If you ever played Poker you probably came in contact with Game Theory. Chess is a 2-player game with complete information about what position is actually on the board. Lot’s of factors are unknown though. Very often you don’t know what the best move is, you can only approximate it. You don’t know your opponent’s next move, you can only anticipate it. You don’t know what your opponent knows, you can only assume it. You don’t know how good his physical or mental constitution is, in fact you can’t even be sure about know your own condition. You don’t know the risk level that your opponent is willing to accept. What is he or her playing for?
The most important question is what to do in must-win situations. Let’s say you are an upcoming player who wants to reach 2700 then you can’t archieve that goal by just drawing 2500 players. Should you follow a pure strategy and only play a certain opening that is supposed to be maximally complicated for instance? Well, in that case you may be setting yourself up as a sitting duck.
Going back to Poker, let’s assume you are playing a heads-up limit-game. Your strategy for acting second is this: If your opponent bets on the river you call, otherwise you check behind. Once your opponent realizes what is going on, he will only bet his strong hands and check with his medium and weak hands. After a while you figure this out and change your strategy: When he bets you fold, when he checks you bet your strong hands. What will he do after a while? He will bet his medium and weak hands and check his monsters in order to check-raise you. Once you notice that, you will start calling his bets and check behind again, and we are back to square one. It’s just an endless cycle of adjustments.
What is the alternative? The alternative is playing a mixed strategy. You act randomly based upon optimal frequencies. Since your opponent is unable to detect a pattern, he cannot adjust to it and prevent you from maximizing your expected value. Finding out what these optimal frequencies are is obviously what this branch of science is all about, but that doesn’tt matter here. What matters is how chess players can make use of it.
Chess players can create a portfolio of openings and randomize. Ivanchuk took this approach to the extreme. He has a memory like a harddisk because he is a savant like Rainman. If you are not a physical freak like Chucky you can still do it to a lesser extent, with two openings for instance. You could even switch between major variations within an opening, let’s say the Nf6 or Bb4 in the French or Bf5 and Nd7 in the Caro Kann. Before the game you simply flip a coin, if you have more than two options you simply throw a dice.
Another concept that comes from Game Theory is the so called “free-roll”. Back in the day some casinos in Las Vegas offered the opportunity play one round of Roulette on the house. In Poker the classical free-roll is A♠ K♠ vs. A♥ K♣ on a Q♠ 7♦ 2♦ flop. If it comes runner-runner spades (~5% chance) the first hand wins, otherwise both players split the pot. In chess the concept can be applied by playing highly irrational forced lines, hence the forced draw. If your opponent makes all the only moves, it will be a draw, otherwise you will have a winning tactic. Grischuk plays like that.
How does all of this relate to Carlsen? What is he doing? Carlsen is taking a calculated risk in order to escape from the prison of the forced draw. He gifts his opponents + 0.25 in the opening, but the game will last at least 50 more moves, where his moves are 0.025 pawns better on average. He will eventually be a pawn up and it will all boil down to technique where he is outclassing everyone.
Continuing the topic I decided to take a look at Leko’s recent own games in the Najdorf. The result was surprising or maybe not.
It is a small sample, but maybe for a reason. The whole setup doesn’t seem suit to his style and apparently he himself came to the conclusion that he shouldn’t play it more often. Overall he played it 12 classical games with it and his score is no wins, 2 losses and 8 draws. This is a reflection of the recent trend: The Najdorf has become a negative freeroll where black has to find the difficult moves to make a draw. Chess is not transitive, therefore Leko’s own problems with the Najdorf may or may not swap over to Keymer, but so far I would say they do.
Keymers results in the Najdorf are not encouraging and it seems to me that it is simply too ambitious for him at this stage of his career. He is sitting there like a dead duck and he obviously has huge problems to get feeling for these positions. Yes, he scored a few wins with the Najdorf, but it was against weaker opposition where any other opening would have done the job just as well.
I fully understand what Leko is trying to do. He has a clear vision of a perfect player who could challenge the World Champion and he wants to project this vision on his pupil. The Force is certainly strong with the young padawan, but he is not that strong (yet).
Last but not least, the Najdorf is a good opening, but it may not suit Keymer’s style. There is a reason why Caruana isn’t playing the Najdorf and it is not that he couldn’t remember the lines. Even Kramnik tried the Najdorf for a few games and gave up on it.
Ok, let’s just assume there is method to the madness. Since young Keymer is the only german prodigy with a shot, he is in somewhat of a do-or-die situation. It’s either him or nobody, and if the Najdorf is the required weapon to rise to the very top, then he has to use it.
