At the last Open of Touquet (France) the talented and young FM Sim Maerevoet told me that he should once store his scoorsheets digitally on his computer. I thought this was a remarkable statement. I am used to that older or lower rated players don't store their games in a personal database but I don't expect that from an ambitious player with virtually approximately 2400 fide.
It appears Sim barely or never analyzes his standard games with an engine. That is the opposite of how I work. Besides now I also start to wonder if my many hours of analyzing with engines is really useful as meanwhile Sim has 100 points more already then I have. On the other hand I can't imagine that I would get better results by playing countless hours of blitz and bullet at lichess like Sim does.
By the way I've being doing that in the past too and it didn't help me to improve but maybe it is related to the age of the person. Older persons make much harder progress than teenagers. Just maintaining their rating is already difficult. I am not doing so bad when I compare with my contemporaries. Maybe it is not very efficient to analyze games with an engine seriously but it definitely doesn't harm.
Nowadays most of my analysis is spent at deeply investigating the opening and sometimes I do get rewarded for it. In the last round of Open Brasschaat 2019 I used an idea which I had discovered in 2014 after some very extensive analysis made of the opening see my article fashion.
However I have to admit that such games are rare. It can take years before I can use some of my analysis and then they are often already outdated. This can be because the newest engine found some holes but also because new trends have popped up in games between grandmasters. Also I do notice that because of the exponential growth of my opening-analysis that I started to forget more often (see e.g. harakiri). In the last Open Brasschaat this was really annoying for me as I suffered from amnesia in 5 out of 9 games. Sometimes I just lost time but in 3 games also my opening went wrong.
I am sure it is much more efficient to read a new openingbook about an opening and immediately learn something about dozens of lines instead of just checking 1 line deeply. Still I do find it 100x more fun to discover 1 idea independently than read 100 new ideas in a book from somebody else. Maybe I should switch to watching dvds as reading books is not only hard work but it is also not easy to remember everything.
Analyzing middlegames is even more doubtful than openings. I spend about 1 hour at it for each game but I don't think that I learn something which can be used in another game. The positions are always very different which makes it almost impossible to find connections between them. Sometimes you also need to accept that we will never have the same tactical skills as the engines. In the earlier mentioned last round of Open Brasschaat I got a chance to become immortal by playing an absolutely insane combination. I showed it to other FM's but everybody agreed that such combination was alien.
Beautiful isn't it but I think the chance is 0% that I will ever be able to execute such combination in another game.
After the last standardgame of the most recent worldchampionship a journalist told Magnus Carlsen that the engine shows +2 after 25...b5. However Magnus answered coldly with "I don't care" see world chess championship game 12 carlsen offers draw in better position to reach tiebreaks. He is right of course because he has proven time over time that an evaluation of an engine doesn't have the last word about how chess will be decided between humans. Although I do use engines to define the quality of my games, I also realize that it is impossible to imitate an engine.
Nevertheless i do recommend analyzing your middlegames with an engine when you are still learning and improving a lot. This is especially important when you don't have access to a good coach. I estimate my chances would've been very slim that I would be today an FM if I had before never access at all to an engine. Today the return of analyzing my games deeply has been diminished a lot for me. I still do because chess is more than just playing games and winning rating. I also enjoy it immensely to discover combinations like the one above. I am interested in the truth so like to know what happened exactly in my games. Chess is not only a game for me but also science and art.
Endgames I consider more interesting to analyze properly. There are only a limited number of games reaching an endgame in which the result isn't fixed yet. You won't win much rating by studying deeply endgames. On the other hand endgames which are misplayed are often very painful. They often make a big difference in open tournaments especially at the end when the prizes are announced. Also endgames can often be categorized in groups. It allows you to formulate rules for a certain category. I gave an example of that in my article exchange pawns when standing worse and at below endgame against the Belgian expert Sterre Dauw I spent quite some time to find out how things should be evaluated.
These type of rook-endgames in which 1 player has an extra lonely pawn at the other side of the board is something which occurs regularly in games. However even some FM's admit after the game that they had not idea how to evaluate such endgame correctly and even less which moves to play. I learned that it can depend from some small details like just 1 square difference of the h-pawn or king defines it it is won or only drawn. I also found it interesting to discover the concept of the 4th row line-up for the white pieces which allows the transfer of the king to the other side.
So experienced players won't gain much by analyzing their own games. It is more useful to track the latest trends in your repertoire and just play a lot. That is how the modern strong master is trying to further develop. Analyzing games is still useful for publication and archiving of course. Very likely this blog couldn't exist without it. The statistics of this blog prove that although many players don't have much time to analyze, they still like to look a bit deeper than just the shallowness of a game.