I can't figure out WHY this is a winning position

So I came up with the commented position during this game and I declined to go for it.  I know stockfish shows other moves leading to this position but I ended up declining the position and I went on to get into time trouble and lose the game. I also know there were mistakes made (eg. taking rooks) on both sides leading up to this.

My question for someone better than me... WHY is this position evaluating as -7 by java/chess.com stockfish at depth 31? I have even tried playing it out but I can't put my finger on why this is such a great position.  Is it something tactical I just can't fathom?

28.  Nxb8?

I think it was posted wrong.  The position I'm interested in is this one:

(17... Bxc5 18. dxc5 Ndxc519. Qd4All on my own, I came up with this position while evaluating my 17th move.  I deemed it mediocre and probably knife edge sharp with really tough ways to take advantage.  I see that his bishop is useless.  I see his king has a bit of a draft and may be easy pickings later.  I see I can castle which would provide some safety. Plus his knight on d2 is pinned and attacked.)

That is, instead of 17. .. Be7, I had contemplated the position after 17. ... Bxc5 18. dxc5 Ndxc5 19 Qd4.  I considered that position poor but stockfish loves it and I can't really understand why.

I had originally tried to post the PGN with the starting position at the alternative line ending in 19. Qd4.  

I lost the game b/c I'm sub 1300, ran out of time, and hung pieces. 8)  I'm not a great player but I love chess and I enjoy simply trying to improve with what little spare time life offers me. 

You posted correctly, it was clear what you meant.

It's tricky when the computer evaluation shows a huge advantage but no clear route to winning material or checkmate.

I guess you've got 2 pawns for the piece, very active pieces and a raging attack. The computer likes the calm 19...O-O at this point - but it looks like there should be a checkmate there somewhere.

Well done for finding 17 ... Bxc5, even though you didn't play it

That sort of sac is actually a common theme in these early c5 lines.

Umm... I'm not sure what to say. It's not a position that can really be explained in words, you have to look at variations. If you've already looked at some lines and it's still confusing, just ignore it. Sometimes positions aren't very useful for us to look at yet because they're too hard for us.

If I played this in a game, I'd be going move by move, I wouldn't have a grand plan other than to keep my pieces active, try to keep white tied up, and push my a pawn.

Good analysis and explanation DeirdreSkye.

I'd have gone for 22...Nb3 having not seen your 22...e5, but the engine prefers 22...e5

Nice analysis DeirdreSkye. I know I would have preferred Black in that line, but to be perfectly honest, I wouldn't have been able to tell you if the evaluation after 19 . . .O-O was -1.5 or -6.5. Black already has two pawns for his piece, the white king is stuck in the mddle, and Black's pieces are super-active. But these kinds of positions, where the losing side simply has no useful moves, are always hard to explain--or even recognize. Black doesn't have one clear winning line, but White is helpless.

I played it out a bit following the engine. After O-O the black rook comes to the open b file. That adds to the pressure on the white king. After a few exchanges black ends up with connected passed pawns, clear access to the white king with rook and queen, and the white rook still stuck in the corner.

I'm definitely not good enough to see deeply into why this is such a great position, but if pushed the biggest weakness is after castling blacks rook has a clear avenue to get in the game.

if white moves his bishop to get their rook in the game Qa6 sticks the king on the back rank keeping it pegged in (block with the knight on d2 and the pawn eats it, becoming a queen threat)

white's bishop is similarly useless, nowhere to go nothing to point at.

so white has no real piece advantage. black is essentially up a rook and it has a pawn white will find very difficult to stop queening if black wants.

After -

17... Bxc5 18. dxc5 Ndxc5 19. Qd4

• Material is equal if I'm counting right (I'm a bit sleepy) +1
• Black gets another passed pawn +1
• Knight gets developed into center with tempo +1
• More dominion and activity for black +1
• Etc.

If you take those points and you couple them with the variations that @DeirdreSkye and everyone else worked out, you've got one great position.

It's a solid move that costs nothing and gains a lot. It can be a bit scary to trade a piece for pawns, but a little endgame-studying can change that view after a while.

*EDIT

Why take with the bishop? Taking with the knight first won't force the queen to move from its current square, in other words -- not as much iniative. Also, those two knights guard each other really well.

A bishop would also be sort of blocked off from the action, as white's kingside is a little closed and clustered (to white's demise), but a knight can jump right in there. Also, it is said that knights and queens make a good fighting pair.

Is it probably "won by resignation" or is it a draw?

Everyone on this thread ( too many to mention ):  THANK YOU.  I swear, it's people like you who are exactly why I love this sport so much.  Your comments are incredibly insightful and helpful.  This is the one that I've taken away from this as my lesson learned:

"the losing side simply has no useful moves" -Laskersnephew

My problem is I need to learn to recognize it AND I need to be brave enough to go for it when I smell a weakness like the evaluated position I should have gone for.  I had suspicions about the position that were so high that after the game I ran it in stockfish.  

I'm also very flattered about IpswichMatt saying that finding the referenced position on my own was good.  It made me feel like a little Kasparov for about 3 seconds. 

This post is totally exemplary of why I love this community.  8)  I don't know why y'all (yes, the plural of you in english.... another story) devoted some of your time to my little 1250 ranked question, but it's deeply appreciated. 8)