No - not the same guy. The one I'm thinking of has much more of a "grad student" look about him and less of a "high school principal" look about him.

but it is also about strategies, some of them must be GTO

More LTO, I think.

(Life Theory Optimal - and that's giving him WAY too much credit, imo!)

If you play the game-theoretically optimal strategy, you are *not* guaranteed to break even. It depends on the game. For example, in heads-up limit poker, if I have the big blind every hand and you have the button, I will lose money to you in the long run (even if I play the optimal strategy for the big blind, assuming you play a strategy that maximally exploits me).

If we instead say that a game of poker is a *pair* of hands - i.e., I play the button once and the big blind once - then by playing optimally I guarantee that I have non-negative EV against any opponent.

When people talk about optimal play guaranteeing that you at least break even in poker, they are tacitly assuming some set up that gives you an equal number of hands on the button as in the big blind.

what is a "dominated strategy decision" and how does GTO play make money?

I feel like if you play "optimally" balanced on the flop/turn etc you must make money?

And on the river if you bet a range so that they are indifferent, you still have all the pot equity etc??

In a rake free environment, GTO will make money against anyone who doesn't know the GTO counter-strategy, but the point has been made again and again that GTO will often make less than an exploitative strategy targeting a non-GTO strategy or player type.

Dominated decisions are those for which no indifference point exists (i.e., one action has the greatest EV and there are no good reasons to ever choose a different option). For example, there are spots where your range is so strong that you can never bluff often enough to make it correct for your opponent to do anything but fold.

That's my understanding of it, anyway.

if a guy always 3bet shoves 72o, GTO play is +EV vs that, obviously less than exploitative but...

it's not always 0 ev..

Okay, that's also true but somewhat different than a dominated strategy decision. It's more or less just a side effect of the opponent's suboptimal range in that spot. To turn it around:

If a guy only ever 3-bet shoves AA, a GTO bot will call a wider range than an exploitative human player would, but the bot's overall expectation versus the opponent doesn't become -EV just because it makes no adjustment to the opponent's suboptimal 3-bet shoving range. Nevertheless, if we pick out an individual hand that the bot calls off with and compare it to a range of {AA}, we might find that the play had a negative expectation when examined in isolation versus that specific, suboptimal range.

I was thinking about this topic a little more today in terms of real world examples.

One place you can see this tension between GTO and exploitative lines is in the NLH HU SnG endgame. Most regulars have a blind depth below which they start just playing the Nash tables from the SB. But you will find that many of the top players place the bar very low, at say 7BB (versus say the 12 or even 15BB cutoff that others might use), and that they play a mixed, exploitative style versus most opponents until that blind depth. One of the HUSNG.com guys dubbed this the ROFL style, and most people I met while studying the game felt this was critical to grasp since the swings during the endgame can be so brutal. Further, Nash from the SB is actually -EV at certain blind depths (the overall strategy being 0 EV if you also get to play the BB).

Similarly, almost everyone realizes that using the Nash table from the BB is only optimal in absence of reads or when you suspect your opponent is shoving Nash. Many common opponent types are not shoving Nash (they are usually quite a bit tighter), however, so arbitrarily calling Nash leads to an unnecessary increase in endgame swings.

A more recent example is the skillbet.com format, which essentially creates a situation where maximizing immediate EV is always appropriate and it would be a mistake to make any balancing plays at all. Obv this is a completely different type of game, but if you play a few of these, you get a really concrete sense for the value of exploitative play as well as how widely different it can be from a more balanced, GTO-like approach.

As for HUHU specifically, just think of any really bad, easily beatable opponent type. Would you really choose to stubbornly adhere to a balanced flop c/r'ing strategy when you know your opponent folds to such raises 75% of the time? Do you absolutely need to know what the GTO bluffing frequency would be in order to make a decision about how to best exploit this behavior? Obv this is a very simplistic example and assumes a very poor opponent, but my own experience is that most in-game decisions against human opponents aren't really that much more difficult. I nevertheless would agree with anyone who might argue that a certain amount of balance is desirable if not necessary in all but the most extreme cases.

The concept of GTO is only relevant against very competent opponents imo. What difference does it make what you should do against someone who is capable of deducing through astronomical iterations of potential lines and counter lines each time choosing the one that incrementally improves on the last, by some faculty be it machine or organic, if the someone you're playing against peels limped pots 85% of the time?

At the lower stakes GTO is just a well performing strategy among other well performing strategies.

This is a bit misleading. The so-called Nash tables are just the GTO play in the restricted game where shoving and folding are the only allowed actions (which I'm sure you already know). This is different from the Nash equilibrium aka GTO strategy in the full game, which means that choosing to not play shove/fold until very small stacks (my bar is at 5 BB and I'm not a "top player"...) is not necessarily an attempt to play exploitively but could be just trying to play closer to GTO in the full game.

That's a very good point ofc. I know Mersenneary presented ROFL (Raise/Open-shove/Fold/Limp) as an exploitative approach in his videos, but you're obv correct to point out that the Nash tables are not a solution for the full game and that the use of a mixed strategy at lower blind depths doesn't in-and-of-itself indicate that a player is not trying to approximate GTO.