Out Now
×

The Golden Rule of Game Theory (Applied)

Posted by

Posted by posted in Mid Stakes

The Golden Rule of Game Theory (Applied)

As I understand it, the Golden Rule of Game Theory is as follows:

When deciding how to play an individual combination, we always take the line which maximizes our EV.


I'm having a very hard time understanding how this relates to certain strategies that rely on information hiding, though.

For example, the common consensus among good players is that limping the Small Blind after it is folded to you in six-max is a very good idea when your opponent is playing close to how they "should" be.  The argument goes as follows:

1)  We are getting immediate odds of the following to call:  .5BB/2BB

2)  We need to 25% equity to make the immediate call profitable (assuming this equation holds, which it doe not, but the general concept seems valid).  Thus, we should be calling very wide.

3)  If we only call our weakest hands, though, then the Big Blind will raise quite wide and we will be forced to often fold or play out of position with a very weak range.

4)  To counter, we limp all of our continuing range.  This hides our information, protects our weakest holdings, and so on.

However, this seems to violate the Golden Rule:

For example, AA is always worth at least + 1.5 BB's to shove AA pre-flop, and so anything we do with AA pre-flop must be greater than or equal to 1.5 BB's in EV.  This should imply that we never limp AA pre-flop, since Villain can simply improve their expected value against our Aces by checking any two cards.

Is the idea that they are somehow forced to raise a certain percentage when we limp, or else we have improved the expected value of our weakest holdings by limping? 

This would counter the "AA" example in a rather hand-waving manner, but it seems it could be valid.


64 Comments

Loading 64 Comments...

arizonabay 10 years, 8 months ago

Since no one else has responded and I have been waiting all day for someone to, I will Take a shot at it. This will be just a guess. I could be way off base.

OK, so if we limp only our weak hands and raise our strong hands, it will incentivize BB to attack our limps aggressively and play fairly tight vs our raises. This would thus incentivize us to start limping our strong hands as well because now raising them is not as profitable as before and since BB is attacking our limps so aggressively the EV of limping AA should be greater than the EV or raising AA because BB is playing so tight vs our opens. Once we reach an equilibrium (of sorts) where BB knows we limp our entire range he should attack us less aggressively thus raising the EV of our weaker hands but reducing the EV of AA (since he is not raising as often) but the EV of limping AA should still be greater than the EV of open raising it because once we start open raising our strong hands and limping our weaker ones, we go back to where we started (BB playing tight and attacking our weak limps).

That's how I see it but I don't know if it is correct or not, and I have yet to develop this part of my game (I just open raise and do not limp). I just think it would be very hard to have a mixed strategy here and we should either limp our entire range or open our entire range. Again this is just my 2 cents and my 2 cents (after currency exchange) is not worth near as much as other posters.

edit: So since we have decided to have a limping range - limping w/ AA is the line that maximizes the EV of AA, doing anything else would allow villain to play better (or at least make our range more transparent and thus allow him to make easier decisions).


Rapha Nogueira 10 years, 8 months ago

"When deciding how to play an individual combination, we always take the line which maximizes our EV."

Just change it to "we always take the line which we think that maximizes our EV. This is the principle of rationality. It is an assumption to every single solution of a game. 


themightyjim 10 years, 8 months ago

the answer is shania.

arizonabay 10 years, 8 months ago

Isn't shania taking a line w/ an individual hand that is actually lower EV than a different line but by doing so increases the overall EV of our entire range?

It has been my understanding that in Game Theory there is no shania because we always take the line that has the highest EV for each and every combonation.

Rapha Nogueira 10 years, 8 months ago

Playing a pure strategy does not hide information, it gives away too much information, actually. Phil discussed it on his theory series, thinking out loud. 

To analyze EV in a proper manner we should make it from the whole hand and not from one street, which is kinda more complicated to explain without taking a lot of time. 

arizonabay 10 years, 8 months ago

I'm not sure I follow you - I do on the EV being over the course of the entire hand and not just one street - but how does always open limping (or open raising)the SB give away more info than sometimes limping, sometimes raising?

Rapha Nogueira 10 years, 8 months ago

If you limp 100% of the hands when is folded to you he already know that you are going to limp so he can adjust his strategy to a pure strategy that is common knowledge. It makes his life way easier. It is similar to one or two bet sizings discussion in HU. 

