In LLSNL, without reads, one could simply adjust by playing in a very unbalanced way of playing too tight.

Playing exploitively is ONLY bad if someone is exploiting you.

I admit, I'm not read-up on exactly what GTO means, but it sounds remeniscient of something I read in Miller's "Playing the Player"

He talks about balanced ranges and playing "optimally". Basically he defined "optimal" as playing a range of good and bad hands that is balanced in such a way that you are virtually unreadable. He then warns that playing this perfectly balanced strategy would inevitably require you to run huge bluffs into calling stations, or make thin call-downs against an aggressive nit. But if you deviate from the "optimal" strategy and avoid those situations, you'll dramatically increase your win rate.

He doesn't need to learn GTO to avoid mistakes like these. In fact, he'll probably be more likely to make them under the guise of "I'm balancing my range."

Balancing your range is part of a GTO strategy. It's not necessary (or possible) to learn a perfect GTO strategy for ring NLHE games. Even HU NLHE is not solved (or bots would be able to beat the best players). That said, learning balance and GTO concepts will certainly help your game by allowing you to prevent regs from exploiting you, and allowing you to recognize exploitable leaks in your opponents' play. I'd highly recommend Matthew Janda's Applications of No Limit Hold'em (haven't read Miller's book).

I have mixed feelings about bumping a 6-year old thread. (Actually was considering curating a list of links to these threads for LLSNL.) But this was worth replying to.

1. I think this whole thread was just before Miller's The Course came out. Ed writes so much about exploiting $2-5 nits in that book that I think that would now be the best recommendation for someone in a similar spot to this 2015 OP. More tailored to this sort of game than Poker's 1%.

2.

I agree that GTO will theoretically put your opponents into "tough spots." If you could somehow play like a GTO bot, and Doug Polk showed up in your $1-3 or $2-5 game, you'd still be putting him in tough spots in HU or maybe 3-way pots.

The opportunity cost is that your $1-3 or $2-5 opponents are already making tough spots for themselves by misplaying not-very-tough spots. The classic: very few players bluff enough on the river (btw many those who do are overbluffing). Maybe they should bet 30% of bad hands if checked to but instead they bet 5%. When they bet big OTR, you should fold all your bluff catchers -- the exploitative fold.

So... GTO says you put them in "tough spots" by calling at the correct frequency so that they're indifferent to betting or giving up. But they don't care. They've created a "tough spot" by hemorrhaging EV, even though psychologically it's very easy to not bluff because you never feel shame about getting caught.

When they bet big, you make the exploitative fold. Are you putting them in an "easy spot"? Couldn't they print money by bluffing you all the time? Of course. But who cares if they don't exploit it?

GTO for poker is GIGO because there is no earthly way to craft a decision matrix (the basic tool that game theory uses to optimize decisions) when you don't know the input parameters. And unless you've been sitting in your opponent's lap for the last 100 or so hours--and seeing all the hands he's played and folded, and how--you have no real idea of those parameters, only guesses. And guesses based on incomplete information pretty much by definition cannot produce optimal outcomes, except by accident.

I hereby submit to being burned at the stake for my heresy.

GTO literally doesn't care at all about how your opponent's actually play.

Playing this year I have heard a lot of 2/5 regs talking about GTO play. It's common enough now I think if you want to play regularly you should understand the idea and what people trying to use GTO play are doing. More about how they are doing it wrong them actually playing GTO.

GTO is not the maximum EV way to play poker. If your goal is to make as much money as possible playing poker you should not be trying to play GTO. Some elements of GTO can be useful when adjusting to a table that contains better then average players. If you have the option switching to a table with worse players will be more profitable.

Agree with all of this... but if the 2-5 regs are using the term GTO at the table, I'd wonder if they have the self-discipline to play GTO or anything close.

In principle (not to maximize win) I believe in helping anyone who wants to learn, but not at the table. Sometimes I slip up and can't help talking strategy among people I know, honestly.

If you want to chill at 1/2 2/5 relax have a good time and win some just do what you're doing.

If you want to move up, "study GTO" starting now.

Jarretman has a good post on this but exploitative play only exists as a relative concept to GTO. Without knowing the GTO actions theres no way you can play exploitatively. There will be someone saying "nahhh man the players at 1/2 are so bad you just have to not bluff them and play nitty and value bet thin a tonnn" well thats basically saying deviate from GTO and underbluff/overfold etc.

As someone said above when you sit yourself at a 10/20 game (or 200NL+ online) and its regville you will be destroyed without further study.

The first sentence is true; the second is completely wrong. Most of the frequency errors made by small stakes players are so egregious and so predictable that you don't really need to understand GTO or the 70% rule or anything that sophisticated. You just need to notice that, for example, when they bet big, they usually have it.

It is indeed good to have an understanding what GTO is so that you understand theoretically why overfolding is correct. But you don't need it to beat these games.

(BTW underbluffing at $1-3 just losing money to opportunity cost in many spots nowadays. You should bluff quite a bit.)

