Need an advice. I almost always see the best move but don`t play it

I have a problem that I don`t know how to solve. I often see the best move in the position (more in complex positions with tactics or something) but i rarely play it. I think part of this is I am not confident enough about myself. May be there is a tip how to fix it. It looks really stupid sometimes. For example in this game (I made a diagram but if someone wants to check it I add a link https://lichess.org/8RWSqAkA/white#0)
So my question is. Maybe someone can give an advice how to deal with this problem? Cause i am just tired of myself. I sometimes play all the best moves on the engine but then I blunder for some reason and most of my blunders are just not making the best move that I see, consider and calculate

Stop thinking about the engine. Sorry to say I took a little peak at your profile, and your rating is lower than mine. And let me tell you friend even in the 1300-1400's the engine will still tell us we play like crap, even when I think I played an absolutely brilliant game and thought I crushed my opponent, I'll analyze it and the engine will tell me I screwed up several times. And that's because the engine is, it's perfect, we aren't. Stop worrying about that and just worry about pressuring your opponent. Make them constantly have to think about, even better, rethink their strategy, until they screw up and leave an opening for you to attack. "Top engine moves", Don't mean a damn thing to guys like us. Get better at applying pressure

Play more puzzles. Because at your current level, pins, forks, and forced moves are what's gonna take you to the next level. And you need to learn how to spot them through pattern recognition

But yeah Iooked at your thing and I can't possibly understand why you would choose the pawn you did to take the knight. Should be second nature to open up the file for the rook, especially if you gotta double a pawn. So yes, that was a bad move. Just play more games and puzzles and you'll get better. Don't quit bud

But me personally would just go queen after taking knight and try to get something on that file anyway. So if you resigned in that position that's also apart of your problem

Qf3

In no way shape or form have you lost the game yet even after taking that way.

Queen h5 looking deadly in too, nah man, quit resigning

At the very least it's nothing but an equal exchange of material

I understand your knight on d4 gets taken, but that's an equal exchange, and you can line something up while they take it. That's tempo. That's what you need to work on

If this position scares you then you don't understand tempo

If one of my pieces is dead, I just focus on my next attack and leave it. I've beaten a lot of opponents by leaving them "bait" if you will so they miss what I'm doing somewhere else.

The knight on d4 is dead sure, but opponents king side pawns looking awfully weak. Game wasn't over. All I'm saying

It's not even dead, you can move it to f5 and offer a bishop trade. I just noticed the king side pawns because I'm always looking for weakness, and that's where I'm aiming

Nvm, my bad you can't move it, it's dead, but still, like it said, this game wasn't over

Denozavvr, if you literally are playing "all the best moves on the engine" most of the time, and, when you don't, it's because you saw the best move but didn't play it, then your chess judgement is super-human, far beyond Magnus. Colour me sceptical.

I think it's much more likely that your chess judgement is about the same as any other player who plays as strong as you actually play. More often than not, a move that looks good to you really will be good. Occasionally it will turn out to be bad. Sometimes you'll consider two or more good-looking moves before picking one of them. And sometimes the one you picked will be a bad one, and the one you rejected will be the best. That's how it goes. For everyone.

In respect of the game you link to, and the specific move (13 hxg3 vs fxg3), you say that "for some reason" you didn't play the right move having correctly analysed it? Well, what was the reason? At a guess, it was because you got scared of the check it opened up. You analysed it, saw that after Kh1 your opponent had no way to continue the attack, but felt scared anyway.

I had a similar situation in one of my games a while back. I saw that the move I was considering opened up a pawn-capture-with-check, but after analysing it, I concluded that there was no way for him to continue his attack. So I played the move.

My analysis was correct as far as it went, but I missed something. There was no way for him to continue attacking, but the move gave him a defensive move which would, if he had found it, have left my attack dead in the water, which would have been losing for me because I'd already sac'd a knight to get the attack in the first place. Fortunately for me, he didn't find the defensive move, my attack continued and I eventually won the game. It was still a bad move though.

Seeing the best move and knowing that it is, are completely different things. It takes only a moment to enumerate all the moves in any given position but a whole lifetime to compare them against each other in many of them.

@OP

The reason why you didn't play fxg3 is that you do not understand why it is better than hxg3.

Sure, engine here takes with the queen and you get material advantage, so it is clear, but human wouldn't play it that way, unless he/she too understands why they should do it.

The reason why fxg3 is better is not straightforward - well certainly not like you showed. I understand it, sort of, but the thing is that you don't get some material advantage immediately.
First of all, fxg3 is not really intuitive because the opponent gets a tempo move with check. But more importantly white gains a half open file and the access to f6 square that is not defended with a pawn, all the while black's king is weak.

So because white is winning, black (engine) understands this and gives up a queen. Why is white winning? Because black's king is weak and if white manages to put a piece on f6 in addition with Qh5, he will create a lot of threats around black king, and eventually if followed up correctly, white will be winning.

So unless you saw that during the game, you didn't know fxg is better, and let me tell you this, there is a chance that I wouldn't see it (during a really long game, say 1 hour per side), and probably some higher rated people than myself would have a chance of missing this as well.

Many lower rated players probably wouldn't understand why fxg is better even during analysis. 
In essence, you need to become a better player and you will miss less.

When you analyze your games, don't always follow engine recommendations. You should sometimes, but when you see something you wouldn't do (for instance you wouldn't really want to give up a queen if you were black), try a different response, and see why it is worse than the engine recommendation (giving up a queen in this case). You can learn a lot like that, and you will understand more going forward.