you'd be surprised, computers on their own are quite bad at it. they need a random sees to achieve true randomness. i.e. white noise or decay times of radioactive atoms

this thread should be moved to RGT

this is going nowhere

Don't know if it's provable but pretty sure Jungle's avatar is GTO.

fyp

this thread is simply evidence to why poker is and will continue to be extremely profitable for the people well versed in the concepts

My understanding is that high level RPS is all about physical tells. The best players are really good at looking at your wrist and deducing what you're getting ready to throw from how your muscles are clenched. (Not joking)

This thread is very interesting when we start to differentiate physical(real world) and purely logical/mathematical solutions. GTO strategy for NLHE and PLO(say Heads up) seems like it can only be solved by very intense(quantum computation or super-computer at the very least) means. In the physical world it seems like there are way too many variables to even attempt to solve a game like that. Online, there are added timing tells, and personal details that could also throw things off.

Mathematically, a GTO strategy was said to be available by Nash. A lot of people here are uncertain on what that means. It just means that you play in such a way (% wise) that you could not be exploited. If an opponent tried to exploit your play it would open up an exploitation in that play. Given all the hands and possibilities, in the long run you simply could not be exploited. Given no tells.

It's funny you mention the wrist aspect...I remember live NLHE hands where wrist information was used to deduce what someone had ha. Regardless of their hidden glasses attempt . Simple situations, but with live tells it changes everything.

I have always imagined a poker player who simply didn't care in the physical world and only thought about/knew an optimal strategy. 9 or 10 handed games are just too difficult, but of course in heads up it would be hell.

I think players are of course getting closer to a GTO strategy. The exact %'s that would take place though...it might be better to round close to it lol. Like within 20%. I would love to see an exact(and huge) spreadsheet of what GTO looks like for games. Then the additions of certain tells alone, would throw it into the abyss.

Wouldn't having a tell on someone be useless if that person was playing GTO anyways? I mean all a tell does is tell you what he is doing but it is useless if he is playing GTO anyways

Not situational tells. If you knew that a person was more likely (or had) a certain hand then you could exploit a GTO strategy. If you picked up on digital or live tells then you could exploit that.

but ya, if it is solved, then someone could play a certain way and no matter what information you got off them you could not beat them. Purely % wise(not live tells). I would love to see that, but it doesn't seem possible at the moment. Someone correct me if I am wrong, but you would essentially need to play out millions(or more) of situations and realize what the overall strategy should be for your play so that no opponent can go against you. If someone tries to exploit you, they are just opening up a leak.

I don't think the strategy we'd be exploiting would be GTO in that case.

Let's say I'm playing a "GTO strategy"—or at least trying to—but I fart whenever I have TPTK against a c-bet. Well, villain c-bets and I fart, so my range is {TPTK}, and a c/c or c/r range consisting solely of TPTK (in all likelihood) isn't GTO, therefore villain's exploitations aren't taking advantage of a GTO strategy.

From what I gather, it doesn't matter on opponent's tells or plays. If you take that into account then you are playing an exploitative strategy.

I may be wrong, but I believe GTO is simply a huge cloud of plays for every hand combination and board texture.

Non imaginative professional blackjack players pretty much disprove your assertion. I think you forgot to consider those who don't fancy themselves good players but only good memorisers. Emotions or not they wouldn't dream of deviating from their memorized strategy.

Ha, ya. I think I have searched through every live table I have played and tried to imagine at least one player being unexploitable and showing no emotion. I always thought it would be possible. Know all the possible plays and be able to read live situations in such that no one could have better analyzed them. The very best player.

just take benzos then you become unreadable

You are right, a gto-strategy would not take physical tells into account at all, but that was not the point of the post you quoted.

Rei's point was that physical tells of a hypothetical "gto-player" take away the random element a gto-strategy relys on, and thus that strategy would stop being gto, even if executed mathematically correct.

Just went through this entire thread. I'm sure it made me smarter, yet I'm also sure I can't put a single sentence about GTO in this thread without getting flamed.

RPS for rollz, tho? Anyone?

Fixed.

Seriously though that's interesting. I've always crushed at RPS but I think i'm more of a "feel" player.

i wondurrrrrrrr if durrr can beat gto yo

Was there every any published work done on the possible magnitude of a GTO solution for NLHE heads up? Even if at some point computer processing was able to discover the solution, would it be compatible with humans? Could it be used without the use of a computer itself, or would you essentially just being playing through a bot anyways.

I was thinking about situations where you were able to view some of the hands that a GTO solution played out. Would using that and trying to implement it into your game even help you? Say you viewed 100 specific hands. You wouldn't be sure on the % of randomness that was truly being used or what the effect would be if you tried playing those 100 hands like that always. If you messed up on a different set of 100 hands, it seems like it doesn't matter that you viewed those 100 hands. If you aren't balanced correctly in other spots, you could just make your strategy worse by playing hands like that, right?

imagine a gto solution a game tree with every branch having probabilities assigned to actions for a certain hand. you would neither know the probabilities by watching x hands nor would you be unexploitable if you always played them the same. and the solution would clearly be massively huge, far away from possibly being memorized by a human.

Yes. It depends on how you use and try to implement it, though, and it could easily be very dangerous.

Ya. I wonder what others think on this too. I could see it go both ways in answers.

Say you can see 100,000 hands played by a GTO strategy bot in Heads Up NLHE.

Would it better your game? What could really be taken away from it that you couldn't already of got without it or that you didn't already know. It seems possible that if you try and used those hands to work on getting your strategy closer to a GTO strategy or better than your current play...you might just make your style even worse.

Would there be any number of finite hands to view that is believed to help your game out and at what number would it start to be useful then?

@franxic Ya, it doesn't seem like a human would come close to being able to execute a possible GTO solution. Having to deal with all the randomization %'s alone would be a huge magnitude of processing and and it couldn't really be used in the live arena.

Yeah, that's definitely possible. That'd probably be the most likely result for someone who doesn't know what they're doing. You'd have to do the heavy lifting: making sense of everything, and making sure you aren't applying things stupidly. I can think of at least a couple dangerous traps.

Hmm. In a 100k-hand sample, you'll get about 50k hands of opening pre, maybe even more if your GTO bot loves to grim people, which is probably enough to draw lots of usable (but not perfect) inferences about optimal play there, at least at some range of stack depths. Other spots (most postflop spots, given the sheer number of run-outs) would probably be less helpful, I imagine.

But even the preflop stuff could be dangerous. One of the traps I was thinking of is that semi-GTO preflop play, even if it's not a total bitch to apply, might be suboptimal w/o being paired with a certain non-intuitive, intricately choreographed game plan on later streets.

But I'm just speculating, lol. I'd like someone more knowledgeable about this than me to chime in.

I realize that these videos would be better suited for a theory post, but seeing as it was posted here I will go ahead and respond here. I have went over this and have some questions for those more poker savvy than myself. When he gets back to the original no limit hold em river decision, how does he get the varying percentages for his river distributions(each hand has a different equity on a scale)? Did he utilize poker stove for each hand individually or did he input a range in its entirety? Thanks.

Cj