Is this a winning or drawn position?

Hi, this is a game I recently played at a USCF tournament. In the opening and middlegame, I dominated him being a pawn up, having the bishop pair, and having all of his pieces in a tangle while mine were very active. Here's the game:

I think white is better:

• black cannot move his bishop from d8-h4 diagonal because of f6 break
• black's light squares are weak and white king can use them to go for black pawns
• white can create a passer on queenside - with play on both sides the power of bishop pair increases and the knight becomes weaker
• black passing pawn is easy for white to stop
• white has active plan while black can only wait and defend - this is not a psychologically pleasant situation

Is this sufficient for a win? I think so - black seems too much overloaded and sooner or later should fall down

Nice positions . Seems there's still a lot of play left .

White has a lot to go for him : Bishop-pair , centralized King , Q-side pawn majority and advanced fg-pawns. The white squared Bishop is a little tied down to the defence of the f5 pawn.

The only real thing Black has , is a strong pawn-centre. If he is allowed to advance it , White can get into trouble.

In the position after 45. .. a5 Black threatens to get his Knight on d4 from where it attacks the Be6 . If Black is allowed to take it , you're left with two weak pawns . The only move is see to prevent Nd4  is 46.c3 and when Black desides to play .. bxc3 , then 47.bxc3 maintaining control over d4 and getting some counter-weight in the centre . Then you start trying to infiltrate with your King.

Don't forget : the Knight is a horrible defender against flank-pawns. For this reason you don't have to fear the trade of Bishops. If Black doesn't trade Bishops , then you have a dangereous threat with Bh6 .

To conclude : I believe White's position is better and with accuracte play he can turn it into a sure win.

@Dzikus ,

I agree with your comment.Black is overloaded .

How did you get the nice layout with the indents , the dots ?

I agree with 46.c3 but I like 47.Bxc3 better. The bishop is putting pressure on a5 so the knight may find it difficult to move. We also get to keep White's pawns together to make a passed pawn.

Yes , 47.Bxc3 looks stronger. I overlooked that when the Black Knight moves , White will take on a5 with check.

White has to prevent positions with opposite colored Bishops , because this can easily end up in a draw.

In the game I played c3 yes, but after analyzing it now, I'm seeing problems with it. 

In the post editor there is "Insert/Remove Bulleted List" button (looking like that in Word) - at least under Google Chrome.

As for 46.c3, I would not rush with that. This is how I understand this endgame:

If my analysis is correct and I did not overlook black's defensive resources than it is best for black to suffer the 2B vs B+N endgame but I think he should finally collapse.

Why is it easier to win the bishop endgame here than B vs. N (like that after 46.c3)? Here are couple of reasons:

• black pawns are on dark squares which drastically restricts their bishop
• both e8 and g8 are light squares so white pawns can only be stopped by the bishop at 7th rank; if white wins g7 pawn then the separation of white passers is a strength not weakness (unlike the knights, bishops easily stop pawn chains but are hopeless against separated pawns)
• white has queenside majority which enables creating a passer after black pieces are tied up to defend against the e6 pawn
• in bishop endgames, zugzwang motives are somewhat easier to implement (probably because bishops never change the colour of squares they operate on)

Bishop endgames are quite often very beautiful and reach in ideas. I strongly suggest Troicky's studies (he composed lots of bishop and knight endgames).

Good point, Frankiebones - but white has a better continuation, namely 49.Bxa5

A sample play could follow:

Again, black is not forced to take on e6 but I wanted to show again the beauty and tactical sharpness of bishop endgame.

I think there is not much black can do after taking on a5, white has two passers and will simply push them

On move 20, I think castling king side is a big mistake. You should have played the pawn moves b4 and then a4 to open up the queen side, followed by king side castling. You needed to make the most of having the bishop pair. Also, if possible, try to create an opposite color bishops middle game. At least try to open the game up.

Endgames confuse me, so i set up a small engine tournament with this position, playing komodo 3.0 against itself.  This is the result out of 73 games:

White won: 53 games

 Black won: 0 games

  Draws:      20 games

If you let komodo analyze this position, it gives only +0.62 for white, yet wins with white about 2/3 of the time.

EDIT: Added more games-still no wins for black

Could you post the games? I am curious how Komodo decides to continue this endgame.

The result however shows that white has more chances to win than black for a draw - this is how I feel this position

Here is 

typical game from the tournament

Here is another game.  I can post some more if you'd like.

Thanks for the games. Imo, it seems dubious to get into the opposite colour bishops endgame, with correct play black should draw even with two pawns down. Yet another example of suboptimal endgame play from an engine.

I do not like c4 plan, it leaves the blockading b4 pawn on place and makes white's task harder. With c3, the position opens which gives the power to the bishops. Well, this might be the piece of positional knowledge which even beginners have but most engines do not

Stockfish is supposedly much better at endgames than most engines, so i decided to rerun the match with the latest stockfish against itself.  Here are the results out of 18 games:

White wins: 9

Black wins:  0

Draws:        9

Stockfish seems to be better than komodo, but the results still are aren't what i expected.  Here is a game from the tournament:

It's amazing how differently stockfish plays this position compared to komodo- I think stockfish is better, based on the win/draw ratio, but it just shows that as strong as they are, engines still have a long way to go.

What was the time control of this tournamnt? I can see the moves were played at depth 17-20 which is rather low. When I put Stockfish into analyse mode it chooses 1.c3 rather than 1.Ke4 and also gives a score greater than +1 since depth 28 (after 13 seconds).

I agree that Stockfish plays better in endgames than other engines. In some kind of positions it is superior to Houdini 3 (at least when both are given the same amount of time - sometimes it takes H3 much longer to find the lines which SF4 produces quite quickly).

One of the reasons I can think of is that Stockfish descends to much higher depths than other engines which is very important in endgames where horizon effect plays more important role. Higher depths let SF4 evaluate positions more precisely because it can "see" further what is going to happen.

The tournament was at the very fast TC of 10 seconds + 100 milliseconds increment.  I would much rather do a longer TC, but unfortunatly I only have one computer, which is the laptop I use for everything else.  I don't have enough time( or patience ) to get a decent amount of games at long TC.  What I really would like is a desktop computer. Then I could run long TC tournaments 24/7!

And yes, stockfish is a very aggressive pruner, so it gets deep very fast, which makes it good for endgames and deep tactics, but it will sometimes choose strange moves that aren't the greatest, which lowers it's value as an analysis engine.