Deviating from GTO

Posted by

Posted by posted in Low Stakes

Deviating from GTO

So I've been an exploitave based player for years and recently dived into the deep end of the game theory pool and started studying GTO based play. I have a hard time grasping a few things I'm hoping some of you guys could help explain to me.

Say we are raising a normal range UTG and get 3 bet by a very nitty player on the button. Say for the examples sake that we have him on a 3bet of 2% in this situation over a huge sample. According to GTO we should have a minimum defense frequency of somewhere in the area of 60%. We have our set strategy with constructed defense ranges which we can almost play robotically. The question I have is in this particular situation are we always deviating from GTO and turning ourselves exploitable by tightening up our defense range substantially? I mean I know some of the GTO talk would be that if he is 3betting so small we will profit by him never 3bet bluffing us plus we will make a lot of money with AA etc. But the thing is we could do that PLUS profit by folding each time he 3bet us except with the very top of our range. So is an exploitave based approach always alright in these spots?

Another example could be where we have a strong hand on the river say AA on 9,9,x,x,x rainbow. We raised preflop IP and bet all streets. On the river we get XR all in for a huge overbet. Our read says the player has never done something like this before and has an extremely low XR%. According to GTO we have to be defending a healthy percent here in order to not get exploited and AA is probably one of those hands. But with my exploitave experience would probably fold nearly everything except boats here. Is this kind of deviation from GTO wrong here?

I know these questions are probably very trivial and I have a good feeling for the answers myself but I still would like to hear what other players say about this.

Thanks

19 Comments

Loading 19 Comments...

BigFiszh 11 years, 11 months ago
GTO is a defense against exploits. If sb. does not exploit you (and you know it) there´s no "need" to defend (by GTO), so the best strategy obviously is to deviate and max. your EV. GTO is best to play against unknown Villains OR in unknown spots where we don´t know or are not sure what the best strategy is - so we take the defensive line and at least stay unexploitable.

Regarding your second example, given Villain is "hugely overbetting", I strongly doubt that AA is in the top of your range. ;) But say you have a hand in the top of your range, say A9, which you still think can´t call - then your question obviously is a good one - namely, how strong of a read we need to deviate from GTO, does a foggy read on the player pool ("nobody plays anything but a boat like that") do the job? As mentioned, if we have no clue, and think Villain is principally capable of doing that as a bluff - we should go with GTO, imho. Otherwise deviate. But there´s a fine line and I think most players overestimate their ability to make "correct reads" or trust their gut too much.
R0b5ter 11 years, 11 months ago
No your right AA is definitely not in the top of my range. But it could in theory still be a hands that's in my GTO calling range? Maybe right at the boarderline? Or perhaps not vs a huge overbet but yeah maybe something like A9 is a better example

But yeah anyway I have a hard time going with GTO in spots like these even vs unknowns since my poker brain just says "out all of the million hands you've played a raise like that is not a bluff often enough for you to call".
João Ferreira 11 years, 11 months ago
A bit off topic, but can you recommend me good material on how to build GTO ranges and stuff like this?

"According to GTO we should have a minimum defense frequency of somewhere in the area of 60%."

I understand that you need around 60% of folds to get instant profit on a 3-bet (with 0% equity), but that means we should defend 40% against a 3-bet to avoid this. How do we get to the 60%? We consider the equity they have in the hand as well? How do we do it?

Thanks,

João
GameTheory 11 years, 11 months ago
This question is really easy. GTO is trying to "run bluffs" with a maximal frequency in order to still stay profitable. The defense against this GTO is to call these "bluffs" down as wide as possible while remaining profitable.

(Note that one can just as easily described this the other way around, since GTO is where the two meet.)

Against someone who "bluffs" too infrequent, whether it be 3-bet preflop or on the 99xxx river, it is no longer the most profitable strategy to call down as frequent as you would against an optimal opponent.

