….I Wanna Learn Endgames!
Being a compendium of both Wisdom and Folly gathered by one smitten with Delusions of Improvement
It has been said that one of the main differences between higher ranked players (say above Elo 2000) and the rest of us is in the way they handle endgames. This is an attempt to chronicle one player’s journey down this particular path to chessic enlightenment. I’ll try to point out common errors, as I commit them.
This position occurred in Walker-DeMaestri, Western Open, July 4 1995, on the 45th move. I’d lost half-points in too many rook endgames of late, so I had been very careful in this one, and was pleased to have survived this far intact.
At the moment, the endgame is fairly level. But it’s not level in any permanent way. Instead, the position rests in a dynamic equilibrium. For the moment, White’s pawns are better (he has just traded a White pawn on g5 for a Black one on a4) but Black can wreck them by forcing …e4. Black’s king is far superior to White’s, but there’s little that can be done to prevent White’s approach to the center. And while Black’s rook is slightly more active, he can’t prevent White from finding a more active place for his rook shortly.
White’s main idea is to get his rook posted aggressively, probably on the fifth rank, to support his own pawns and be ready to switch over to attacking Black’s pawns from behind. If he plants his rook on d5 he can accomplish this, and leave his king to deal with the splintered Black pawns, assisted by the d-pawn. From d5, his rook can also fence off the black king either from the support of his own pawns or from the white c-pawn, possibly both. In any case, White can hold the passed c-pawn’s possible advancement over Black’s head as a threat to maintain the balance the game.
Black’s main idea is to keep his king in position to deal with the passed White c-pawn, use his e-pawn to keep White’s d-pawn from rolling, perhaps even as a lever to remove the support from White’s c-pawn. and drive his g-pawn hard toward the finish line, protected by his rook.
While both plans sound easy, they will collide. The likeliest scenario is a trade of the central pawns, leaving neither side able to queen the outside ones, as the kings will be needed to defend against the opponent’s pawn and a rook cannot force a queen by itself. The only winning possibilities lie in who can get his pawn rolling soonest, with enough support to make it happen. It’s hard to see a forced winning line for either side, though the board abounds with losing ones.
45. Ra5+ Kd4?? This particular endgame sin is called looking for too much. In a delicately balanced endgame like this one, the side who tries to win almost inevitably loses. It’s better to hold off the mad rush for glory until you have carefully searched the path completely for landmines. In this case, foreseeing White’s reply was far short of clairvoyance. Black needed to decide whether he wanted to assist his pawns or stop White’s c-pawn, with Kd6 for the former intention and Kc6 for the latter. Instead, Black is trying to attack White’s pawn chain, letting his king get trapped out of position, unable to help either his own pawns or his rook deal with White’s advancing pawn. This move allows White to win.
46. Rd5+ Kc3 47. c5? White sees the possibility of victory before his eyes and loses all his patience. In the endgame, victory is a shy creature, and often runs away from those who make such bold approaches. The reason this move fails is explained in the note to Black’s 48th. 47. Ke3! is called for here. White should eliminate Black’s e-pawn before turning loose his passed pawn on the world. There’s no hurry. Black’s king is bottled up, and there’s no time for Black to get behind his g-pawn and drive it to glory. White’s arrangement of rook and pawns is too strong to break up immediately, before he has used it. White will win the e-pawn easily, now, and Black will have to give up his rook for White’s c-pawn without ever getting his g-pawn rolling, e.g.: 47. Ke3 Rf6 48. Ke4 Rh6 49. Re5 Rh4+ 50. Kd5 Kd3 51. c5 and White’s pawn runs for daylight.
47. … Rf5 48. c6 Rf6? 48. … Rf7? fails in the same way. But on 48. … Rf8 would follow: 49. Re5 Rc8 50. Rc5+ Kd4!, and White cannot hold the c-pawn.
49. Rc5+ Kb4 50. c7? This gets a query not because it loses so much as it throws away a perfectly clear and simple win. I had found 50. c7 and judged it as winning with a cute rook sac, so I didn’t look for a simpler way to win. Here, simply 50. Rc4+ lets the pawn queen without giving up the rook to do it. After Black moves his king, 51. c7 forces Black to give up his rook for the pawn. From that point the win is a simple mop-up operation. If one move wins, look for a simpler way.
50. … Kc5 51. c8/Q+ Rc6 52. Qf8+ Kd4 53. Qb4+ Kd5 54. Ke3?! An inaccuracy, which makes life more difficult for White. I knew I had to bring my king up in order to make progress, but chose the wrong moment to do so. White’s idea at the moment, is to grab off Black’s pawns and then trade queen for rook to get into a won pawn ending. 54. Qe4+ forces Black to cough up a pawn on the spot. 54. … Kc5 55. Qe5+ and White can now shepherd the d-pawn up the board as he chases the Black king, and Black will find it difficult to protect the g-pawn without separating king and rook, which would allow White to get the full benefit of the multiple attack possiblities of the queen. On the other hand, 54. … Kd6 allows 55. Qg6+ and if Black moves his king to the fifth rank, Qf5 will make the e-pawn hard to hold. And Black has some other worries here, as well: 54. … Kd6 55. Qf5 Kd4 56. Qd7+ Kc5 57. Ke3 and now if Black tries to build a bridge to get his king back to the defense of his pawn with 57. … Rd6?? then 58. Qc7+! wins either the pawn (59. … Rc6, 60. Qe5+) the rook (59. … K anywhere but d5) or the game (59. … Kd5??, 60. Qc4++). And if he doesn’t get back to the pawn, White will be able to trade down to a won pawn ending.
