Thanks for the info.

Trying to apply game theory to behavior and future oppurtunity value has become more of an interest to me. Both of which the current iteration of solvers don't take into account. I just think too many players have become too fixated on the current iteration of solvers. Anyone have an opinion on this?

as long as no one plays close to gto u can never say phils game is not winning in his live sitngo element and first sentence is 100% true. Second, phils lines are so unconventional, evan the best players have no clue what gto is in that spot when he makes his weird plays, no one evan ran that into a solver so its imposible evan for linus to play gto when phil takes a wonky line.

what players are calling gto is a bad painting copy of a famous masterpiece. gto is just a fancy term applied in poker so pros feel better about their winnings, no, all pros money came from casuals who play 50% of their range vs 30% of pros range in position. and just becouse some regs play each other meanwhile and one wins more doesnt mean he won because he plays better gto.

to me gto its like teaching a 16 yrs old quantum physics before an exam when the exam is about basic physics and expekting to have better results than a normal student.

GTO doesn't change based on your opponents play.

You guys will waste so much time arguing about **** that doesn't matter to yourself its amazing.

Read again, he has a valid point. Daniel is so green in GTO thinking that he may have the solves for basic stuff but surely not for all the goofy stuff Hellmuth is doing.

If Daniel has "the solves" for "basic stuff" but not the "goofy stuff" then he is not playing "GTO" against the "basic stuff." If you vary your play based on your opponent doing "basic stuff" vs."goofy stuff" you are playing exploitatively and not GTO. This is not a bad thing. GTO is only the optimal strategy against an opponent who is also playing a perfect GTO strategy (which no one is).

A better way to phrase it might be that "basic stuff" does not exploit the the non GTO aspects of Daniel's strategy as effectively as the "goofy stuff" PH is doing.

+

A GTO strategy is maximally exploitative of the opponent's strategy when the opponent is also playing GTO. In this situation, neither player can increase their EV by changing their strategy. The GTO strategy is the best strategy, period.

Now suppose the opponent starts playing a bit "goofy." We know by the fact above he has lowered his EV versus the GTO strategy. By playing goofy, now, he is making a mistake losing EV.

For Hero, when the opponent is not playing GTO, some other strategy can be +EV compared to hero's original GTO strategy. Hence node locking in solver software. When you know something about your opponent's strategy, you can calculate the maximally exploitative "best response" strategy and this may be different than the original strategy you had before you knew that information about your opponent.

However, the GTO strategy is still +EV against the "goofy strategy" even though it may not be the best response to the goofy strategy. The only player losing is the one playing the "goofy strategy."

Yes, and I think the approach I have in mind would be a Bayesian one. And I think you could create strong bots this way with enough computing power.


Basically, my idea at present is that you'd assume your opponent is playing some set of strategies, one of which is GTO and the others are like the "tilting" strategy and "tightening up close to the money" strategy and so on.


You'd start at hand 0 that the opponent is playing GTO and you'd play GTO until you have enough evidence to start mixing in maximally exploitative counter strategies that take advantage of the information you have about your opponent. You can mix in these other non GTO strategies at some low rate so you don't expose yourself to too much EV loss if you are wrong and your opponent is playing GTO, but you should be able to gauge when the opponent is playing far from GTO.

The issue here is that playing this mix of strategies yourself opens you up to being exploited so you'd want to make sure the EV gain you get from deviating is really worth it. For some spots, you might only need a small sample size to be highly confident your opponent spews on the river when an A hits, for a simple example. In a multi-player game in practice though this sounds hard and playing GTO might be the best solution, but that's that's I've been thinking about at least

my point is for example when u only know the gto strat vs a 3x chk raise when someone is gonna chk raise u 5x u wont play gto vs that if u never chkd a solution for a weird chk raise like that because the gto frequencies change, and this is just one example, so than u just start guessing what the correct play is, this is same vs other weird plays like a weird bet size that u never checked the gto solution for etc.

Yea, I think that's fair. However, knowing your strategy in the 3x game will make you much better at estimating the 5x solution on the fly during the game.

thats just a guess, maybe someone like hellmuth is more used with his goofy lines than someone like negreanu, nobody knows this

Nah, you took as much time to write this post as others spend talking about GTO play. And the fact that it doesn't matter is laughable. Understanding the concept of GTO will always be helpful to a poker player of any level.

* * * * *

I think I get what the "GTO doubters" are trying to say, though. (I'm putting that in quotes because I know it's inaccurate but I don't know what else to call them.) While true GTO play can't lose, there will be some specific situations in which it's also not the most profitable strategy.

The toy example I like to think of is a game of rock, paper, scissors. Suppose you and an opponent play this simple game for $10 per throw. The GTO strategy would be to employ the three options completely randomly (i.e. with no discernible patterns) and such that the long-run distribution is exactly one-third rock, one-third paper and one-third scissors. If you could somehow master that, you could never be beaten. Sure, you would break even against someone else who also does it, but you would never encounter a negative-EV game.