You don't give information about which part of your range you are but you give information about your strategy, which is what define ranges. 

arizonabay 10 years, 8 months ago

Im not sure what he can do about it - he is the BB and has to act after we do....so what advantage does he gain by knowing that when we play a hand it will be by limping? I am still not following. I am not advocating limping a 100% range (in fact I am not even advocating limping period) but lets say we have decided to play 60% of hands - how does the BB gain any advantage if we a.) open every one of those 60% or b.) limp all 60%? What do you mean the BB knows if it is folded to us that we will limp and so he can use a pure strategy? What pure strategy is that? I think somewhere we are getting our wires crossed. Because I think it is obvious that if we say open top 20% and limp the 40% below that - BB would have a much easier time playing than if we just raised the entire 60% or just limped the entire 60%

edit: Any confusion is probably on my end.


Rapha Nogueira 10 years, 8 months ago

It is much easier to play against pure strategies (you just play one strategy 100% of the time) than mixed strategies. 

If you are limping 100% of time you are playing your one street game as a linear preflop decision (you can understand the top 60% as your playable range so it gets to 100% of your range). Without more information I can not say that the BB playing a pure strategy in response preflop is good or not but I think it is not. 

The advantage for BB if you limp is the equity realisation and against raise is the EV increase on 3bets. If you play a pure strategy he can hand over hand realise how you decide your limp/call, limp/fold, limp/3b ranges without giving you a lot of information. 

Suppose that you decide to limp top 60%. If you don't mix (and merge it) your limp/call-limp/3b ranges with sometime polarized and linear ranges your pure strategy becomes an easy decision for BB. 

After a long day I don't know if I am clear here, but feel free to ask. 

arizonabay 10 years, 8 months ago

I want to be clear - I have not studied open limping the sb, but there are plenty of players (really, really good ones) that talk about it and I really don't see how the BB can play a pure strategy because he knows you are going to limp. So what. (BTW - letting the BB realize his equity is partly why I don't open limp at all from the sb - the other reasons are I haven't studied it at all. But Sean Lefort made 2 videos about it - and he is way, way better than me) but what pure strategy can he take? But the only reason I am even talking about it at all was to try to answer OP question which I think is very interesting and was hoping some of the game theory guys would show up when this thread started getting active.


arizonabay 10 years, 8 months ago

I still say BB has an easier decision if you just limp weak hands and raise strong hands than if you were to only do one or the other w/ all hands.

Rapha Nogueira 10 years, 8 months ago

BB does not play a pure strategy on this situation, or at least should not play if SB is playing a pure strategy, in general. I think I never said that or mean it. 

arizonabay 10 years, 8 months ago

yeah, thats where we were getting crossed up because you said 

"If you limp 100% of the hands when is folded to you he already know that you are going to limp so he can adjust his strategy to a pure strategy that is common knowledge" 

But you meant something else - I will comeback to this after I have had some sleep.

AF3 10 years, 8 months ago

Playing a pure strategy does not hide information, it gives away too
much information, actually. 

As a logical statement, this is obviously false.  Take the example raising first in pre-flop. 

I think that what you're talking about is splitting your range.  It's certainly possible to split your range by playing a pure strategy, but it's not guaranteed (see the example above). 


themightyjim: the answer is shania.

I've heard this phrase before, but I don't know what it means.  When I did an internet search for "shania", I got like a million pages of Shania Twain. 



Rapha Nogueira 10 years, 8 months ago

If you present me the logical explanation to say it is false I would be more than happy to hear. If you see poker as a game of strategy, having a pure strategy is in general worse than having a mixed one. If you see poker as a game of ranges (I always 3b AA vs EP for example) the talk is different. I prefer to think the game as strategies, it makes easier to use game theory on it. 

I don't think I am talking about splitting ranges because it is probably one of the things I try to avoid doing. But can be a bad choice of words and I need some rest to coming WCOOP. Brb :D

arizonabay 10 years, 8 months ago

Poker has plenty of pure strategies....betting the nuts IP on the river is a pure strategy (right?) or am I mixing up terms (a dominated strategy = pure strategy?)

AF3 10 years, 8 months ago

I just looked it up and found it on another poker site.  It seems like total horse-shit. 

The concept they're talking about is quite different from balance as we think of it today, and the first thing that Mathematics of Poker does is argue that "shania" does not exist.  Didn't Sauce make a video showing the concept of "Shania" as they refer to it is actually not possible? 

arizonabay 10 years, 8 months ago

I have always understood that it does not exist in a game theory sense at all - because adding a negative ev hand to your range doesn't work. You would increase your ev by removing that hand.

themightyjim 10 years, 8 months ago
I think shania as a concept was explained very poorly in it's inception.  I think as a theorem or idea it has a lot in common with range balancing and eventually (long down the logic line) GTO play.  But you're correct in criticizing it's initial hypothesizing, because it wasn't properly laid out.

