Could I have won this endgame?

Black is not better, but he has equalized. But i'd like to point out that 3...g6 is much worse position than if I could play 23...Bxe5. 

lol

Dont be too hard on him, he is not Fischer.

I had calculated Bxh2 and saw my bishop is trapped after g3.

Actually, Fischer had fallen for Bxh2 before losing the bishop for 2 pawns (and eventually losing the game).

(vs Spassky WCC game 2 if I remember correctly)

Now that I think about it, I would get 2 pawns and an entry for the king which might be worth it... though it would likely take a lot of analysis/calculation, but as I was playing for a win it would likely be something worth calculating.

First off game 1. Second after Bxh2 it was still a draw. (most masters know this by now). 

Also in your game I checked the computer and it says that you're much worse after Bxh2 (figures). and Bxe5 is +1.2 which is a winning advantage. 

Why would the knight vs bishop endgame give white a winning advantage? It seems to me like black has a lot of winning chances there.

I saw 26...a5 mentioned. Isn't 27.a4 a simple response to that?  If you want to try to roll the Queen side pawns, you need to play 26...b5 first intending 27...a5.  But then you have to consider both 27.a4 and 27.b4.  Not saying they work or don't work, but both must be considered.

Of course, the one person that made the comment that becasuse there are no heavy pieces that it must be as draw and good vs bad or active vs inactive bishops are meaningless without heavy pieces clearly is clueless about chess.  Minor piece ending often result in superiority one side or the other.  If I had to choose a side here, it would be Black, but you can't blitz this position to victory.  Often a critical factor is king placement in minor piece endings.

+1.2 for black. which is winning (+or-) for the person to play.

It appears that the knight vs bishop endgame would have been great for black.

I saw that during the game, but after it was too late.

I suppose the lesson is to turn off the autopilot, especially at critical moments like exchanging.

Specially on a 45/45 game.

Sit on your hands lol.

I had Stockfish 8 play the position analyzing about 17 moves deep and Black's edge remained too small to win at about a 0.3 to 0.4 pawn advantage.  Usually you need a 2.0 or more advantage in evaluation score to be sure of a win.  There is simply no way for Black to pick off one of White's pawns, create a passed pawn or penetrate with the king.

GET AWAY FROM ME. THIS WAS SUPPOSED TO BE NATURAL KIBITZING.

ThrillerFan is right. The idea of playing on the queenside is correct but a5 is refuted by a4.

You should have played 26...b5 and prayed for the best.

It's quite clear that black is better in the knight vs bishop endgame, but is there a clear win?
I'm quite curious as to a sample variation/plan that could lead for a win in the knight vs bishop endgame. 

I haven't done any analysis of this yet, but it feels like the sort of won position that Belyavsky wrote about in Winning Endgame Technique. The basic plan is to freeze the e-pawn then force White to defend it and create a second weakness on one or both wings. 

It is a position in which both sides want their pawns on white squares, like your opponent has realized on move 30. The advantage you had is imho disappeared at that time. I would therefor try to get my pawns on white squares and let his pawns be on black squares. In that way I improve my bishop and restrict the use of the bishop of my opponent.

Furthermore should I try to get a majority on the queen side, hence accepting that white has a passed pawn in the center. I consider a passer on the far queen side a bigger threat than a passer in the center.

However I should make the first move on the king side and play 25... h5. I wonder if 25.Kf2 is his best move. Should it not have been 25.g4? It is very difficult for black to make progress on the king side after 25.g4 and it is very difficult for black to penetrate the queen side as long as the pawn on a2 stays at it place. That is why I would play 25... h5.

At move 28 I should not answer with d4. If white captures on d5, then is he helping black to get his king in a good position. So, that gives black a free extra move. I think I would have played 28... a5. I would expect white to answer with b3, After b3 could play have continued 29.b3 c4 30.Ke4 cxb3 31.cxb3 a4 and you have your passed pawn. If white would play exd5 then will white help black to improve the position of the black king. That is the move I would not expect.  Therefor do I think that 28.e4 is not the best move for white. If you play a pawn and create a threat to capture a pawn of the other side, then should it be a threat for the other site too.

Wut lol

Yes, I think so!

ciarli, do you know the endgame?