Maybe 17. ...Bc5 was better than Bb4? That way the king is basically trapped in the center and if 18. Ne4 the e5 pawn hangs.
Post Mortem against a Chess Expert
I have to say, my intuition says th sacrifice should be wrong and losing, if white plays correctly.
I can't find any clear attack for black after a few accurate moves by white. And in any endgame, black would be lost.
That said, white started off with the strange 15.Bg2. I think black is already starting to have some compensation (even if still not sufficient, in my opinion).
15.Qa4!? seems interesting. Qxf3 Rg1 just makes white's two bishops (one of which is an extra piece, sooo much more active. Materially, it'll be fine for black, but in piece vs. 3 pawns in middlegame positions, if the piece can become active and the pawns aren't passed or otherwise dangerous, it's just a decisive advantage for the piece. White is almost winning there in my opinion.
Qa4 introduces an active defense, which seems logical since white is up a piece and under no severe pressure yet. Using the extra piece should just win: 15.Qa4 prevents 15...Nxd4 because of 16.Bxd4, and if black dares to play 16...Rxd4 17.Qa8+ Kd7 18.Bb5+ ouch (can't play Ke7 because of Qe8#:)
All that said, still after 15.Bg2 white is doing ok, even if it's more complicated. 15...Nxd4 16.0-0 seems almost forced, although I have to admit I wasn't 100% certain about that at first. I am now.
16.Bxd4? and now, black is the one almost winning:)
yeah, 17...Bc5 would be devestating. Punishing the white mistakes in the last two moves.
After that, 18...Rhd8 would keep the position very complicated, instead of simplifying into a lost position with 18...Bxc3.
19.Rd2 is wrong. Your comment shows a lot about the problem in your way of thinking:"making the position more simple was my plan". WRONG.
That must not be your plan. You're a piece down, and the pawns you have for it won't be dangerous anytime soon. The two pawns you have for it aren't nearly enough. Even three won't be enough if his pieces can become active.
Simplifying in such a position is simplifying to a lost position: making the position an easy win for your opponent. So the problem wasn't only missing a move, but rather, looking for the wrong plan. You do NOT want simplification here.
White is probably already with the better chances to win here, whatever you play, but if you simplify before something special happens in the position, you will lose the game.

You're definitely right about Rd2, which I noted in my own analysis... It's simply not right.
Here's something I should point out. Since I made this all from memory I'm not actually sure if white played Bg2 or Qe2 first... However the position that transposed was the same either way...
You're definitely right about Rd2, which I noted in my own analysis... It's simply not right.
Here's something I should point out. Since I made this all from memory I'm not actually sure if white played Bg2 or Qe2 first... However the position that transposed was the same either way...
Unfortunately, the exact moves are the important thing, since, as I've mentioned, white had ways to easily gain a huge advantage after the sac.
Every change of move order may change everything else too along the way.
However, the important thing was that as I've mentioned, the sacrifice was incorrect, and a bit of active play by white would have been enough to crush black, even before that position was reached.

No in this case the change of move order affects nothing, the exchange takes place and the bishop moves to G2,the exchange takes place and the queen moves to E2.
The position is the same.

You're definitely right about Rd2, which I noted in my own analysis... It's simply not right.
Here's something I should point out. Since I made this all from memory I'm not actually sure if white played Bg2 or Qe2 first... However the position that transposed was the same either way...
Unfortunately, the exact moves are the important thing, since, as I've mentioned, white had ways to easily gain a huge advantage after the sac.
Every change of move order may change everything else too along the way.
However, the important thing was that as I've mentioned, the sacrifice was incorrect, and a bit of active play by white would have been enough to crush black, even before that position was reached.
You're probably right.
So today I was making one of my bi weekly excursions to the Second Story used book store in Canmore. I had retired to the library with a chess book and was using the libraries chess board and pieces to look through some stuff halfheartedly. Suddenly a young-ish man (late 20's early 30's mayhaps) approached me and we started chatting chess.
Turns out he was a 2000+ rated CFC (USCF equivalent I believe) member, making him a chess expert, and he used to run a chess club in Canmore.
So we had a few games, the first was a mainline Ruy Lopez Zaitsev variation (I being white), he made some nifty moves and I was slowly squeezed to death... and I lost a pawn, what of it? Well then the second game started. He opened with E4 and I looked through my options, did I want to play E5 and face either the Lopez (I surely didn't want this from the black side this time), the Kings Gambit (shudder), or the Scotch?I decided to play an opening which I've never played before in a game... the ALEKHINES DEFENSE!
What follows is a very (in my opinion) interesting game.Hhowever I'm really unsure about how it could have played out. I lost in the endgame, but I think late middlegame I had definite drawing chances. I sacrificed a knight for 2 pawns and an attack, with interesting possibilities.
I'm playing through the game here from memory, so I hope I get everything right, if something looks wrong, tell me and I'll edit it.