Take care when exchanging into pure pawn endings!

[Kevin Bonham]

I think this is a general lesson so important especially to many (but by no means all) U1500 players that it deserves a thread of its own even though there is another pure pawn thread at present.

This weekend in the Tas Open I had a couple of difficult games, one against a player rated much lower than me and one against an unrated player. Both had played well, the unrated player even missed a win, and in each case a rook ending was reached. One of the endings was pretty obviously drawn and in the other I was a pawn up with winning chances but there was still plenty of fight left in black's position.

In both cases the opponent completely unnecessarily allowed me to exchange rooks into a pawn ending that was an easy win for me. Both of them did it with heaps of time on the clock, spending not much time analysing the position, and wasted their efforts of hours with a single thoughtless move.

I don't know why people do this so often. I've lost count of the number of games I've won or seen won through the needless exchange of the last piece leaving a pawn ending.

Things to remember here.

* Pawn endings are very unbalanced. Often the exchange of the last piece will change a position from a draw to a win for one side.

* Most likely if you do go into a pure pawn ending you are not going to need that much time to play it out, especially if you've already analysed it. You can usually afford to get down to a few minutes (especially if there's a decent increment) deciding whether or not to exchange. If during that time you decide not to exchange because the pawn ending would be a loss, then yes, you are playing on short of time, but that is better than playing on with lots of time on your clock in a dead lost position.

* Many pure pawn endings can be calculated quickly, leaving even less excuse for exchanging into one without trying to analyse it out. However, others are incredibly complex, especially some of those that end up with both sides promoting. If you're worried that deciding whether or not to go into an unclear pawn ending from a given position will really burn up too much time as you weren't able to calculate the outcome more or less right away, then a simple option is: if in doubt, don't do it. Especially not if the outcome of the game is otherwise looking OK for you (eg if you should draw against a higher-rated player with the material kept on.)

* Just as you should be careful about exchanging off into a pure pawn ending, so you should also be careful about allowing the opponent to force an exchange. Putting your rook behind your king where your opponent can check and get rooks off is a common example.

No point playing a mostly excellent game and then falling at the final hurdle - especially not if you want to get your rating up! Your rating doesn't say how well you played most of the game, all it cares about is the result.

[Igor_Goldenberg]

Happens at every level. Bobby Cheng against Leonid Sandler (Vic champ 2012) allowed an exchange from slightly better/roughly equal rook ending to a lost pawn ending.

[Capablanca-Fan]

A very promising thread, with good thoughts above. All we need now are some concrete examples. To give a contrary case, here is Réti swapping rooks against Alekhine, because he foresaw that the pawn endgame is drawn.

Most of the time, it's not a good idea. Less experienced players don't realize that an otherwise ordinary P endgame with am extra outside passed P is easier to win than an ordinary middlegame a rook ahead. There might also be an effect often seen in simuls: the weaker player loses a little material then just happily swaps off most pieces making a very easy endgame win for the stronger player. Maybe the former thinks he is "holding out longer", and many beginners think that a longer holdout means greater skill, but in reality it's very weak resistance.

[pax]

Most of the time, it's not a good idea. Less experienced players don't realize that an otherwise ordinary P endgame with am extra outside passed P is easier to win than an ordinary middlegame a rook ahead. There might also be an effect often seen in simuls: the weaker player loses a little material then just happily swaps off most pieces making a very easy endgame win for the stronger player. Maybe the former thinks he is "holding out longer", and many beginners think that a longer holdout means greater skill, but in reality it's very weak resistance.

I can recall this very thought process when playing as a junior against a much stronger player. Material was level, and I can remember thinking that each exchange was getting me closer to the draw I was seeking when in fact I was just playing myself into a lost endgame (though not a purely pawn game) that I didn't fully understand. I think the mistake is in thinking of "king vs king" as the objective in seeking a draw rather than simply a protected position where the opponent cannot make progress.

[Igor_Goldenberg]

Many players neglect studying basic endings. Indeed, when it happens on the board they often are able to work it out. However, if you know the basic ending, you can determine:
a) whether you can swap into it.
b) what kind of ending you need to be aiming for.

[Capablanca-Fan]

Many players neglect studying basic endings. Indeed, when it happens on the board they often are able to work it out. However, if you know the basic ending, you can determine:
a) whether you can swap into it.
b) what kind of ending you need to be aiming for.

Indeed. I'm actually mystified. Most of my serious chess was played under the unlamented adjournment system, where people could look up many endgame positions. Of course, what you say is true, that they would need to foresee what the endgame would be. Now that this outside help is impossible, one would think that players would be more motivated to study the endgame techniques more thoroughly. It doesn't seem to have happened though.

[Kevin Bonham]

These are my examples from last weekend:

FEN Viewer

Black played 55...Rc6?? losing instead of 55...Kd7 which is pretty much dead drawn (barring the same sort of blunder later).

FEN Viewer

White was the exchange up but has just given up the exchange for a pawn on d5. Black plays 32...Re5?? losing. (I would expect to probably win this with white anyway against an inexperienced opponent but not against someone who knew what they were doing. Indeed in the latter case I would not have exchanged into this position in the first place.)

[Kevin Bonham]

Cheng - Morris, as just published on Kerry's blog (assuming moves as given are correct), is a rather advanced and difficult example:

PGN Viewer

 

Black's ...Bxd7 is at least risky because it gives white the option of exchanging into the pawn ending, and while the outcome if this occurs immediately should be a draw, the annoying point for black is that neither rook nor bishop can move after the capture, allowing white to improve his king position and then exchange into the pawn ending if he can reach a won position to do so in. Even more annoying, the natural attempt to lift the pin, ...Kc7, runs into Nd5+ and black loses material (at least the g-pawn).

