King and Pawn Endgame Notes

Ken Whaley


Playing and studying endgames is one of my very favorite parts of chess.   Contrasted against middlegame play, where the player often must weigh the advantages and disadvantages of various plans that may not lead clearly to victory, in endgames there often is only one way to achieve the win or to hold the draw.   This "there's just one right answer" quality of endgames is especially keen in king and pawn endings, where a single ill-timed pawn push can throw away a win in a hard-fought game.

King and pawn endings appear deceptively simple -- all the major and minor pieces are off the board, and each side is just trying to promote a pawn or stop the opponent from doing the same.   Indeed, NM Dan Heisman says "king and pawn endings are the easiest type of endgame to win."    True enough, if you happen already to know and to understand the concepts of:
 and are faimilar with scenarios with colorful names like "the Trebuchet," "the Squeeze," and "the Underpass."  

While there's definitely a lot to know and to learn about king and pawn endings, much of you need to know to play king and pawn endgames really is simple.  The annotations that follow are part of my chess homework suggested to me by my teacher, NM Dan Heisman, to highlight these ideas in the context of a game I played..   He suggested this while we were reviewing a game of mine from a recent weekend slow chess tournament on ICC (http://www.chessclub.com/)   The time control for the game was 45+5 (45 minutes for the game plus 5 seconds additional for every move you make), which is "slow" for ICC) hosted by the Slow Time Control Bunch (http://www.stcbunch.net/). After the dust of middlegame (where both I and my opponent made mistakes) had settled, I was a pawn up in a king and pawn ending.    In a previous lesson, Mr. Heisman had shown me how early in his chess career he had played many "boring" (his term!) games where he got a pawn ahead, traded off the pieces, and won the resulting king and pawn ending.  Hah!  I would love to be on the winning side of many  such "boring" games!

Well, here I was, I had my "up a pawn" position, and... I blew it.   I played the incorrect move not just once, but on three separate occasions in a key position where I had the opoprtunity to win.   As a result, the game ended in a draw.   To add insult to injury, I was way ahead of my opponent on time as well: about 12 minutes to 4 minutes, at the time when the last major piece came off the board.

Below is the full game for your amusement.   I played Black.   Below, I will highlight important king and pawn endgame concepts.   There are probably many lines I've missed, but I'm going to concentrate on those moves that are the most closely related to the moves made in the actual game.   Moves from the game will be in bold and  variations in normal type.  I've used Fritz 8 to help me evaluate side variations.  Something I've noticed is that you must be careful when using computers to evaluate endgame positions, especially king and pawn endings, because unless the computer can "calculate" a forced sequence of moves all the way to promoting a pawn, they often play very strange, questionable, or even wrong moves because they usually don't operate on the general guidelines and principles and techniques that human players can use.   Searching 12 or 13 (or even 20!) ply ahead in king and pawn endgames is often not sufficient to find the winning move (of course endgame tablebases play 100% perfectly, but those are still limited to only a few pieces (4 or 5) left on the board.)

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. d3 d6 5. Nc3 Na5 6. Bb3 Nxb3 7. axb3 Be7 8.
h3 Be6 9. d4 exd4 10. Nxd4 O-O 11. O-O d5 12. e5 Ne4 13. Nxe4 dxe4 14. Be3 Qd5
15. Nxe6 Qxe6 16. f4 exf3 17. Qxf3 Qxe5 18. c3 Bc5 19. Rae1 Bxe3+ 20. Rxe3 Qc5
21. Kh2 Qc6 22. Qxc6 bxc6 23. Re7 Rac8 24. Rd1 Rfe8 25. Red7 Kf8 26. R1d2 Re7
27. Rd8+ Rxd8 28. Rxd8+ Re8 29. Rd7 Rc8 30. b4 Ke8 31. Rd3 Rd8 32. Re3+ Kd7 33.
Rd3+ Ke7 34. Rxd8 Kxd8 35. Kg3 Ke7 36. Kf4 Ke6 37. g4 f6 38. h4 g6 39. c4 h6
40. b3 g5+ 41. hxg5 hxg5+ 42. Ke4 f5+ 43. gxf5+ Kf6 44. b5 cxb5 45. cxb5 g4 46.
Kf4 g3 47. Kxg3 Kxf5 48. Kf3 Ke5 49. Ke3 Kd5 50. Kd3 Kc5 51. Kc3 Kxb5 52. b4 c6
53. Kb3 a5 54. bxa5 Kxa5 55. Kc4 Kb6 56. Kb4 c5+ 57. Kc4 Kc6 {
Game drawn by mutual agreement} 1/2-1/2

