Your surprise seems to imply the expectation that "GTO" is better than "non-GTO", which is apparently wrong - or at least not necessarily right.

If, for example, I'd be sure that my opponents massively overfold the river when I overshove, then I should do it all the time. Non-GTO obviously, but optimal (in that case). If instead I had no clue if my opponents overfolded or overcalled, I should stick to GTO - to avoid of getting exploited (accidentally).

Now, when Magnus has a good idea of how his opponents would react in those fields they are unexperienced in, than it's a good idea to deviate. And your understanding that he "sacrifices EV" is clearly wrong. If he did not know how his opponent reacts, it could be that his opponent - known or unknown - reacts optimally which will crush Magnus, but apparently that is not the case, so he knows what he does - and does not sacrifice EV at all (all he does is sacrifying some early value in trade of later return).

=> When my students asked me, why they should study / apply GTO when their opponents would not play optimal (or even good) anyways, that there's always the risk that an unknown player "accidentally reacts optimally". I call that "accidental exploitation" - and obviously, regarding our EV, it's irrelevant if we got exploited consciously or unconsciously. :-)

BigFiszh

Woah thanks for posting this great analogy with poker sirin

I think what he’s doing would be comparable to play a pretty unorthodox style preflop on poker such as “EccentricBG” did; coming up with a limp only strategy on 6 Max. I don’t know much about his personal story but I heard he had great results with it for some time because population was adjusting poorly against it.

In conclusion, I think the most important thing when playing an unusual strategy or when including unusual plays on your game is to have a good idea of how GTO works and how you’re deviating from it while playing the way you chosen to. This would help you a lot to react to possible counter strategies that you might find and re-counter them

Have you seen movie about Bobby Fisher? He used similar tactic vs his opponents.
And even if we dig in history (biography of great people) many did same in different areas of live Napoleon(war) and other examples. They choose ''-EV turns'' in order to take opponents in unfamiliar situations.
That's basic strategy of War

Have you ever 3B a fish because you know they play badly in 3B pots even though your particular combination is -ev as a 3B typically and you're not doing it because they overfold pre flop? I think that's a fairly typical exploit that a lot of people run which seems to follow this line of thinking

Yeah for sure of course I adjust based on exploitable reads, and I understand that GTO is only a framework for understanding the game theoretically, based on which we deviate when facing real opponents.

What I find interesting about the chess analogy is that even in a game which has been as extensively studied an analysed as chess, there are still opportunities for massive deviations from optimal play while playing against the best players in the world.

Much of the conventional wisdom around exploits in poker revolves around understanding the equilibrium and then deviating slightly based on our reads, so if we think someone is a little hesitant to pull the trigger, we fold a few extra combos of bluffcatchers on the river, that sort of thing. Or maybe we defend a few extra % in the BB vs someone who we expect to let us get to showdown too often.

But it sounds like the chess thing is more along the lines of what Nuno mentioned - having a limp only strategy at six max. IE making plays PIO makes 0% of the time, because we have thought about the implications of this and villain won't have.

[I am not a very strong chess player. I'm just familiar with it, having played it and followed it for many years.]

I just saw this thread. So, sorry for necro.

I heard about this claim that "Carlsen is deliberately taking -EV lines in the opening to try to get an advantage later on " in this other thread, and responded there.

The TLDR is that this claim is made by a commentator in a blog post. This commentator may or may not be right, but Carlsen himself never says anything like this in his interview.

Finally, one should keep in mind that high level chess, like high level poker, relies a lot on psychology. People often make inflated claims to try to rattle their opponents or get an edge.

Also, related to a post above. I don't think Fischer ever tried to deliberately play bad moves to try to rattle opponents. In Fischer's world championship match against Spassky, he did play openings he had never played before in his life, but he played all of them in a "theoretically correct" manner. His treatment of the Tartakower Variation of the Queen's Gambit declined in Game 6 is a model of logical and theoretically sound play.

Thats just pretty obvious. Not follow the “gto strat” in some street to make your opponent make bigger mistakes in future ones. Just the pseudo-mimic-gto guys dont understand that

What he is doing is purposefully making bad opening moves, but preparing ahead of time against whatever plays his opponents might make, so he knows the possible lines and they don't. It is sort of related to not playing GTO. Similarly, in poker you can make unusual or deceptive plays that your opponent may not respond well to.

There are implication for NLHE in the GTO era. For example, using nonstandard sizings preflop or on the flop. Or making a slightly less than GTO play that is deceptive or aggressive and hard for opponents to counter.

The Comparison with Carlsen, for making explotative plays isnt reasonable. Since first of all Carlsen didnt necessarly work with a equivalent to Pio. Or rather; strong chess computer have been around for years and everyone works with them. He recently worked with an AI named AlphaZero, and got some not findings out of it. His sacrifiecing "EV" isnt as Huge as you think. IN chess Positions are measured in Pawns rather than EV. So for example White has a positional Advantage of 0.20 Pawns. Carlsen in Opening Lines with White which are seamling even, but he plays those position the Best (White should in Theory aim for a Opening advanatge) Or he goes in Lines with Black which are evaluate maybe -0.1-30 but not more, so he deviates not as Hugely as the Comparison was made to Poker.
The Reason why Carlsen is crushing is, that he is the most constistend player of them all. He isnt Top on makeing the best move, though he's the one who makes by far the most second/third best moves according to Engine analysis.

In Short you cant really Comapre Poker Explotative Plays with Chess. Just out of the Reason, that those games are very diffrent, foremost, that you have all information in Chess in Front of you, but in Poker you have a Game of mising information:)

Just my 2Cents