# Thread: Take care when exchanging into pure pawn endings!

1. Originally Posted by Rincewind
I don't have an analysis but my impression is that this looks winning for Black. Looking at the pawn position doesn't black a-pawn block whites b-pawn and the king side is held by blacks king and g-pawn so White's king has to do something about the d-pawn and while that is happening Black can win a k-side pawn and break through with the g-pawn. Something like 1...Qd4+ 2.Qxd4 cxd4 3.Kf2 Ke6. Now if 4.Ke2 black can play 4... Kf5 and use d3-d2 to divert the king and allow a breakthrough into win the k-side pawns. If 4.Kf3 d3 5.Ke3 Kf5 with the same idea. White wants to swap off all the pawns but can't force this because needs to be in a position to win both d-pawns when this occurs. Could be totally wrong as it is just a brief impression.
Comment below:
I agree with you until 1...Qd4+ 2.Qxd4 cxd4 3.Kf2 Ke6 4.Ke2. If black plays 4... Kf5 what would be the answer on 5.Kf3? If 5... d3 6.g4+ Ke6 7.Ke3. I must admit I am also not sure I have all answers in the position, because the zugzwang elements are so diverse. Black can insert a5-a4, or switch back between d5 and f5 etc.

2. I'll need to look when I have access to a chess set.

3. I've had a bit more of a look at this and while Black seems close to winning White seems to have just enough to hold on. Rather than following the line I gave earlier Black should I think wait at e6 and play 4... a5 but then if White plays 5.b3 he holds whereas I think 5.g4 (for example) loses. I would not be surprised if there is an unexpected resource to tip the balance as it seems finely poised.

4. Following Barry's line above:

After a5, b3 d3+, Kxd3 Kf5, doesn't white have problems?

5. I don't think so. In that case White's king comes to e4 (I don't think it matters if via d4 or e3) and if Black takes g3 then White plays f5. If Black instead tries d5+ then Kxd5 Kxg3 Kd5 and all is well.

Edit: Scratch this. I loaded this with an engine and actually White *is* busted after d5+ Kd5 Kxg3 line. White follows up with Kf3 and after say f5 gxf5 Kxf5 Ke3 wins easily.

6. Originally Posted by Kevin Bonham
Following Barry's line above:

After a5, b3 d3+, Kxd3 Kf5, doesn't white have problems?
Let's follow that up - since the finer point of this particular position deserves a look. After 4...a5 5.b3 d3+ 6.Kxd3 Kf5 7.Ke3 (threatening Kf3 and g4+)
7...Kg4 8.Ke4 now 8...Kxg3 doesn't win because after 9.f5! d5+ 10.Ke5! gxf5 11.Kxf5 Black is in time to run over and capture the d-pawn and a-pawn. Then 8...d5+! is correct but after 9.Ke5! White needs to find a study-like move 9...Kf3!! and zugzwang. 10.Kxd5 Kxg3 11.Ke4 Kf2! doesn't work, and likewise 10.f5 gxf5 11.Kxf5 d4! wins (not 11...Kxg3? 12.Ke5 which transposes to the earlier mentioned drawing lines). Also it is worth mentioning that after 9.Ke5! d4? doesn't win because of 10.Kxd4! Kxg3 11.Ke3! winning the opposition. I think this already shows why I would be very hesitant to enter such a pawn endgame in a normal OTB game even with half an hour on the clock, because there are so many fine points where a slight concrete error can be fatal, when only looking at the position with a holistic view. However, it does seem to be winning in all lines.

That leaves another important branch that you mentioned. If 4...a5 5.g4 how does Black win this?

7. Originally Posted by Jesper Norgaard
That leaves another important branch that you mentioned. If 4...a5 5.g4 how does Black win this?
I looked at this yesterday and from memory I thought it was quite easy with Kd5 Kd3 then a4 with White in zugzwang. which is why I considered after a5 White should play b3 and not g4.