Here we are at the start of the king and pawn endgame, after 34...Kxd8.   We both have 3 messed up pawns on the queenside (mine are worse since they're all isolated while White's are connected), but isolated and doubled pawns are perfectly good at stopping enemy pawns on adjacent files from promoting and at hindering the Black king from penetrating, and I have a healthy  3 versus 2 advantage on the kingside.   But, how to proceed?
after 34...Kxd8
after 34...Kxd8

In almost every type of king and pawn endgame it's essential to activate and centralize your king, in order to escort your pawns to promotion and to stop your opponent from doing the same.   White's king is on the 2nd rank but in the corner, while Black's king is on its first rank but on a central file.  They're both 3 moves away from the center but it's White's move, so his king can get one rank farther up the board than Black's king can before they meet.   We both start out by centralizing our kings: 35.Kg3 Ke7  36.Kf4 Ke6.

after 36...Ke6
after 36...Ke6

37.g4.    

There's not much for White to do here except push pawns, shuffle his king back and forth, try to stop Black's pawn advances, and hope for Black to make a mistake.  White's king can't penetrate Black's crippled queenside pawns without wasting too much time, he'd have to take a trip like Ke4-d4-c4-b3-a4-a5-a6-Kxa7-b7-Kxc7-Kxc6 before his queenside pawns could promote b5-b6-b7-b8 a sequence of 15 moves.  But Once White's king hits d4 Black plays Kf5-f4-g3-Kxg2 then pushes the f-pawn f5-f4-f3-f2-f1, and gets a queen 6 moves before White does.  When you're in a queening race, it's a very helpful technique to count the moves needed to queen a pawn for both sides and then compare the counts.    If you can queen much faster then your opponent, feel free to enter the race, and when your opponent can queen in fewer steps than you, avoid the race if you have another way of promoting.  Comparing queening counts  is much easier and far less error prone than trying to calculate a sequence of moves for both sides ("I go here, he goes there, I go here, he goes there").  There are some slightly shorter sequences for White that include b5 but they still take about 12-13 moves to queen versus white's 9.   And Black doesn't have to let White penetrate the queenside without a fight -- Black's king can switch over to the queenside to hold down the fort while advancing his 3 vs. 2 kingside pawn majority (you have a pawn majority is when you have more pawns than your opponent across a set of consecutive files) to promote a pawn on the kingside.

I briefly entertained 37...Kd5-c4-b3-xb2 to gobble up White's queenside pawn chain to draw White's king away from the kingside, but noticed that White shuts out the Black king with 38.b3.   I reazlied that I didn't know what to do here to win, even though I thought for sure that I had a won position.  

 I.continued 37...f6.  

 It turns out this is the correct move!  The principle involved here is that you want to lead with the pawn that can become a passed pawn, which is usually the pawn in your majority that has no enemy pawn on the same file (such an unopposed pawn is called a "candidate").   Another very good principle is to avoid creating backward pawns.    A backward pawn is a pawn that cannot advance without losing material.    Backward pawns can be bad because king and pawn endgames almost always include the notion of Zugzwang, or the state of having to move when you don't want to because it worsens your position.   When you have backward pawns and your king is in zugzwang elsewhere on the board, you can't free yourself by pushing the backward pawn to waste a tempo because you end up losing material (the pawn you moved, usually).

