# Can Black save this Knight vs Bishop endgame with a pawn down?

After playing horribly in the opening, my opponent played a couple of suboptimal moves which suddenly allowed me to exchange some pieces and get the following unclear position.

I thought about giving the h pawn away, giving White two potentially useless h pawns if he captures gxh4. However, I thought that making use of it would require trading my g pawn for the White f pawn, which seems to be impossible.

[White "Villain (2200)"]
[Black "Hero (2000)"]
[fen "2B3k1/5pp1/4pn2/8/3P3p/4P1PP/5P1K/8 b - - 0 1"]

1... Ne4 2.Kg2 hxg3 3.fxg3 *


The position after fxg3 was when I needed to formulate a plan. I thought of several options, all of which are more or less passive defenses, as I think moving my pawns would only weaken them and gain nothing in any position.

First idea: Leave my king on g8, and sacrifice the knight to get a fortress. More precisely, force White to move his pawn to g6 and trade the knight and f7 and e6 pawns for White's other pawns (king, pawn g6, light-squared bishop vs. king g8, pawn g7).

[Event "Draw"]
[fen "6k1/6p1/3K2P1/8/4B3/8/8/8 b - - 0 1"]


However, I didn't see any way to force this: if White can keep any pawn not in g6, he wins if I sacrifice the knight.

Second idea: Leave my king to protect the g pawn. This forces White to exchange the e pawns. Then I can use my king to stop the passed d pawn and my knight to stop the kingside pawns. However, from what I calculated (and analyzed afterwards), it seems it is not sufficient. This is just an example of what I think could happen; I haven't checked every move thoroughly, but the general idea is still the same: Black can block White's d pawn, but White can always force a win due to Black knight's limited mobility and zugzwang.

[White "Villain (2200)"]
[Black "Hero (2000)"]
[fen "2B3k1/5pp1/4p3/8/3Pn3/4P1PP/6K1/8 b - - 0 3"]

3... Nd6 4.Ba6 Kf8 5.Bd3 Kg8 6.Kf3 Kf8 7.e4 Kg8 8.d5 exd5 ( 8...Kf8 9.e5 ) 9.exd5 Kf8 10.Ke3 Kg8 11.Kd4 Kf8 12.Kc5 Ke7 13.Kc6 g6 ( 13...f6 14.Kc7 Ne8+ 15.Kc6 ) 14.Kc7 Ne8+ 15.Kc6 f5 16.g4 fxg4 17.hxg4 g5 18.Bg6 Nf6 19.d6+ Kd8 20.Bf5 Ng8 21.Kd5 Nf6+ 22.Ke6 Nd7 23.Kf7 Ne5+ 24.Kf6 Nc4 25.d7 1-0


During the game I remembered that KBPvKN with the advanced pawn can be a win in some positions. However, I didn't remember when that happens, and I also couldn't see any way to force the exchange of all kingside pawns before allowing White to advance his pawn this far. In post mortem, I was able to blitz a win against the opponent in a position similar to this (without the g pawns), even though he claimed it is an easy draw.

Third idea: Use my king to stop the d pawn, exchange my e pawn to White's d or e pawn and force him to defend the isolated pawn so he can't make progress. This is what I decided to do in the game, making by endgame skills look ridiculous.

[White "Villain (2200)"]
[Black "Hero (2000)"]
[fen "2B3k1/5pp1/4p3/8/3Pn3/4P1PP/6K1/8 b - - 0 3"]

3... Kf8 4.Ba6 Ke7 5.Bd3 Nf6 6.Kf3 Kd6 ( 6...e5 7.dxe5 Nd7 8.Kf4 Ke6 9.Bc4+ ) 7.Bc4 Nh7 8.h4 Nf6 9.g4 Nd7 10.g5 f6 11.Kg4 Nb6 12.Bb3 e5 13.gxf6 gxf6 14.h5 Ke7 15.h6 Kf8 16.Kf5 1-0


Something like 6...e5 was my idea but I couldn't make it work in any position. For example, after 6...Kd6 7.Bc4, I realized the f7 pawn would hang. I saw no other way to make my position any better: any pawn move would weaken them and any king move would give White more space. Only after 13.gxf6 I realized that my plan of moving king to stop the d pawn allowed White's king to break through easily.

Question: Did I miss something? Was one of these plans drawing after all, or is there a resource I didn't see, or is the game simply lost?

I think it is good to notice the possibility of different fortresses, like the one you mention, and of course the wrong coloured bishop for the h-pawn. But to turn reaching such a fortress into a "plan" is way to early in this position.

I think fxg3 was probably ok. At the first glance it looks ugly that now there are white prospects for a passer on the d-line and on the h-line, but your g and f-pawns can actually deal with the white g and h-pawns. Against g4 and h5 you should play f6 and against h4 and g5 (like in the game) you could have played g6. These setups prevent an h-passer.

But maybe 10.f6 was still ok. Then you certainly should have played 11…fg 12.Kg5 Nf8 which prevents the intrusion of the white king. But even here h5-e4-e5-h6 and Kf6 might just win for white.

An engine will be able to come up with better lines, but my feeling is that

• Keeping the white king away.
• Preventing a h-passer.

might have earned you a draw. (Coincidentally you violate all these guidelines with the move e5, after which you are surely completely lost.)