8. Originally Posted by Rincewind
I looked at this yesterday and from memory I thought it was quite easy with Kd5 Kd3 then a4 with White in zugzwang. which is why I considered after a5 White should play b3 and not g4.

True, it is not that difficult. After 4...a5 5.g4 Kd5 6.Kd3 a4 7.f5 gxf5 8.gxf5 Ke5 9.f6 Kxf6 10.Kxd4 Ke6 11.Kc4 (if White stays with the d-pawn, black will eventually give up the d-pawn for the b-pawn) 11...Ke5 (a move you have to be careful with since it just advanced to the place where b8=Q+ will be with check, but here it is right) 12.Kb4 d5 13.Kxa4 Ke4! (The point and the only move that wins - Black is ready to meet Kb3 with Kd3 and b4 with d4 and support the pawn for instance 14.b4 d4 15.b5 d3 16.Kb3 Ke3! and the d-pawn will always promote with check. If 16.b6 d2 17.b7 d1=Q+ 18.Kb5 Qd6 Black is busted) 14.Kb3 Kd3 15.Kb4 d4 16.Kc5 Ke3 17.b4 d3 18.b5 d2 19.b6 d1=Q 20.b7 Qd8 -+

Another branch we can quickly reject is 4...d3+ 5.Kxd3 Kf5 6.b4! and because of the pawn constellation with black a6 against white b4, this is now a draw. For instance 6...Kg4 7.Ke4 d5+ 8.Kxd5! Kxg3 9.Ke5 Kf3 10.f5=

9. Black wins, but it's awfully complicated. Below is the analysis checked with the engine:

1... Qd4+ 2. Qxd4 cxd4 3. Kf2 a5! (the only winning move)
a) 4. Ke2 Ke6 5. Kd3 Kf5 6. Kxd4 Kg4 7. Ke4 a4!
b) 4. b3 Ke6 5. Ke2 (5. g4 Kd5 6. Kf3 d3 7. Ke3 d2 8. Kxd2 Ke4-+) d3+ 6. Kxd3 Kf5 7. Kd4 Kg4 8. Ke4 d5+ 9. Ke5 Kf3!! 10. Kxd5 Kxg3 11. Ke4 Kf2
c) 4. g4 Ke6 5. Ke2 Kd5 6. Kd3 a4 7. f5 gxf5 8. gxf5 Ke5 9. f6 Kxf6 10. Kxd4 Ke6 11. Kc4 d5+ 12. Kb4 Ke5 13. Kxa4 Ke4!! (as Jesper suggested)

Before swapping into pawn ending black has to find 7...a4, 9...Kf3!! (9 moves ahead) and 13...Ke4! (13 moves ahead) in each line. I would definitely not find it OTB (despite being relatively good with pawn endings) so far in advance. Even analysing at home I had to check every line with computer to make sure it works.

Practical solution is to play very promising queen endgame

10. Originally Posted by Igor_Goldenberg
Black wins, but it's awfully complicated. Below is the analysis checked with the engine:

1... Qd4+ 2. Qxd4 cxd4 3. Kf2 a5! (the only winning move)
a) 4. Ke2 Ke6 5. Kd3 Kf5 6. Kxd4 Kg4 7. Ke4 a4!
b) 4. b3 Ke6 5. Ke2 (5. g4 Kd5 6. Kf3 d3 7. Ke3 d2 8. Kxd2 Ke4-+) d3+ 6. Kxd3 Kf5 7. Kd4 Kg4 8. Ke4 d5+ 9. Ke5 Kf3!! 10. Kxd5 Kxg3 11. Ke4 Kf2
c) 4. g4 Ke6 5. Ke2 Kd5 6. Kd3 a4 7. f5 gxf5 8. gxf5 Ke5 9. f6 Kxf6 10. Kxd4 Ke6 11. Kc4 d5+ 12. Kb4 Ke5 13. Kxa4 Ke4!! (as Jesper suggested)

Before swapping into pawn ending black has to find 7...a4, 9...Kf3!! (9 moves ahead) and 13...Ke4! (13 moves ahead) in each line. I would definitely not find it OTB (despite being relatively good with pawn endings) so far in advance. Even analysing at home I had to check every line with computer to make sure it works.