38.h4 g6
after 38...g6
after 38...g6

38...g6 is also the correct move, even though the g pawn is officially "backwards."  In this case it doesn't matter because the g-pawn is not going to advance.  Instead, pawn on g6 wiill support the subsequent advance of the f-pawn f5, in order to create a passed pawn.  38...h6? is terrible because the g-pawn becomes badly backwards after 39.h5!  where Black's pawns are stopped cold and his advantage neutralized..  38...f5? is bad of course because it just throws away the f-pawn for nothing.

39.c4

White is starting to use up his safe reserve tempos, or pawn moves that waste time to force Black to move again.   When White runs out of safe reserve tempos his pawn moves will lose more material, and once Black creates a passed pawn White will not want to retreat his king, allowing the Black pawn to advance towards promotion. Black's 39th move is the first key position and opportunity that I missed in the endgame.

39...h6

39...h6 doesn't toss away the win but it's not to the point, which is to advance the outside f-pawn to create a passed pawn.   I think I was concerned about the g-pawn being backward, but didn't realize that this didn't matter because, again, the g-pawn's job here is to suppport the advance of the f-pawn.   The principle of advancing the candidate pawn still applies here.  Fritz agrees: he rates 39...f5 is the best move at 18 ply.  As Dan said in my lesson, "after 39...f5, what's White going to do?"  39...f5  40.gxf5 gxf5 and Black's created a passed pawn.   After this all of White's pawn moves go nowhere, and once he makes them all he must end up retreating his king or trying to enter a queening race that he'll lose via Kg5-h6-xh7-g7 followed by h5-h6-h7-h8Q.xg2- Ke5 where White queens in  8 moves and Black queens in 5 moves Ke5, f4-f3-f2-f1Q.   Also no good is 40.g5, which just gives the passed pawn to Black on a silver platter, with a bonus: Black's f5 pawn then becomes a protected passed pawn, (protected by the g6 pawn).  With the f5 passed pawn protected, Black's king is freed to wander around the board wreaking havoc on White's remaining pawns. 40.h5 is of no concern at all because  40...gxh5 41.gxf5+ Kf6 creates two passed pawns on the h file for Black, tying the Black king down to the h file, allowing the Black king to mop up White's queenside pawns and promote on the queenside. And after 40.h5 gxh5 41.gxh5 h6 White's king must give way after he's exhausted his pawn moves and Black achieves 2 passed pawns on the kingside (e.g., 42.b3 Kf6  43.b5 c5  44.Kf3 Kg5  45.Kg3 Kxh5).   But White can't be careless, 40.h5 Kd6 draws because then 41.hxg6 hxg6 42.gxf5 gxf5 43.Kxf5 and good-bye extra pawn.

40.b3 g5+

Still missing f5, the advance of the outside pawn, but still not tossing away the win on the kingside yet .

after 40...g5+
after 40...g5+
41.hxg5 hxg5+ 42.Ke4.  I saw the opportunity to create a passed pawn via a neat little Zugzwang here, and did:  42...f5+ 43.gxf5+ Kf6 44.b5

after 44.b5
after 44.b5
When White next moves his king his f-pawn must fall.   I thought it very unsporting of him to keep pushing his other pawns simply in order to delay my certain victory (!).  However, when given a choice between making a move that loses right away and one which loses a little bit later, choose the move that loses later, your opponent may make a mistake, especially when low on time.  At this point I had about 3:30 left on my clock to my opponent's 1:30.    I spent one third of my remaining time (about 1 minute)  thinking and came up with the following "gem," for this second critical position of this endgame:

44...cxb5?

I honestly don't recall what was going through my head.  I think I was worried about White's king penetrating after  44...c5 45. Kd5, but a quick queening count after 44...c5 45.Kd5 shows 8 moves for White and 5 moves for Black with Black to move in order to queen.  That counting should have taken about 10 seconds to prove that 44...c5 was safe.   I think I was impatient and just wanted White to move his king so I could win his f-pawn.   But still, I didn't move quickly, I did spend almost one minute of my remaining 3 minutes thinking before playing  44...cxb5.  In my defense I'd been playing about 3 consecutive hours of serious brain-burning chess at this point, which goes to show that mental endurance is definitely a requirement for good chess playing.   One must strive to play  the best moves on each move of the game!   However as Dan pointed out, I could have taken almost all 3 minutes and then won fairly easily after 44...c5 since it was just pushing pawns, and I'd be receiving an extra 5 seconds for every pawn push to create a comfortable time cusion to win after I queened.

