Transitioning from Middlegame to Endgame

It was Yasser Seirawan who first taught me, in his book “Winning Chess Endgames”, how endgame knowledge helps you play the middlegame. I’ve had multiple students coincidentally ask me the same question recently: how do I do better moving from the middlegame to the endgame? Gukesh and Ding have given us a good example of (sorry Ding) what not to do with their 14th and final game of the World Championship match.

I considered the question, and came up with two aspects. One is, as Yasser described, having a good theoretical knowledge of endgames. If you know about the ‘wrong colour bishop’, for example, you will make sure from well in advance that you do not get stuck in it, unable to win with an extra bishop and pawn. The second is considering what is needed for the player to win, or to hold the draw. That might mean “I need to create a passed pawn on the queenside,” or “I need to get my king in front of their pawn”.

Let’s take a look at a few moments of game 14 to illustrate. The full game PGN is at the bottom.

The first critical moment of the game was Gukesh’s 18. …b5!. Ding thought about it, then went for the sequence 19. cxb5 axb5 20. Nf4!?, leading to a liquidation of pieces. While it was consistent with his match habit of playing for safety, the game showed how such a strategy can backfire, if you are not willing to throw punches of your own. A few moves later, they reached the following position.

This was the moment where our topic really comes into focus. How should White prepare for the endgame? I will not declare myself to be an endgame whiz, but many of the Grandmaster commentators were very surprised that Ding’s 10 minute think did not produce 24. Bd7, forcing a trade of bishops. While the engines will say almost anything is equal, White should be careful to make their life easy, and major piece endgames tend to be a lot less tricky to defend. After 24. …Bxd7 25. Rxd7 (diagram below), you can imagine, for instance, it may be hard for Black to activate their rook from f8 if White can target the pawn on f7 with their queen.

Instead Ding retreated 24. Bf3, and over the next few moves gave up his extra queenside pawn to enter the following endgame.

Theoretically, yes, three pawns vs. two on the same side of the board is a frequent draw, whether it’s rooks, bishops, knights, or some combination thereof left on the board. But Ding was under no obligation to accept this defensive task. In our first diagram after …b5, White had other moves to fight for control of the game.

Now Black has the only realistic winning chances. If they are going to win, they need to do something like create a passed pawn (via piece or pawn trade), infiltrate White’s camp, or inflict some other damage to White’s remaining structure. To hold, White needs to prevent the black forces from encroaching into their end, and can’t allow a weak point for Black to target. 23 moves of slow advancement later, we reached the fateful position where Ding played 55. Rf2??.

Instead 55. Ra4 patiently blocks the fourth rank and leaves Black to show how they can make further progress. What Gukesh suddenly realized on Rf2 is that Ding had given him a chance to do what he needed to win: trade in a way that something changes in his favour. 55. …Rxf2 56. Kxf2 Bd5! forces the bishops off too; with them still on the board, Black simply would not have enough resources left to win, which is what Ding had in mind. But the exchange leaves the kings in such a place that Black has an easy technical win. It is Gukesh’s knowledge of basic king and pawn endgames that allows him to appreciate this, and go for it.

They played out 57. Bxd5 Kxd5 58. Ke3 Ke5, when Ding resigned and Gukesh became World Champion. A fitting final picture of a championship match, if I do say so.

Black’s king either bulls its way in starting eg. 59. Kf2 Kd4, and soon wins the last white pawn. Or else 59. Kd3 attempts to stave off the black king, but the tidy win is 59. …f4! White must take, or …f3 is too strong a passer. 60. gxf4+ Kxf4 61. Ke2 Kg3 Pushing …g3 works too. 62. Kf1 Kh2 and the pawn strides to promotion.

I’d be willing to guess that many chess players around the world will be more reluctant to enter the rook/bishop vs. rook/bishop endgame on the defensive, down-a-pawn side. That’s good! It means they learned something, and will be watching out for it from now on.

Our takeaways: learn what a good endgame looks like, from king-and-pawn endgames upwards. When you have a grasp of that, it enables you to consider during the middlegames which endgames you can lead towards, and even within an endgame which simplifications should or should not be allowed. The scope is different, of course, but it applies all the way from sub-1000 to the World Championship level!