No doubt, but this is "Discussion of up to 3/5 live no-limit..."; there's another forum for 10-20 NL hands.

It's great to discuss GTO about these small stakes hands as a sort of boundary stone, for reasons you've laid out. I just think the emphasis gets way out of whack because game theory is fun to think about and betting a lot of the time until they raise feels like getting away with something unsustainable that will catch up to you.

By all means learn both, talk about both, but make it clear that ruthless exploitation makes the most money in small stakes.

I agree with you just the way we articulate it is different.
I'd say folding a hand pure that is an over 50% calling frequency because "when they bet big they have it' is actually already a deviation from GTO just that the said hero doesn't know it. I'm not saying you need to be a robot to play exploitatively.

I talked about 10/20 since OP claimed to wanting to move up in stakes as fast as possible.

I dont play in the 1/2 live games so didn't know overbluffing was better

Thanks for the reply.

In LLSNL, playing GTO doesn't put your opponents in a tough spot in most cases.

Take these couple examples:

Player A suffers from "I am not here to fold" syndrome.

-Any time you are "balancing your range" by bluffing, you are losing value.
-Any time you are "balancing your sizing" by betting less than maximum, you are losing value.

Player B suffers from "GIamasupernitG" syndrome.
-Any time you are "balancing your range" by calling with bottom half of your range, you are value owning yourself.

You are actually making it easier every time you are not exploiting them because you are "balancing."

Can’t believe this was over five years ago…

It depends what your goals are. Are you planning to only ever play 2-5 live or lower? Then no, you don't need to be running sims to beat the game (in fact the highest EV thing you can do at LLSNL is aggressively table change but who wants to be that douchebag).

Everything is relative and it depends. Do you need to learn GTO? No. If you are planning to be/are a professional poker player should you learn GTO play? Yes. Probably should've started 5 years ago.

Recall that six years ago, GTO was a very popular buzzword, but there wasn't even any good evidence that what was being called GTO actually was.

Now there are multiple sim engines using different learning algorithms, and the fact that they have largely converged on the same strat is a strong argument that they've found something close to GTO, though it's still not proven.

To update my opinion from then, I still wouldn't want to play GTO unless I thought I was being exploited, but there is much more value now in knowing what the GTO base is in order to recognize your opponents' deviations from it. Most importantly, though, you need to learn how to exploit those deviations, which is generally not by playing GTO yourself, but by unbalancing your strat in a counter to the way that their strat is unbalanced, thus taking maximum advantage.

I don't think any of these solvers have the computing power to solve the entire game, though. My understanding is that you can use them to solve sub-games where you have a feel for your opponent's range before the sub-game starts. They also almost exclusively are used for heads-up situations (bots that can crush humans only play heads-up) and so all of the caveats that apply when you move from heads-up to multiway still apply.

In short, there is no proof that solvers are learning GTO because there is no reason to think they are. Solving a sub-game is very different from solving the entire game or even proving that the sub-game you are solving ever occurs in a true equilibrium.

I agree from a theoretical standpoint, but I think that the evidence that has built up for successful sub-games at least hints that people learning "solver strat" are learning something acceptable within a GTO strat, if not actual game-wide equilibrium.

No, it means that people learning solver strat are learning a strategy that will beat humans, because most humans are not smart enough to pick apart what the solver is doing. It's still not necessarily to do with actual GTO.

There is a HUGE gap between how good the best humans are capable of playing, and truly "perfect" play (if that can even be said to exist in poker, for multiway reasons). The same holds in chess, where computers have been able to beat humans for about 25 years but there are still computers being made that can beat the previous computers. Beating humans is still very, very far from solving the game.

I believe the solver is closely approximating a gto solution to that particular subgame. For the reasons you state, that doesn't cover the universe of all plausible subgames, and if you set it up with faulty assumptions then you're just solving a situation that never actually exists. I wouldn't say they have nothing to do with GTO; I'd say they find GTO solutions to problems that almost certainly aren't themselves GT equilibrium situations to begin with.

Im not big into solvers though so please show me where I'm wrong.

As has been said over and over if that's the case, it's not real GTO. GTO is the same (if solution is unique) across multiple situations. It maybe 'solving' optimal plays for that situation, but it's not GTO.

They're finding solutions, but it's not GTO.

Simply put: which solver would be more popular.

1. One that helps you win more.

2. One that is closer to GTO.

"solvers don't actually know what GTO is and are essentially just EV maximizing calculators"

yes this is right. afaik there is no guarantee that gto even exists in multiway nlhe, so solvers are certainly not trying to work within a full game solution. once the hand goes heads up it becomes solvable (given the players ranges at that point) and as you said the solver would approximate the gto solution to that subgame.

This was discussed a ton in the old game theory COTM. GTO certainly does exist multiway--but there is no guarantee that it behaves the same as your intuition from heads-up games would tell you it does. The most relevant difference is that you can play GTO in a multiway game and someone else's deviation(s) from the NE can cause you to lose EV even if you are sticking to GTO. So by any reasonable definition you can no longer call GTO "unexploitable".