Monday, February 02, 2015

Grandmaster Preparation: (King and pawn) Endgame Play

Black to play
Wang Yue - Liu Qingnan, Xinghua 2012

I do not want to pretend that I can add anything significant to your understanding of pawn endings ...
My only observation is that a great number of pawn endings are misplayed if they get slightly complicated - even by grandmasters. It is not obvious what the reason for this is, but it could be because pawn endings are primarily about calculation, Most of us find calculation impossible even at the best of times, many grandmasters included.
Jacob Aagaard, Grandmaster Preparation: Endgame Play (Quality Chess, 2014)

Or as Angus reflected after we’d studied the position at the head of today’s blog (which comes from Aagaard’s excellent book), "There’s nothing you can learn that will help you with this. You just have to work it out."

This is rook endgames:

Either side to move: draw

White to play: White wins
Black to play: draw

Either side to move: draw

Move a position one file to the left and a draw becomes a win. Move the king forward one square and it’s back to being a draw again. It all seems rather arbitrary, but if you know why Position One is a draw, you can work out why the evaluation changes for Positions Two and Three without too much trouble. There’s a common thread that runs through them all.

Similarly the Philidor position ...

1 ... Ra6!=

... is linked to the Short-Side Defence ...

1 ... Re1! =

... which in turn can transform into Ending 57 ...

... if you move your rook to the long side prematurely (as Carlsen did against Aronian in 2006, for example). Not to worry, it’s still a draw. Just make sure you maintain checking distance. If you don’t you could end up in Ending 58 ...

... and lose the game.

This is rook endgames. The positions are interrelated, each one building upon what came before. Invest some study time and in return you get a very concrete sense that you are building up both your skill and your knowledge.

Black to play and draw

King and pawn endgames are an entirely different kettle of fish. Take today’s position, for example. Yes, there is one concrete theoretical position

that will help if you already know it, but finding how Black draws from the starting position is much more about themes and concepts than specifics. You already know about the opposition, shouldering off, zugzwang and reciprocal zugzwang, right? It's still going to take a lot of brain work to discover how they apply in this particular case.

Invest some study time and ... next time you get something similar you'll have to work it out all over again anyway.

This is king and pawn endgames.

an ordinary chessplayer said...

Nice one. I saw that 1...Kc5(?) 2.Ke4(?) f6! 3.Ke3 Kd5 4.Kd3 h6? loses to Kh4-h5xh6-g6 and *still* almost fell for 1...Kc5? 2.f6!

So 1...f6! 2.Kf4 Kc6 3.g5 fxg5+ 4.Kxg5 Kd5! (4...Kd7? 5.Kh6 Ke7(e8) 6.Kg7 +-. 4...Kd6? 5.Kf6! h5 6.Kg7 h4 7.f6 +-) 5.Kh6 (5.Kf6 h5 6.Kg7 h4 =) 5...Ke5 6.Kg5!? Kd5! =.

I'm still internally debating the idea that rook endings are more positional (heuristical!?) than pawn endings. Maybe it only seems that way because you have studied rook endings more?

Solving studies is good practice, but there is a big difference between solving a study and playing a game. When solving a study, you *know* there is more than one idea. So if you find just one idea, it is sure to be the *try*, and you know to keep looking. When playing a game, if you find one idea, there is a tendency to relax, even more so because you are tired and want to relax anyway. It helps to be afraid of the opponent. If you can't be afraid of the opponent, at least be afraid of the chessboard.

Anonymous said...

It's a six piece ending, so Shredder's tablebase can be your friend. If you play out the lines, you get wins by a single tempo.

A premise, advocated semi-seriously by some GMs, is that King and pawn endings are best avoided.

RdC

Jonathan B said...

"A premise, advocated semi-seriously by some GMs, is that King and pawn endings are best avoided."

Is it?

Jonathan B said...

... there is a big difference between solving a study and playing a game

For sure. You can’t look it up on a tablebase, for a start.

Anonymous said...

The GM who recommended avoiding King and Pawn endings was Soltis in a book Grandmaster Secrets:Endings.

http://www.amazon.com/Grandmaster-Secrets-Endings-Andy-Soltis/dp/0938650661

http://likesforests.blogspot.co.uk/2007/08/soltis-on-pawn-endings.html (for the exact quote)

RdC

an ordinary chessplayer said...

@RdC - Are you (Shredder) saying I made a mistake in my analysis? I don't much analyse with an engine, even when I have one available. Nor do I typically look up endings in a tablebase. (I have done it once or twice.) It's not interesting.

Soltis is far from dogmatic. An amusing story from him (Chess Life perhaps) was about strong player X studying R+B vs R, and strong player Y asking why? "You will never get that." Later in the same tournament these players were paired and the ending was R+B vs R! Soltis's comment was "score one for preparation". Player Y managed to draw with the B. Soltis's comment was "score one for practicality".

Soltis gave some statistics about endgame frequencies in master play. I get far more pawn endings than the % he gave. (Can't remember the details but I have this in a file somewhere, I will look for it later.)

Anonymous said...

1...f6 2.g5 looks good enough...

Anonymous said...

The tablebase solution is that 1. .. Kc6 is the only drawing move. The idea is still to play .. f6.

With the best defence, the tablebase gives the refutation of 1. .. f6 as

1. .. f6 2. g5 fxg5 3. f6

That's just the standard trick with the defending King outside the "square".

A drawing line is
1. .. Kc6 (only move) 2. Kf4 f6 (now) 3. g5 Kd7 (only move) 4. g6 hxg6 (only) 5. fxg6 Ke8 (only) 6. Kf5 Ke7 and the draw is now clear.

an ordinary chessplayer said...

@Anonymous(1) - Right. Slightly embarrassing on my part.

@Anonymous(2) - It's still not interesting. Somehow I prefer to be proven wrong by a human. Maybe that's just me.

Niall Doran said...

Jeremy Silman (who I know isn't a GM), in a sub-chapter of his "Complete Chess Course" entitled "All King and Pawn Endgames are Confusing!" provides the following advice:

Never enter a pawn endgame lightly!