racing his e-pawn right away. If 2.f5 e5 and Black will win the race. If White doesn't make a pawn move at all then 2...gxf4 3.gxf4 f5 and Black's f-pawn will promote.

In actual play, I ran across this variation: 1...g5 2.fxg5 hxg5 3.h4?! gxh4 4. gxh4 Black to move. What now?


Continuing the race loses! You must reappraise a pawn ending every time you see an unexpected move. Black wins with accurate play. He's in the square of White's passer so he can capture it. Black's connected passers are untouchable, because if White attacks the back-pawn, the front-pawn promotes.

4...Ke4! 5.h5 Kf5! 6.h6 Kg6 7.h7 Kxh7 -/+.

Grigoriev, 1933


This one comes from my favorite composer. It's obvious the immediate race only draws. In 5 tempos White queens, but in 4 tempos Black's pawn is in the perfect position to draw the resulting Queen vs Pawn ending.

We can use the "Rule of Exclusion". Since b4 draws for certain, and Kf6 (getting into the square of the h-pawn) doesn't lose, play it immediately. One trick to pawn endings is knowing when not to calculate!

1.Kf6! h5 2.Kg5! Kg2


Now Black is one tempo closer to queening, and is threatening to get inside the square of our pawn, so advancing our pawn makes sense.

3.b4! Kg3 4.b5! h4 5.b6! h3 6.b7! h2 7.b8=Q!


Now it's Queen vs Pawn. Against a rook pawn, these positions are usually drawn, so we must calculate very carefully in order to win.

8.Qb2+ Kg3 9.Qc3+ Kg2 10.Qd2+ Kg3 11.Qe3+ kg2


Our Queen has approached. Now it's time for our king to approach, but again we must calculate the next few move very carefully.

12.Kg4 h1=Q 13.Qe2+ Kg1 14.Kg3 Qh4+ 15.Kxh4 +/-

Pawn Races

All positions are White to move unless otherwise noted.

1. Smerdon-Humphrey, Penrith 2003


Counting tells you White queens, then Black queens. If you imagine a White queen on h8 and a Black queen on b1, you'll see Black queens with check, but she can't skewer White's king, so the position's drawn.

But--that evaluation's all wrong! White's king is inside the square of the Black pawn, so instead he captures it and wins. 52.Ke4 Ke2 53.Kd4 Kd2 54.Kc5 1-0.

Lesson: Consider "the square of the pawn" before counting.

2. Leick, 1948


At first this looks like the opposite case. White is outside the square of the Black pawn. Black is inside the square of the White pawn.

However, White can draw by threatening two things at once--his king moves towards both his pawn and the enemy pawn on the same move! 26. Ke7 b5 27. g4 Kf4 28. Kf6 Kxg4 29. Ke5 b4 30. Kd4 b3 31. Kc3 b2 32. Kxb2 1/2-1/2.

Lesson: Consider moves with "dual aims" before counting.

3. Lahno-Shilin, Donetsk 2003


Black is 3 tempos away from queening (e3-e2-e1). White must queen about the same time to have any chance of drawing. 1.d5! e3 2.dxc6 e2 3.d7 e1=Q 4.d8=Q =. Wrong is 1.Kxc5 e3 2.Kc7 e2 3.c6 e1=Q -/+.

Lesson: In a close pawn race, every tempo counts.

4. Vusatiuk-Panchenko, Kiev 2003


It would take White 6 tempi to promote his h-pawn, but Black only 3 tempi to promote his c-pawn. An immediate race is not in White's best interest.

1.bxc4 bxc4. White is still 6 tempi away, but now Black is 5 tempi away. In addition, White has the first move and will queen with check, so he's well-prepared for the resulting pawn race. 2.Kf6 Kc2 3.Kg7 Kxb2 4.Kxh7 c3 5.Kg6 Kb1 6.h7 c2 7.h8=Q c1=Q =.

Lesson: Visualize the result before you enter a pawn race.

5. Shishkin-Kislinksy, Kiev 2003



White is 5 tempi from promoting, has the move, and his queening square happens to check the Black king. Black is 6 tempi from promoting--obviously, a pawn race is not in Black's best interests. But he has a trick up his sleeve!

1.d4 Kb6 2.Ke6 Kc7 3.Ke7 Kc6 4.Ke6 Kc7 =

6. Ljaksa-Domuls, USSR 1974


With Black to move, counting would indicate the position is drawn. Both sides are 5 tempi away from promoting and neither side queens with check.

However, Black can use to his advantage that he's inside the square of White's pawn! 26... h5 27. c4 h4 28. c5 Kf6 29. Kb5 Ke6 30. c6 h3 31. c7 Kd7 32. Kb6 h2 33. Kb7 h1=Q+ 34. Ka6 -/+. To defend his pawn, White was forced to step onto a square that allow the Black pawn to queen with check and win.

Lesson: Watch for king moves that force your opponent to move his king.


7. Van Scheltinga-Cortlever, Beverwijk 1946


This is the first endgame of this bunch that I got wrong. Van Scheltinga and Cortlever played it accurately! No wonder they scored first and second place, respectively, in the 1947 Beverwijk chess tournament.

1... Kc4 2. Ke4 Kb5 (2... h5 3. f5 h4 4. f6 h3 5. Kf3 +/-) 3. Ke5 Kc6 (3... h5
4. Kf5 Kc4 5. Kg5 Kd5 6. Kxh5 Ke6 7. Kg6 +/-) 4. Ke6 h5 (4... Kc7 5. f5 Kd8 6.
Kf6 Ke8 7. Kg7 +/-) 5. f5 h4 6. f6 h3 7. f7 h2 8. f8=Q h1=Q 9. Qa8+ +/-

No single variation is tricky, but every turn White has to calculate anew whether to shoulder out the black king, gobble up the black pawn, or simply race his pawn towards promotion.

Lesson: In pawn endgames, calculate each move carefully. Sometimes there's no option but to calculate all the variations.

8. Colle-Gruenfeld, Carlsbad 1929



This position is deceptively simple, like pawn endings in general. White races to promote his g-pawn and Black races to promote his f-pawn. King maneuvers seem unlikely since the black king is outside the square of White's g-pawn, and Black's king is able to protect his f-pawn.

1.g6 f4+. The only possible escape squares are h3, h2, g2. Kh3 would allow f1=Q+, so it's a bad idea. Kg2 allows g2+, so it also looks bad. 2.Kh2 f3 3.g7 f2 4.g8=Q f1=Q =. There are no skewers or queen exchanges, and with rook pawns the position is clearly drawish.

Then I checked my notes. Colle won this position?! 2.Kh2? was a mistake, and only 2.Kg2! wins. The trick is, after 2.Kg2 Black must play 2...Ke2 to prevent 3.Kf1. The line plays out 2.Kg2! Ke2 3.g7 f3+ 4.Kg3! f2 5.g8=Q f1=Q +/-. White can force a queen exchange and then win the resulting pawn endgame!

Lesson: Always watch for king moves that force your opponent to move his king!! Even if the enemy thwarts them, his king may wind up on a bad square.