All is not lost!

This is one of my most recent games. It is one that I hastily resigned, not realising that the position was not only wholly salvageable, but worth a considerable amount. My opponent shattered my plans with an unexpected move that I thought was dangerous; seeing my hopes for some neat tactical ideas demolished, I angrily put in the towel, ceding that I was too short-sighted. 

I shall post the opening and then leave the salvaging move as a cheeky puzzle (I have to keep you all interested), before I ask the question that the analysis can't solve.

Having resigned, I still decided to run the actual game past the computer analysis, more interested in what it made of 14. Nb5 as opposed to other offerings, rather than having it suggest that there was a wholly accurate continuation for white that I missed. That's the puzzle:

I hadn't even properly considered b4 as a move, but I realise now the simplcity. I think there's a lesson here right from the off about genuinely looking for every continuation and not simply bowing out because your opponent plays one unexpected move, or because your best laid plans are wholly dismembered.

I am not really sure of how to add additional lines and to show where all this goes within the context of the puzzle, suffice to say this: if Black proceeds with Qxb4? then c3 chases away the queen, allowing Nxd6+. From there, white's queen and rook can make dangerous forays into black territory. Surprisingly, white's defence is more harmonious that it appears. 

Now, the computer analysis can tell me about what happened in the game; it can point out that the position was really not lost and that I am a fool. That alone, with the puzzle and example, I wanted to share as a little piece of chess wisdom - all is not always lost!

But, the big money question is: given that the computer puts me with a score of 8.68 with Qa5+, do moves like Nc3 still allow white to win or at least continue somewhat evenly, or is the loss of a piece too great? To put it another way, even if I hadn't seen b4, did I resign a little too prematurely?

I recognise that white does have a few things in its favour. The rook and queen are nicely active and bringing my other pieces into the game wouldn't be difficult; black's king isn't going to be safe on either side and he has more to develop.

On the other hand, white is now down a whole piece and my queenside pawns are looking decidedly vulnerable. 

So, I am putting it to the community to take a look, to learn a little as I have done in that I probably do need to think a little harder both when I resign and about tactics, and to hopefully offer some evaluation into the rest of the game. I'd like to know which side your money would be on. If you have anything to say about the game, don't be afraid to post. 

you missed 15.Qc3 while 15.b4 loses.

no Qc3 loses to Qxc3, b4 then c3 survives after queen takes bishop and night takes night check

Exactky what i mean!

Thanks to all those who had a look. To get to all these in order:

2...Nf6 is perfectly playable. I was hoping to engage some tactics and assert an aggressive style by playing Bg5. Maybe e5 was a little calmer - certainly the computer analysis preferred that. I do think that my opponent played his opening a little awkwardly and led himself into a lot of trouble...but I didn't capitalise.

Qc3 is an interesting alternative. I had considered it, but losing the minor piece did displease me somewhat. However, I hadn't seen Nc7+ which wins back some material, though as you say, black is still better off since the knight can't be saved. It leaves quite an interesting game - rook and messy pawns with safe king vs. two minor pieces and connected pawns with weak king. I'd probably rather have black, but white is not lost. Whatever the case, Qc3 is a whole lot better than my other suggestion of Nc3.

I'd be quite interested if someone could put a numerical figure on the value of the game at that point - after the Qc3 continuation.