The idea of shania is that we should make certain decisions with portions of our range that may not be the most +EV decision in a vacuum, but add to the EV of our entire range.  IE there are some hands that as UTG raises in 6max NLHE are basically breakeven. But having them as part of our range adds equity to our other hands by allowing us to value bet and bluff certain board textures that would otherwise be excluded by our range.  More importantly they allow us to meet minimum defense frequencies so we aren't just exploitatively playing too tight.

I think shania as a concept has a lot to do with OP.  I think it is a theorem that is a baby step towards the much larger understanding of GTO poker and range balancing that we frequently discuss today.

I think there are merits to going back and looking at the discussions that led us to our current understanding of the game.


oblioo 10 years, 8 months ago

I haven't read the other replies yet but my short answer is that the assumptions you are making in your question may be wrong. For example, even in a vacuum I'm not sure if open raising AA from SB is higher EV than limping. Even if villain checks back pre with air that he'd fold to a raise, sometimes we end up making more money by either letting him catch up a little bit or bluff, plus we often get to play a bigger pot by limp/3betting vs some hands that would just call a raise but would raise/call vs a limp. Obviously there are arguments for raising as well, but there's a lot to consider. 

Edit: This plus the adjustment points that arizonabay mentioned in his first post. Also your specific example assumes that you want to open limp SB, which you may or may not want to do. Basically you haven't demonstrated the "contradiction" that you say you are confused about.

On the other hand, clearly there are sometimes specific 0 EV lines we should take over other 0 EV lines in order to maximize the EV of our overall strategy. If this is the case, then are there really no -.001bb EV lines which do better for our overall EV than a 0 EV line with the same hand? I'm not sure... 





themightyjim 10 years, 8 months ago

I'm not sure that most of you guys know what GTO means.  It does not mean making a play that is most +EV in a vacuum given available information.  It is a strategy that would be (if actually solved which based on current computational power is essentially impossible) unexploitable.  That does not mean that it is the most profitable strategy, it just means that no matter what strategy you would choose against it the best you could do would be to break even.  And if you make any mistakes in range construction, essentially you do not play a GTO strategy in response, you will be giving up some EV vs a GTO playing opponent's range. 

You guys are confusing optimal exploitation vs GTO.  Assuming that your opponent does not play GTO (which is a safe assumption considering no one really knows what a GTO strategy would be) you should shoot for a strategy that maximally exploits your opponents tendencies.  The closer that opponent is to playing a GTO strategy the closer your maximally exploitative strategy will be to a GTO strategy.  It's a ying and yang give and take situation.  

edit: and if you don't really understand your opponents strategy (ie you don't have enough or have been unable to synthesize available information into a cogent exploitative counter strategy) the advantage of attempting to play a GTO strategy is that it protects you from exploitation in situations where the inherent rules of the game stack the odds against you (playing OOP like in the SB for example).


oblioo 10 years, 8 months ago

Sorry but I'm not sure how this is relevant or why you think it goes against what has been said ITT. First of all, you wrote: "And if you make any mistakes in range construction, essentially you do not play a GTO strategy in response, you will be giving up some EV vs a GTO playing opponent's range."

This is not always true; it depends on the situation. To use a cliched analogy, if you always throw rock in RPS you will still break even vs. a GTO strategy even though you are making a mistake in range construction.

More relevantly, you wrote in a previous post: "IE there are some hands that as UTG raises in 6max NLHE are basically breakeven. But having them as part of our range adds equity to our other hands by allowing us to value bet and bluff certain board textures that would otherwise be excluded by our range."

Notice you are not advocating taking a line which reduces the EV of any hand "in a vaccum"; you are simply taking a specific 0 EV line instead of folding, which would also be 0 EV. 