So the option for black that does not raise immediate pawn-ending problems is 1...Bxa2. That's far from simple in its own right but the idea is that if white goes for a7, black will play for K-d6-e7 and Be6 and he is fine.

After 2.Kf2 an idea for black is to try to lift the threat on the g-pawn by playing 2...g5. Now black is threatening to free his bishop with ...Kc7 so if 3.Ke3 Kc7 we get this after the swap-off:

FEN Viewer

...which is probably a draw but some lines are still complicated. To be very confident this was OK you would probably need to have this position in mind and be thinking about it for 10 minutes or so when analysing ...Bxd7.

I believe Black's choice of 2...f4 loses to either 3.h4 as played or 3.Ke2 but the Ke2 g5 line with immediate exchange is tricky. For instance, white might have to anticipate this position:

FEN Viewer

Now the subtle Kd4 Kd6, Kc4! Kc6, h4 appears to be the winning idea.

It happens that in the actual game black escaped unscathed because white miscalculated and took a draw in a position where the outcome of a pawn race is a queen ending that white should win. Presumably the clock was a factor on one side or the other in all this.

So maybe no care is needed in exchanging into pure pawn endings because nobody knows how to play them anyway.

[that Caesar guy]

Cheng - Morris, as just published on Kerry's blog (assuming moves as given are correct), is a rather advanced and difficult example:

PGN Viewer

 

Black's ...Bxd7 is at least risky because it gives white the option of exchanging into the pawn ending, and while the outcome if this occurs immediately should be a draw, the annoying point for black is that neither rook nor bishop can move after the capture, allowing white to improve his king position and then exchange into the pawn ending if he can reach a won position to do so in. Even more annoying, the natural attempt to lift the pin, ...Kc7, runs into Nd5+ and black loses material (at least the g-pawn).

So the option for black that does not raise immediate pawn-ending problems is 1...Bxa2. That's far from simple in its own right but the idea is that if white goes for a7, black will play for K-d6-e7 and Be6 and he is fine.

After 2.Kf2 an idea for black is to try to lift the threat on the g-pawn by playing 2...g5. Now black is threatening to free his bishop with ...Kc7 so if 3.Ke3 Kc7 we get this after the swap-off:

FEN Viewer

...which is probably a draw but some lines are still complicated. To be very confident this was OK you would probably need to have this position in mind and be thinking about it for 10 minutes or so when analysing ...Bxd7.

I believe Black's choice of 2...f4 loses to either 3.h4 as played or 3.Ke2 but the Ke2 g5 line with immediate exchange is tricky. For instance, white might have to anticipate this position:

FEN Viewer

Now the subtle Kd4 Kd6, Kc4! Kc6, h4 appears to be the winning idea.

It happens that in the actual game black escaped unscathed because white miscalculated and took a draw in a position where the outcome of a pawn race is a queen ending that white should win. Presumably the clock was a factor on one side or the other in all this.

So maybe no care is needed in exchanging into pure pawn endings because nobody knows how to play them anyway.

The whole ending (and a lot of the actual game) was filled with hallucinations and rather silly miscalculations, a lot of it based in time trouble. I saw Bxa2 and thought 'Draw but considering the amount of stuff I've missed tonight it's just safer to take on d7 and get an easy draw'. Then I realized I was actually pinned...which is just sad on reflection but it was legitimate horror at the time. As Bobby played the instant (and correct) Kf2 I went into panic mode and with not much time played f4, with a great sense of regret...

Not exactly one for the highlights reel that game. Loads of simple things missed. And when the time scramble came it was miscalculation city hahaha

[Kevin Bonham]

Another example from a game won by Ethan Lim in the Vic U12 today - a rook ending a pawn up where the defender must beware forced liquidation because the pawn ending will be lost. In this case the defender could actually not only have avoided the exchange of rooks, but instead forced pawn exchanges into a drawn position, so Ethan was a bit lucky that didn't happen.

PGN Viewer

 

[Capablanca-Fan]

2... Rd4 or Kf5 would force 3. fxg5, and RPP v RP is almost dead drawn here.

[Kevin Bonham]

Yes. It's also a good example of how leaving a pin on your king can be a bad idea even when there is nothing too obviously wrong with it.

[Jesper Norgaard]

In fact not all what could be learned from this endgame has been commented. First you might even criticize 1...Kg6!? because it goes up into a pin, but since 2.f4 does not really improve white's position and pawn exchange helps the defender, then it could be cunning. However, it led to the disaster in the game. Then 2...Rd5?? is the real howler (not commented above) because it loses on the spot - instead of defending it puts the rook on the only square where the moves 3.fxg5 fxg5 4.h4 are actually winning. Of course 2...Kf5 or 2...Rd4 would be easier, but even 2...Rd2 would be OK because of 3.fxg5 fxg5 4.h4 Kh5! (only move) 5.hxg5 (5.Rxg5+,Kxh4 is also a draw) Kg6! and although black will have to suffer because he can't capture the g5-pawn, it should be a draw.

I would certainly have started out with 1...Kf7 to avoid the bind and run the king towards the center, that should (also) be an easy draw.

About 90% of all pawn endings are winning - a statistical fact. All rook endings are drawn, especially those with pawns on only one flank - an exaggeration, but still telling a truth about rook endings being drawish.

[Kevin Bonham]

Really all of the moves 2...Rd5, 3.h4, 3...Rf5 are "blunders" in that they dramatically change the evaluation of the position with best play, but the first two were not punished while the third one was. As noted it was a junior event; time limit was 40 mins + 30 secs/move.

[Kevin Bonham]

Also posted to HICC thread.

FEN Viewer

White to move. Should white take the bishop?