Really weird Eval

So i just got a really interesting thing. This position is fascinating. 

So Stockfish calls Nxe5 a blunder and says it is -4.2. What i saw and the Tablebase backs me up on this is that it is a draw because wrong color bishop. Is Stockfish not equipped with a database? Or is there another thing that could explain this? 

I don't know.

I think it is a win for them, because of their passed pawn.

bruhhh

what?

Just a draw no matter what

the bishop gaurds the pawn, when you run your king over, and the king gets the other pawn, and forces you to move.

The king puts you in zugzwang after winning your pawn.

I put the drawing move. Nxe5+ is just a draw. The Bishop is the wrong color

This is way above my head lol

What tablebase says is correct. If stockfish contradicts the tablebase, then stockfish is wrong. Tablebase is never wrong. Yes, this is indeed the wrong color bishop endgame and black can't win it. 

engines cant do endgames like that

That is why i played it. I was definitely confused when stockfish called it a blunder. So i put it into the tablebase. It seems lately like Stockfish is very generous with giving Blunders and very stingy with Brilliancies. 

It is super instructive. Basically if the promotion square and the Bishop are opposite colors and the king can get there it is a draw.

So what would the computer have played there? Would it have done some weird shuffling? i just thought that was the simplest way to hold and i had like 13 seconds on my clock.

engines are bad at evaluating endgames

I mean i knew that but this seems a bit over the top. i mean even an idiot like me saw it.

I think its because the browser-base stockfish is weak, its just an elementary draw

yeah but thats a fortress. Engines don't see those

OH that qualifies as a fortress! Makes a lot more sense now.