• That seems to be what I wrote as the "second idea" but without 15...f5. Stockfish says +1.68 in the position before 15...f5 after a minute, so perhaps a draw is still possible in that position.
– JiK
Commented Dec 28, 2014 at 0:34
• Well, in that case you wouldn't be able to prevent a passed h-pawn. So if you don't play 15...f5, you probably shouldn't play 13...g6 either. But you actually have the 13…f6 variation there, you just don't explore it deeper. I seems to me that the Nd6 has enough squares not to get into zugzwang. Commented Dec 28, 2014 at 9:53

In the second idea, it looks to me like 8... exd5 is a mistake. I would try 8... e5 instead, accomplishing your idea of putting the pawns on dark squares, followed by f6, Kf8 and Ke7. If the white king comes to g6, you just move back to f8. The knight will always have a move that the bishop doesn't cover. The most challenging idea looks like playing h4 and g5. For example:

[White "Villain (2200)"]
[Black "Hero (2000)"]
[fen "2B3k1/5pp1/4p3/8/3Pn3/4P1PP/6K1/8 b - - 0 3"]

3... Nd6 4.Ba6 Kf8 5.Bd3 Kg8 6.Kf3 Kf8 7.e4 Kg8 8.d5 e5 ( 8...exd5) 9. h4 Kf8 10. g4 f6 11. g5 fxg5 12. hxg5 Kf7


and it doesn't look to me like White can make progress. If White brings the king to support the g-pawn instead advancing the h-pawn, Black could recapture on f6 with the king and that also looks like a fortress.

I haven't checked the moves leading up to 8. d5.

What if you don't play fxg3? It seems that white h-pawn and g-pawn that can create a passer is a big head-ache for black. It does not look like white can capture h-pawn easily, their doubled pawns on the h-file would not look very good.

Maybe in your initial position you could play Kh7, and then to g6. The f7-pawn is safe while the knight is on f6.

• One problem with that might be the plan to play g4-f4-g5 and bring the king to g4 to attack the h-pawn. Commented Dec 28, 2014 at 10:29

In your second plan, you say:

Black can block White's d pawn, but White can always force a win due to Black knight's limited mobility and zugzwang.

That is White's hope for sure, but it should not happen. For instance, in the line you give, 18...Nf6?? is very collaborative: why not 18....Nd6 with a clear draw ? Your knight is not going to be zugzwanged since White cannot control c4, b5, e4, f7, e8, c8 and b7 at the same time. If the white king aims for f5, simply Kf6 will stop him.

[White "Villain (2200)"]
[Black "Hero (2000)"]
[fen "2B3k1/5pp1/4p3/8/3Pn3/4P1PP/6K1/8 b - - 0 3"]

3... Nd6 4.Ba6 Kf8 5.Bd3 Kg8 6.Kf3 Kf8 7.e4 Kg8 8.d5 exd5 ( 8...Kf8 9.e5 ) 9.exd5 Kf8 10.Ke3 Kg8 11.Kd4 Kf8 12.Kc5 Ke7 13.Kc6 g6 ( 13...f6 14.Kc7 Ne8+ 15.Kc6 ) 14.Kc7 Ne8+ 15.Kc6 f5 16.g4 ( 16.h4 Nd6 17.Bc2 Ne8 18.g4 {White gets a h-passer, but the g-pawn gives enough counterplay for Black.} fxg4 19.Bxg6 Nd6 20.Bd3 g3 21.Bf1 Nf5 22.h5 Kf6 {is drawn as well}) fxg4 17.hxg4 g5 18.Bg6 Nd6! ( 18...Nf6? 19.d6+ Kd8 20.Bf5 Ng8 21.Kd5 Nf6+ 22.Ke6 Nd7 23.Kf7 Ne5+ 24.Kf6 Nc4 25.d7 ) 19.Bd3 Ne8 20.Bg6 Nd6 21.Kc5 Nb7 22.Kd4 Kf6 23.Bd3 Nd6


Admittedly, earlier in your line, White could aim for a passed h-pawn instead of exchanging with g4. And 8.d5?! doesn't look very smart either. But the position should be drawn anyway.

Anyway, this "second plan" looks like the most natural: ask white to demonstrate how he plans to makes progress, exchanges pawns and block the passer on a dark square.

I know I am late to the party, and it should be a draw, but none of the answers really hit on the easiest plan, and that was f5, making is impossible for white to create a passed pawn, except for an h-pawn, but the white bishop cannot control the queening square.

The structure is similar in reasoning to the question I posted here.

f5 looks weakening, but the only weakness, e6, is easily defended by the black king, and the white king does not have time to run to d7 to help the bishop attack it, or the black knight will wreak havoc.

 [FEN "2B3k1/5pp1/4pn2/8/3P3p/4P1PP/5P1K/8 b - - 0 1"]

1... Ne4 2. Kg2 hxg3 3. fxg3 Nd6 4. Ba6 (4. Bd7 Kf8 5. Kf3 Ke7 6. Ba4 f5 7. Kf4 Kf6 8. h4 Nf7 9. Bc2 g5+ 10. Kf3 (10. hxg5+ Nxg5=) 10... e5 $1 11. Bb3 (11. d5 e4+ 12. Kg2 Ke5 13. Kf2 gxh4 14. gxh4 f4= {Wrong bishop after the knight takes on d5 soon.}) 11... exd4 12. Bxf7 dxe3$1 13. Bc4 gxh4 14. gxh4 Kg7=) 4... f5 5. Kf3 g5 6. h4 gxh4 7. gxh4 Kf7 8. Kf4 Kf6 9. Be2 Ne4 10. Bf3 Nd6 11. Bc6 Nc4 12. h5 e5+ 13. dxe5+ Nxe5 14. Bd5 Nd3+ 15. Kg3 (15. Kf3 Kg5 16. Bf7 $4 Ne5+$19) 15... Kg5 16. Bf7 f4+ 17. exf4+ Nxf4=


Against a human it could certainly happen. Black should not resign! But against a computer maybe, my intuition say no.

More importantly is how good are white and black at endgames. That could be worth more than a pawn.

I would play on as white and expect to win. But I always expect to win endgames:)

I would play on as black and expect to draw. But I am confident in my endgame play.