If you play poker to win real money instead of theoretical money, don't call those nits in these spots!
BigFiszh 11 years, 11 months ago
Kinda disagree. GTO is not intending to "stay profitable", it´s goal is to maximize EV - which can simultaneously mean "minimize losses" - that can result in negative EV, it can result in 0EV, it can result in +EV. Staying profitable is not the goal of GTO. It can´t be by definition as two GTO players playing against each other can only achieve break-even (ignoring rake), agree?
GameTheory 11 years, 11 months ago
You read "stay profitable" in the wrong sense here.
I mean it in he sense that two players start with typical human strategies and that they keep adapting against each other and that both of them will reach GTO strategies.

Then during that process, both players want to maximize the amount of value they steal from the other player by bluffing, but at the same time hey don't want to bluff so often that hey are giving away value to the other player.

AFTER they both reach GTO all this figurative language becomes meaningless.
Dont confuse yourself there, a pair of GTO strategies is just a pair of strategies where neither player can unilaterally improve his value by changing his strategy. Barring the freedoms they have that don't affect the value of their strategy those are fixed strategies. They don't move, they don't have a goal, they don't "want" to maximize or minimize anything. They just are!
Michael Gazonda 11 years, 11 months ago
One of the big problems with GTO or exploitative strategies is that you need to know how your opponent sees the situation.

"Another example could be where we have a strong hand on the river say AA on 9,9,x,x,x rainbow. We raised preflop IP and bet all streets. On the river we get XR all in for a huge overbet."

Maybe this player sees all 99XXX boards the same. What if they see 998Q5 differently than 99232? They may take one line on one board, and another line on the other board. If you see these as the same spot, and your opponent doesn't, your opponent has the opportunity to be one level ahead of you no matter what strategy you're employing.
BigFiszh 11 years, 11 months ago
But this is only a problem if we use exploitative strategies, right? If we mis-estimate Villain´s ranges we´re likely to make (costly) mistakes by trying to exploit him.

Once we were able to play 100% GTO though, we just wouldn´t care about Villain´s range or what he thinks / does whatever. We´re keeping our lumps together and Villain can jump up and down or run in circles, there´s nothing he can do to increase his winrate.
Michael Gazonda 11 years, 11 months ago
It's a problem with GTO as well... if not an even bigger problem.

How can you calculate optimal frequencies for a situation if you're not seeing it properly?
R0b5ter 11 years, 11 months ago
This is exactly my problem with GTO. It's basically a bot style. Perfect math but readless. Sure starting off with trying to play GTO is probably good but as soon as you get any kind of feel for your opponent (first couple of hands really) then it's all about exploit exploit exploit. And a good opponent does the same with us until we after an infinite number of hands both end up playing GTO again. Correct?
Michael Gazonda 11 years, 11 months ago
Yeah, it assumes that the depth of awareness and thinking goes "so far", and no further. If you GTO a spot that I split into 4 different spots, and have 4 different ranges, then GTO has failed.

Not saying GTO isn't good, or that it doesn't have good points... just that it can only go so far, and has some major flaws.
BigFiszh 11 years, 11 months ago
Maybe we´re talking about different things here? What I mean is, if I play a GTO strategy there´s nothing you can do to increase your EV. You can have one range or four ranges, I´d just not care about it. You could do anything you want, as soon as you deviate from the nash - you´re losing money against me.

Now, I know that this mostly is "theory" as nobody knows the nash ... but just take the "rock paper scissors" game. If I choose any option with exactly 33% you can do what you want - you won´t win. Ever.

So, if we´re talking about the same thing - GTO can´t have any "flaws" - by definition. :) Unless you mean by flaw that we don´t maximize EV against non-optimal players. That is true, obv. but that means we have to KNOW the strategy / leaks we can exploit.
Michael Gazonda 11 years, 11 months ago
My example was used to illustrate how people "think" they're playing GTO, but they're not. No one is. Comparing poker to rock paper scissors is not useful here, because game theory would look more provable there.

I guess I'll need to make some videos on this soon...

Be the first to add a comment

Runitonce.com uses cookies to give you the best experience. Learn more about our Cookie Policy