After five rounds it’s time to take a look at Keymer’s performance:
In the first game against Carlsen he was simply unable to build up the attack on the kingside, to construct the machine so to say.
In the second round it was quite similar. Keymer got maneuvered in a position where he had to demonstrate something on the kingside. Once again he failed to build up and allowed Anand to take the wind out of his sails. He hesitated to take a pawn and lost.
Against Caruana he tried another kingside demonstration, missed the best continuation, underestimated the counterplay and lost.
In the game against Naiditsch he didn’t get his queenside going, underesimated Arkadij’s kingside attack and got crushed.
And finally in the fifth round he didn’t even bother with building up on the queenside and sold his huge advantage for a single pawn. Very important here is that Leko called the resulting position drawish and it turned out to be equal. Once Keymer got confronted with a study-like problem he found the solution. This game raises all red flags, because it shows that Keymer has problems in evaluating positions: He undervalued his attack and overvalued the pawn.
You may have noticed that I am repeating certain patterns, because I think that’s where the problem is. Keymer was simply unable to build up positions and not because the moves weren’t there, he just didn’t find them. Once it came down to solving concrete problems he did very well. The big problem here is that he isn’t a fast superficial player. He is actually taking quite a while to come up with not that much. The game in the fourth round is the best example, because Naiditsch showed him how to build up.
To make it short: Keymer is rather weak strategically, tactically about average for his level and rather precise in the endgame.
What can be done about it? The typical suggestion would be to “study the classics”. Look at Nimzowitch vs. NN and do it exactly like that. Such advice worked for Botvinnik in 1925, but nowadays it’s simply outdated. If you want to study something from the past then you could start with the late 1950s, because they were playing the structures back then that Carlsen is playing now. One question remains though: What can you really learn from them?
There is another way to make progress though and that is to either skip the middlegame or to choose openings where certain elements are missing. For instance there is no middlegame in the Marshall Attack. It’s all tactics until it’s either mate or an endgame. There are no kingside or queenside build-ups in any of the Open Games. In such a sense Keymer is playing the wrong variations.
On the upside I think he is quite old for his age already, in other words he is not acting like a child in interviews. He can solve concrete problems and his openings with white lead to positions that are at least objectively advantageous. This is also a typical problem of modern preparation. You know a line until the point where the engine shows an advantage and just go from there. Sometimes you don’t even know why the position is supposed to be that good and have to figure it out over the board. That is where they are catching him right now, and it also explains his timetrouble. Positive is that he fighting and not trying to gain Elo by drawing.
Will he make a bigger impact on the scene than the 37 other kids who became GM before the age of 15? I guess it’s too early to tell.