ALSO, if you really want to get nit-picky then some would argue that exploitation is actually part of a GTO strategy, and what you are talking about is called Nash equilibrium. 



themightyjim 10 years, 8 months ago

yeah I know the RPS analogy, and essentially you're correct that in some cases if you've made previous range constructions mistakes earlier in the hand you could result in a situation where you can make one action with all of your range and not be giving up EV.  But I do believe the idea of a GTO strategy is based around avoiding exploitation, and not maximally exploiting an opponent.

and my previous post was addressing the idea of Shania (an old 2p2 concept when discussing poker theory) that I believed was in some way relevant to the OP.  I was not trying to suggest that Shania as a concept represented a GTO strategy, or that the hypothesis by the orginal theorist that proposed it was in some way representing a GTO strategy.  More that unexploitable range construction as we think of it consists of having hands that have differing EV against opponents actual holding.  The idea is that we increase the value of our value hand by bluffing, and conversely increase the value of our bluffs by value betting (as our opponent can't can value against our range by simply always calling or always folding).  I believe that this is that idea that shania was eventually attempting to explain.  It's how I always understood shania, even if it wasn't actually laid out in that way initially.

fwiw I don't see how exploitation can be part of a GTO strategy at least in the way that we commonly use the term when discussing poker strategies.  A GTO strategy would inherently ignore the available information that would in fact inform an exploitative strategy.  at least that is my understanding of the concept.


arizonabay 10 years, 8 months ago

I don't really like most conversations about "GTO" because no one knows what a full GTO strategy would look like. However a GTO strategy would involve taking the most +EV line vs another opponent who was also playing a GTO strategy. (At least that is my understanding of it). Basically it involves 2 players exploiting the shit out of each other, both of them constantly taking the most +EV line that they can until they reach a point where they are at a stand still and neither one has the incentive to deviate, they can do no better. They are both maximally exploiting each other. This is the Nash Equilibrium right? And a solved GTO strategy would be the NE for the full game right? Or am I missing something. 

But I will admit to not really knowing the difference between N.E. and GTO - I always thought N.E. was the goal of GTO.

edit: I should say I actually do like conversations about GTO, I just don't like the arguments it often creates when no one really knows what a GTO solution looks like.


themightyjim 10 years, 8 months ago
AB let me say I enjoy discussing GTO as a concept in order to inform poker decision making, but I agree I don't like when it dominates a particular hand's strategy discussion as it rarely clearly guides us towards a most profitable strategy.

However I'm not sure that the way I understand GTO and the way you understand GTO are exactly the same, although I believe we eventually reach the same conclusions.  I do not believe that a GTO strategy has to be achieved by a process of alternating maximal exploitation until we reached an equilibrium, although I do believe it COULD be achieved via those means.

However when we discuss GTO in terms of 6max or full ring poker obviously there is much more than simply maximally exploiting an opponent in terms of strategy decision making.

I think the key word to focus on when trying to understand GTO as a concept is UNEXPLOITABLE.  we're making decisions by which our opponents at best (meaning they make optimal decisions) can break even (minus rake) against our strategy. That does not mean that we are playing a strategy that makes us the most money versus our opponents actual strategies (ie our opponents will never actually be playing GTO so a maximally exploitative strategy (MES) will almost never be the same as a GTO strategy).  It just means that our opponent can vary their strategy as they like but they can never do better than breaking even against us.

but, tbh I hate focusing solely on this concept because we are so far from understanding GTO play in poker (or even being able to calculate it) that we should hesitate to put so much of time and energy into it's understanding.  If you're goal is to totally understand GTO play in poker you'd be much better off spending your time getting a PHD in mathematics from a prestigious university.  If you're goal is to make money playing poker you'd be better off studying your opponents and trying to develop MES against them (while keeping in mind GTO concepts when facing new spots and new opponents that aren't well understood).


arizonabay 10 years, 8 months ago

I agree w/ most everything you said in this last post here....especially the part of our time being better spent on MES vs actual villains. 

Rapha Nogueira 10 years, 8 months ago

I don't finished Jandas book yet but the Nash equilibrium (I hope it is the same as GTO) is the best response we can give to our opponent actions. It is exactly making the play that gives the higher EV given our opponent actions and information available. It becomes unexploitable when played, because our opponent also knows that this strategy is the one which will give us the highest EV compared to any other strategy that we may play. 

sigis123 10 years, 8 months ago

I think you do not qiute understand what GTO means. As it was already mensioned ITT GTO is by definition, a two strategies which are maximally exploiting each other. So if ever somebody will solve NLHE it will be a huge strategy pair. And if you learn it and use it vs other strategy it might be NOT GTO anymore.

oblioo 10 years, 8 months ago

I don't want to get into a big discussion about semantics but the way I understand it is that technically a single GTO strategy is the MES vs. whatever strategy you're playing against. It's the optimal strategy because it makes you the most money, and exploitation is often involved.