Now consider a hypothetical opponent who throws rock 45 percent of the time, scissors 45 percent of the time, and paper just 10 percent of the time. He isn't aware of his distribution, nor is he savvy enough to notice when someone adjusts to him. He just keeps going about his way, handing $10 bill after $10 bill as long as someone wants to play. (I can only assume such a player is named Kah Ching.)

Against such an opponent, you become more profitable by abandoning the true, third/third/third GTO strategy and throwing rock much more often.

I'll pause here... Variance51, is this what you're getting at?

The problem with this example is that it's a highly stylized situation, one that almost cannot exist in real life. For starters, the whole 45/45/10 idea is laughably extreme. On top of that, even the densest, thickest opponent would eventually realize he's getting milked dry. He would either quit the game, or would realize sooner or later just how often you're throwing rock, then adjust accordingly. If/when that happens, your only foolproof recourse is to return to your GTO strategy.

Poker is obviously much more complicated than rock, paper, scissors, but the basic concept still applies. Yes, there are always certain players with tendencies you can exploit (e.g. too tight, too loose, bluffs too often, doesn't bluff enough, and so on.). But the problem is that they will adjust as you continue to take advantage. And even if this one opponent doesn't adjust, at a non-heads-up table – which is a far more common situation – other players will be able to exploit you.

Thus, you're always better off getting as close to GTO as possible as often as possible: reaching it means you can never be beaten. You would be +EV against all players except for the few who have somehow hit that same GTO benchmark.

Besides, you have to first master GTO play before you can truly know how/when to best depart from it in certain situations – much in the same way you blackjack card counters have to first master basic strategy.

Anyway, I would never claim to be a GTO expert, so I'll let others pick apart my post here. Helps me learn it better myself!

Your argument for Phil is that he's taking his opponents into a different game and that neither of them knows GTO in that game but Phil has more experience in that game, so he'll do better. It's true, based on what I know about GTO and math, that if Phil includes something like a donk min bet range on the turn, the opponent cannot "just treat it like a check" and expect the same EV with their strategy as if Phil had checked. We also know from game theory that adding any strategic option cannot decrease your EV. So simply having the option to donk min bet the turn is at least as good if not better than not having that option, so I agree with you there. Theoretically, the solution to NLH has to consider all bet sizes which is very hard. That's why we model the game by simplifying the strategic options like bet sizes when running solutions in practice.

This problem is not that hard to solve for the GTO player v Phil though.They just modify the set up of the game to include a few more bet sizes and then they're done. You can be smart about how you group your opponents actions (like bet sizes) and not give up that much EV compared to not grouping them at all. And then Phil is crushed again. It can still be true that Phil is a tougher opponent because he uses more options than other players. There is no way for the non GTO player to win this tit for tat, back and forth, arms race. The other issue is that the non GTO player has to respond to the GTO player's actions, too, and they're likely making big mistakes there to support their "goofy" strategy. It's not just a simple river call situation where we just have to respond to some weird bet size we never modeled.

And one more point that kinda supports Iatchi even though I'm on the other side of the argument in a way is the Doug v Daniel match.

There was a reason Doug and team had to run new solutions when Daniel introduced new bet sizes. However, as also mentioned, having existing solutions with the other bet sizes made it easier for their models to estimate, say, the optimal preflop ranges for the new sizes.

Doug mentions this in the post match interview when he's thanking the team.

I think many people here are ignoring one key aspect from live poker, reads, and yes playing live can give u this gut feeling over someones hand better than playing someone behind a pc, to what degree? this is open for debate, phill says all his plays are based on this "White magic". imo this is where phils winrate comes from in this format hes doing good at. btw i think phill would have close to no chance vs negreanu if they would play online same format as doug negreanu match

master it, is a big exageration, there is no way humans can evan get close to gto strategy without aditional help. in the end gto is mega simple at its core if u had time, if someone bets on river 100% pot bet u have to defend 50% of the time so that he doesnt make profit with his bluffs or value, now if u had 1h to think about ur range and hand select each range and best combos to defend u could always make a very close to correct decision, complexity starts when u realize all this range has to be perfectly balanced from flop turn until this point across multiple game trees to have best range for defending 50% of the time vs his river bet. I dont think anyone plays evan close to any form of pure gto and most big winners are biggest exploitive players in more subtle ways some without evan knowing it.

I think I saw match #3 is middle of June.

where?

I can't find the tweet that I saw. Probably Poker Go again.

Did he insinuate that new york is part of new england?

Daniel Negreanu
22 May

It’s all fun and games now… but the BATTLE resumes June 23rd @PokerGO
#HighStakesDuel

Ya´ll understand that the chance of one of the two going 2-0 in two sng´s was 50%?

Surprised no one posted about their third match

GOAT gonna GOAT. Not much more to say.