54. … Rf6 55. Qe7 In this position we want the king to interfere with the operation of his rook as much as possible. If the rook keeps tripping over the king, it can’t do a good job of defending both pawns. So, 55. Qb5+ is indicated. The idea is to drive the king either to f5, where Qd5 will pin and eventually win the e-pawn, or to e7, when White’s king can approach the e-pawn with impunity.
55. … Re6 56. Qf7 Kd6 57. Ke4 g5 58. Qg7 g4 59. Qg4?! Better was 59. Kf5!. Black is now in zugzwang. Any king move loses the rook. There are also no good moves for the rook, as its only “free square” (d8) loses: 59. … Rd8 60. Qg6+ Ke7 (Kd7 61. Ke5) 61. Qg4 and Black’s pieces are too far from his last pawn to keep it alive. The best of a bad lot is 59. … g3 60. Qg3 and White’s dominating king makes it impossible for Black to get his rook back into the game, e.g., 60. … Re7 (60. Re8 61. Qg6+ Kd7 62. Qe8+ Ke8 63. Ke5 and White wins) 61. Qg6+ followed by Qf6, winning.
59. … Rf6 60. Qg8 Rf4+ 61. Ke3 Rd4 62. Qd8+ Ke6 63. Qa5 Again, White chooses the wrong idea. The right idea here is to come up on the pawn from behind, making Black move away from it. There are two key ideas here for Black: First, he must always be aware of the possibility that White can exchange on d4 and get a winning pawn ending. For that reason, he must be able to play Kd6 any time White threatens to capture the rook on d4. Second, the square e7 is the “magic” square in this position. If White can occupy e7 with his queen, he’ll win the pawn. So Black also needs to remember while shifting his king to prevent White from gaining that key square. White needs to make these two goals mutually exclusive. 63. Qc7 is the beginning. If 63. … Kd6(to stay near the rook) then 64. Qe7 (the “magic” square!) Rf4 65. Qd7+ Kc4 66. Qe6 wins the pawn. Black’s best hope is 63. … Kf6 (63. … Kf5 falls to 64. Qf7+ followed by Qe6). White’s winning pattern after 63. … Kf6 is 64. Qc8! Moving the king to the g-file falls to 65. Qe6, keeping e7 under observation with 64. … Ke7 (Kf7 65. Qf5+) 65. Qf5 Kd6 66. Qf6+ Kd5 67. Qe7 (the vital square is reached!) Rf4 68. Qd7+ Kc5 69. Qe6, and attempting to lose a tempo with 64. … Rf4 is met by 65. Qd7 Rf1 (… Rd4 66. Qd4! trading down into a won pawn ending) 66. Qd6+ and on 66. … Kf5 67. Qf8+ wins the rook while anything else is met by 67. Ke4, not only attacking the pawn, but threatening to corral the king in a mating net if Black isn’t careful.
63. … Kd6 64. Qb5 Ke6 65. Qc6+ Rd6?! The rook leaves its protected fortress on its own. The last few moves were played more on the theory that if you give a B player the opportunity to weaken his position if the path is at all unclear, he probably will. Now White will aim for the e-pawn.
66. Qe8+ Kf5 (Kf6 67. Ke4) 67. Qf7+ Rf6 (anything else meets 68. Ke4) 68. Qd5 The target is now pinned. 68. … Rh6 Nothing looks good here. 68. … Re6 loses to 69.Qd7 Kf6 70. Ke4 69. d4 Rh3+ 70. Kf2 Rh2+ 71.Kg3 Rh5?? (Black had intended trying to save the pawn with 71. … Re2??, but at the last minute noticed 72. Qf3+. This probably unsettled him a little and resulted in this piece of miscalculation. Even, so, there’s very little he can do now, anyway.
72. Qe5+ Kg6 73. Qh5+ Kh5 What follows is another example of limited thinking in endgames. Black hasn’t yet realized that White’s king can go beyond the pawn’s file and still enable it to queen. It’s a simple calculation error, but it’s amazing how many times it occurs in mid-level (B&C) class endgame play. True, if White had to get his king between Black’s king and the pawn, without crossing the queening path, he wouldn’t make it. But all he has to do is control the queening path, and he can do that from the c-file as well as the e-file. The “key squares” for the pawn on d4 are c6, d6, and e6. If White’s king can reach any of those squares before the pawn, he wins. In this case, Black cannot prevent it. 74. Kf4 Kg6 75. Ke5 Kf7 76. Kd6 A key square has been reached! 76. … Kf6 The light has dawned. 77. d5 Resigns