The strategy for beating underdogs is to avoid sharp forced lines in order to outclass them in the long run. This usually means playing second rate openings to a certain extent. One could even call it some sort of positional gambit. In the first round of the Tata Steel it didn’t always work since even the world’s best Anti-Drawmeister couldn’t break his solid opponent. Nice try by Carlsen though!
It went rather smooth for Anand, who just played normal moves in the slightly passive Caro Kann waiting for his opponent to blunder.
The successful hustle by Nepo should also be mentioned.
Recently there has been lots of talk about the german prodigy Vincent Keymer. The kid is certainly a huge talent and obviously the only realistic hope for german chess to produce a world class player. He worked with Jussupow and currently he gets coached by Leko. Yet there is a slight problem and that is his performance curve. It’s simply way too flat, especially for the massive support that he is getting. If this is some sort of valid indication, he may even peak out barely around 2600. What do you expect from young prodigies these days? You expect them to be lightning fast and improvise a lot, especially in the openings. Keymer is totally different. His approach is very objective and very scientfic. He is not trying to hustle his opponents, he is trying to play precise. But that’s probably what is holding him back. Engines have more or less killed the scientfic approach to chess. Modern chess is all about physical ability and performing well under pressure.
On the positive side Keymer managed to avoid the forced draw, but eventually it was him who got outplayed. That is pretty bad news. Paehtz on the other hand had probably set her goal on a draw all the way, therefore she accepted it, even in a winning position. That’s also pretty bad. Overall I would say that this game was very interesting, but certainly nothing to cheer about.
Keymer’s performance according to the database:
2016: white 2418 – black 2360 – avg. 2387
2017: white 2497 – black 2333 – avg. 2414
2018: white 2525 – black 2504 – avg. 2515
Let’s talk about the specific problem a bit more in detail. The human brain grows out around the age of 15. This means that Keymer has about one year left to reach his full mental capabilities. That’s about one year, maybe one and a half, to close a gap naturally by just growing up. The other way to gain strength is to work hard on his openings. Let’s take Gustafsson for example. He was just an international master with Elo 2450 who was basically playing the trash that you could learn from Batsford books at the time. But then around the year 2000 he picked up “Openings According to Kramnik” by Khalifman, switched on the Shredder engine, checked the lines in detail and developed a world class repertoire which shot him up all the way to 2650. That sort of jump is simply not available to Keymer, because he already gets his openings from Leko. There is simply not much room left for massive improvements. Leko managed to improve the results with black, but that’s about it. Winning straight from the opening is practically not possible anymore, because everyone has access to critical evaluations within a couple of minutes, not within a couple of days as it used to be in the past. That’s the reason why Kasparov retired, that’s the reason why Leko dropped out of the Top-20 and that’s the reason why Gustafsson gave up on his ambitions after all. The only way for Keymer to gain a lot of Elo-points fast is to play the Drawmeister in closed tournaments or to play matches with some overrated veterans, guys like Igor Rausis for example. They have to select his opponents similar to managing a boxer.
This is what Stockfish thinks about the starting position:
This is what Leela comes up with almost instantly:
I find it rather amusing, because this line of the Classical French is also my repertoire suggestion. On higher depth Leela will switch to the d3-Caro Kann by the way, which is at least mildly surprising.
Well, sometimes things turn out to be much different than expected and the same goes for my “novelty” in game 2 of the World Championship. Initially I thought that I had discovered something amazing, but then Ramirez said on air that the line had already been played in a correspondence game. Ok, fine…I can live with that.
At the time of writing I only knew the article from Shankland who doesn’t mention the line at all. Later I found an article by Saddler on the match in NIC 8/2018 and he dismissed it as good for white. Guess my surprise when I realized today that Dreev had already given the line up to move 25 in his book on the Bf4-QGD quoting the following game:
Therefore my final take on the variation starting with 11…d4 is this: It’s a novelty unless you already knew it.
Since I already mentioned certain problems with published theory, here are five examples where the improvement can be found by simply clicking the engine button. This isn’t unfair to the authors, because they used engines to check these lines too.
The first one comes from “The Sicilian Accelerated Dragon” by Peter Heine-Nielsen. This book dates back to 1998, but even then Fritz 3 was already strong enough to find the refutation (you don’t even need an engine because it’s obvious, way too obvious).
The second one can be found in “Openings for White according to Anand Vol. 7” from 2006:
The third example comes from Kasimdzhanov’s DVD on the Classical French where he analyzed the following line for white:
The fourth one is another recommendation by Nielsen in his DVD on the Sicilian Dragon:
The last example is from “Attacking the Caro Kann” by Dreev:
To be continued…