How did Nakamura win three of the first five matches against Carlsen in the final of the Magnus Carlsen Chess Tour?

After studying the games, I finally cracked Nakamura's secret for myself:

Nakamura does not care at all about the objective evaluation of the position! 

This superpower allows Nakamura to find practically dangerous opening ideas, and defend more tenaciously than any other player in blitz or rapid chess.

Instead of caring what the engine would think, it's much more effective to ask, 'How can we make the opponent's life as difficult as possible?'

Here are some examples:

What would you play as White, in the diagram below?



In this position, the computer gives 14.0-0 0-0 15.Rfe1 with a small edge for White, but no real clear indication of how to press Black's c4-pawn or break through. Carlsen's move in the game, 14.h4, is practically much more unpleasant, as it discourages kingside castling (due to Ng5) and sets up Rh3-g3 as well as h5-h6 to target Black's kingside.

The computer recommends 14...f5!? in reply, to stop Ne4 in many positions and get counterplay down the g-file after 15.exf6 gxf6. Instead, Nakamura played 14...Bc6-a4, which is a mistake objectively, but it baited Carlsen into a double-edged exchange sacrifice later, which he misplayed and went on to lose.

After all, Carlsen sensed that his position was very strong, and this led him to spend a lot of time looking for a strong move that was not straightforward in the position - and that proved to be his downfall. 

Another example, from this same tricky 1.d4 Nf6 2.c4 e6 3.Nf3 d5 4.Nc3 dxc4 5.e4 b5!? Vienna line Nakamura used to beat Carlsen twice with Black!



In this position, Carlsen and his team had analyzed the move 17...0-0-0 very deeply. Therefore, Nakamura deviated from the engine's first choice and played 17...Qd5 18.Rg3 Qxc5, simply grabbing a pawn and asking White to show his compensation!

According to the computer, 19.Qf4! is good for White, but from a human perspective, it feels unnatural to allow ...Nd5 with tempo. Instead, White quickly went wrong with 19.b4!? cxb3 20.Ba3?, which is the sort of thing you play when you think there's a very strong move in the position, but are uncertain what it is. Nakamura grabbed the material with 20...Qxa3! 21.Qxa5 Bd5! 22.Rxg7 c5!, and in just a few moves since deliberately avoiding the engine-approved moves, he had a winning position.

A key lesson I've learned from these games, and from studying many books, is that even very strong players often make mistakes when they are surprised!

If a player is objectively stronger than you, you can use their strength against them by playing a move that the stronger opponent 'knows' is inferior, based on their superior understanding of chess, but which is very hard to prove over the board! In fact, this is how many lower-rated players have defeated me in the past. 

Nakamura got to the no.1 rating in blitz (both on FIDE rating and Chess.com blitz) by mastering this approach, and I have a feeling we're going to see the other elite chess players follow suit soon as well. After all, one of the other semi-final qualifiers, Daniil Dubov, already exemplifies this approach in his own games, with his very original ideas and dynamic style. 

- GM Max Ilingworth

it is very big paragraph!