After  44...c5, all of White's moves lose material: 45. K(anywhere) Kxf5, 45.b5 cxb6 and Black's pawns form a  connected chain in a 3 vs. 2 scenario.  45.b4 cxb4 and Black has passed pawns on both sides of the board and White is totally, easily lost.

45.cxb5! and now Black's in Zugwang!  

after 45.cxb5!
after 45.cxb5!

 Black's a- and c-pawns are dangerously backwards: moving either one allows White to capture and queen quickly.  Any move of Black's king results in the loss of his g-pawn, e.g., 45...Kf7 46.Kf3 Kf6 47.Kg4 Kf7 48.Kxg5 and White has the passed pawn and the win!   The only hope is to push the g-pawn, which forces the White king to chase and capture on g3, which lets Black mop up white's dangerous f-pawn and his b5 pawn as well::

45...g4 46.Kf4 g3 47. Kxg3 Kxf5 and now the kings race to the queenside  48. Kf3 Ke5 49. Ke3 Kd5 50. Kd3 Kc5 51. Kc3 Kxb5
after 51...Kxb5
after 51...Kxb5
52.b4

A good move, moving the White king results in a reasonably easy win for Black, e.g.,  52.Kb2 Kb4 53.Kc2 a5 54.Kb2 a4 55.bxa4 Kxa4 56.Kc3 Kb5 57.Kb3 Kc5 58.Kc3 c6 and we're in a standard "every Russian schoolboy knows how to play this" winning king and one pawn versus king position where Black has the opposition with this king in front of his pawn.   I saw that exchanging either Black pawn right away draws:  52...a5 53.bxa5 Kxa5 54.Kc4 Kb6 55.Kb4 with the opposition for White and a standard drawn position.  52...c5 53.bxc5 Kxc5 and Black's a pawn up but unfortunatley it's a rook pawn and rook pawns are a draw in king and one pawn versus king endings if the defending king  can get to the queening square ahead of the attacking king (because the attacker cannot dislodge the defender from the coner).

At this point I let my clock get down to 3 seconds before playing out for a draw:  Believe it or not, however, after 52.b4 the position is still won for Black, and so I'm calling Black's 52nd move the third  critical position in the endgame.    It's a fairly tricky one, however, and I just didn't have the time to think it out.

52.. c6  53. Kb3 a5  54. bxa5 Kxa5  55. Kc4 Kb6  56. Kb4 c5+  57. Kc4 Kc6 {Game drawn by mutual agreement}

I was so upset with myself that I didn't play out the ending to stalemate.   In hindsight I violated the principle of give your opponent the opportunity to make mistakes.  Just because I knew how to draw the positon as White didn't mean my opponent knew!   And in time pressure he may have made a mistake.

Now back to after 52.b4 and the third critical position in the endgame.:

after 52.b4
after 52.b4
The correct move to win is 52...Ka4!   In all fairness to myself, I  got down to 3 seconds on my clock, and winning takes a pretty tricky maneuver where black actually gets behind the White king!   53.Kc4 anything else loses the pawn.  53...a6 forces the White king to retreat (if 54.Kc5 Kb3 and white must move the pawn or lose it outright next turn: (55.Kc6 Kxb4 56.Kxc7 a5 and Black wins easily) and if 55.b5 then 55...a5 56.Kc6 Kb4 and Black picks up White's b-pawn and queens his a-pawn ). Continuing the winning line: 54.Kc3 Ka3 55.Kc4 Kb2 behind the White king!  56.b5 a5 and Black again wins the queening race by one move because White must spend time clearing away Black's c-pawn.

Well, that's it!   I hope you've enjoyed this article, and found some useful technques and guidelines that you can use in your own games.   Good Chess to you!

Ken Whaley
Cupertino, CA, USA
October 2006
Copyright (c) 2006 Ken Whaley