why endgames are so hard ?

I am playing a lot of classical games lately ,because i want to improve , but in most of times I achieve an drawn endgame , or I have some advantage and at the end I lose like this game

Any advices ?
it is better to keep myself in the middle game till I am sure the exchange will give me a winning endgame ?

because all that's left in the endgame are the problems caused by positional mistakes

Here are some good books:

"Shereshevsky, M - Endgame strategy". Will teach you the feeling for the endgame.

"Rosen, Bernd - Chess Endgame Training". Will teach you the endgame standards (except the fundamental mates). E.g. opposition, key squares, reserve tempi, rook activity, etc. Has three chapters about rook endgames.

archive.org/download/CHESSCollection has these two books contained.

A third book, which is big but will transform you into a rook endgame pro, is the book by Lewenfish and Smyslov. Unfortunately, this doesn't exist online as PDF. I have worked through parts of that book, and I would have easily drawn this rook endgame, starting with 33.Kd3, giving him the b2 pawn, but activating my pieces.

The raw pawn endgame was dangerous, because he had more reserve tempi. Always try to avoid raw pawn endgames, when you are not 100% sure you will draw them. After 43.g4?? the catastrophe happened (43.Ke3-f3 instead, he can not make progress), he had the opposition after 45...Kxf5 and zugzwanged you away from your d4 pawn.

g4?? was suicide, yes.

You played the London system, of course you got punished. LS players at your rating are mostly lazy players who want to play the same position every game, which indicates limited learning ability and bad endgame skills...

The London System is easily refuted, at master level White only has 17% win rate vs. Black's 33% after 1. d4 Nf6 2. Bf4 c5 3. e3 g6. Its performance for White is even worse than the Alekhine defense for Black.

@OSpengler I am asking about endgame , but if you have experience give some new opening to start with , i know that it's better to stick on one opening if you are under 2000 and focus on middle and endgame , and after you reach 2000 , you start building an opening repertoire

Maybe generalizing from myself, I reckon the reason that endgames can seem extra hard is because a) we don't get as much practice at them as we do at middlegames (because most games have a middlegame but a lot of them at lower levels will be essentially over before the endgame) b) we've already been thinking hard for a long time when we get to an endgame, so we might be less focused, and c) we probably don't study them as well as we do middlegames (like, how often do you play through an annotated game, go through all the middlegame stuff in reasonable detail, and then get to an endgame and think "okay, and then it's just some endgame stuff, apparently that move was a mistake, okay, whatever..."?)

"it is better to keep myself in the middle game till I am sure the exchange will give me a winning endgame ?"

Yes.
And even better is to win in the middlegame. Winning the endgame requires more precision.

The game you showed was a difficult endgame. I had no idea what is going on. Also your opponent played a perfect game. Don't expect to beat a 2100 player, especially in the endgame.