Practical solution is to play very promising queen endgame

I am not quite sure I agree that the queen endgame is so promising. In the actual game I goofed up the sequence of moves going into a queen ending, that was even more drawish. In the analyzed position it is still not easy to coordinate the attempt to avoid perpetual check if looking for another move than 1...Qd4+

Otherwise to get a full understanding of the pawn endgame, you probably could do it fairly comfortable if there were one hour on the clock, but if only half an hour, it would be really tough.

Actually I did not catch Rincewind's strong move 4...a5 but I don't think it is the only one that wins.

1... Qd4+ 2. Qxd4 cxd4 3. Kf2 Ke6 4.Ke2 Kd5!? was my first idea, and it works quite similarly (and with a lot of transpositions) to the already analyzed lines after 5.Kd3 a5 6.b3 Ke6 7.g4 Kd5 (curiously Black has just stepped twice to e6 and twice to d5 with the king, but is still winning): 8.f5 gxf5 9.gxf5 Ke5 10.f6 Kxf6 11.Kxd4 Ke6 12.Kc4 Ke5 13.Kb5 d5 14.Kxa5 d4 15.b4 d3 16.b5 d2 17.b6 d1=Q 18.b7 Qd6 and Black is still winning.

I think it is a very rewarding pawn endgame to analyze because of the many transpositional lines that can happen. The whole idea with 9...Kf3!! for instance only works because the pawns are fixed on a5 vs. b3 where with a6 vs. b2 to react the same way would only be a draw because there is no way to combine the defense of a6 with attack on b2, the final moves in the former line (a5 vs. b3) being 1...Kc3 2. Kc5 Kxb3 3. Kb5 a4 etc.

11. Great example position.

It seems then that ...a5 wins on any of three moves:

1...Qd4+ 2.Qxd4 cxd4 3.Kf2 and now:

(i) 3...a5 wins
(ii) 3...Ke6 4.Ke2 a5 wins
(iii) 3...Ke6 4.Ke2 Kd5 5.Kd3 a5 wins

12. ## another example of swapping into pawn endgame

 PGN Viewer

13. Originally Posted by Igor_Goldenberg
 PGN Viewer

Nice example endgame.

It is noteworthy how the element of queening rook pawn against opposite side rook pawn can not only win a tempo, but prevent promotion alltogether, with an easy win.
Hence 43.f5!+- would have been decisive.

An option to exchange queens from a queen ending (or generally the last piece in any endgame) must be considered as a deadly threat, just as much as winning a piece or a pawn, because 90% of pawn endgames win, a percentage way higher than for any other endgame. So after 35.Kf3 Black should consider 36.Qe4 a deadly threat.

(1) The only easily winning queen endgame is one where there are decisive threats to exchange queens, or one side gets an extra queen.

Which sometimes leads to a new queen endgame after mutual promoting, leading recursively back to (1).

"With easy win in queen endgame that took another 30 moves and mutual blunder" - QED

14. If black didn't play 32...e6 swapping queens wouldn't be a threat.
As for "easy win in queen endgame" - the fear of queen endgame is overrated. The one in the game was indeed an easy win. There are many checks and you have to calculate variations, but checks will come to an end sooner or later. It took almost 30 moves, but the result was never in doubt (even after I blundered a pawn and black reciprocated by resigning).
I'd say one of the reason I missed 43.f5 is that I had no doubt queen ending is won and didn't think hard - which is inexcusable.

15. There needs to also be a thread called "Take care when not exchanging into pure pawn endings!"

 FEN Viewer

In this position in a rapid tonight white (Markovitz) played 1.Rb3. Rather than spend a while calculating the pure pawn ending (which if anything is good for black and certainly not lost) I instead more or less instinctively played the fabulous 1...Rc7