When two strategies playing against each other are both GTO, then you're at equilibrium so the strategy is called Nash equilibrium. When most people talk about "GTO strategy" what they mean is Nash equilibrium strategy.


themightyjim 10 years, 8 months ago
I'll agree that we're basically arguing semantics, but the idea of reaching a NE as GTO as part of a two MES from a pairing doesn't really fit with the discussion brought up by OP (dealing with a 6max situation in a current 2014 game where it is folded to you in the SB).  I feel like when we use GTO in the context of such a situation we're focusing on the unexploitable NE end point of the strategy (as a concept since it clearly isn't a reality) and not the MES sextupling of 6 nemesis opponents it would take to get to a NE. 

I very likely may be wrong in where I've jumped into the conversation and how I'm applying the terms.  I just haven't seen in the past where people have expressed that an MES against a field of current players was the same as a GTO strategy.  I thought the general concept when we suggested that someone play in a more GTO-like manner was that they played as though the other players in the field were also attempting GTO and thus we were trying to achieve the NE end point.  Since clearly most of our opponents are not doing such a thing the MES of almost any current game type is going to be far from such an end point and thus look very different.  But maybe that should also be classified as GTO and I'm misapplying the term.  Obviously I'm not a PHD in mathematics and don't have the same understanding of these concepts as some others.


arizonabay 10 years, 8 months ago
fwiw I don't see how exploitation can be part of a GTO strategy at least in the way that we commonly use the term when discussing poker strategies.  A GTO strategy would inherently ignore the available information that would in fact inform an exploitative strategy.  at least that is my understanding of the concept.

I won't claim to truly know the answer to this but I believe this is wrong, a GTO strategy only ignores the other strategy after it is solved. But in order to find the GTO solution it is very much concerned w/ exploitation. Again, I have been under the impression that a GTO solution is a maximally exploitative strategy only it is exploiting a nemesis villain who is also maximally exploiting us. Both players are exploiting each other so much that they both lose incentive to deviate thus once this point is reached we are at the N.E. - once we have found this, we could always use it and never be exploited by another opponent. This strategy might not always make the most money (almost certainly not unless our opponent was also using this strategy) but it would prevent us from being exploited accidentally by an unknown opponent and the best that unknown opponent could do vs us was to play the GTO strategy himself. 

themightyjim 10 years, 8 months ago
I think our misunderstanding of terminology is suggesting that we don't agree on a conclusion when in reality we do.  I'm focusing on the concept of GTO in an actual poker game being played currently as opposed to an MES using available information from that game.  You're discussing the development of a GTO strategy via a process of MES against nemesis opponent(s).  We're reaching the same conclusions, but I'm suggesting that in a 2014 online poker game an MES will make us more money than a GTO strategy (which is obviously unknown).  And I think (based on your responses) that you agree with this.

but this can all get very confusing ;-)



arizonabay 10 years, 8 months ago

Yes, I agree w/ this - maximally exploiting our actual opponents will make us way more money than trying to play a strategy that doesn't exist yet. However, I think studying game theory and how a GTO strategy could be reached is of immense help in the construction of MES vs our actual villains. I hope that makes sense as I am getting really tired and was going to sleep 2 hours ago and thought I would check RIO one last time.

And yes all this gets really confusing.

BigFiszh 10 years, 8 months ago

Skimmed over the responses, and don´t feel like it´s necessary to involve into the general "GTO"-discussion, most has been said already. :)

So, back to OP´s question:

"For example, AA is always worth at least + 1.5 BB's to shove AA
pre-flop, and so anything we do with AA pre-flop must be greater than or
equal to 1.5 BB's in EV.  This should imply that we never limp AA
pre-flop, since Villain can simply improve their expected value against
our Aces by checking any two cards."

You always have to think about it as strategy vs. strategy, not hand vs. hand. Take your example. We shove AA preflop (and say, we´re only shoving AA and nothing else) and Villain correctly folds, as we always have the nuts. This strategy "guarantees" us 1.5bb / hand with AA.

Now, as you correctly say, any other strategy, limping / raising smaller included have to yield a bigger profit than 1.5bb. Now, assume, for the sake of whatever I implemented the following strategy: Besides raising some of my top-range (excluded AA), I limp with some crap hands (say 76s-54s) that aren´t good enough to raise, but for getting 3:1, I still pay the price. To not let you roll over me, I limp with AA as well. If you raise (say to 5 bb), I´ll 3bet my AA and fold all my junk. When I 3-bet you, you fold your entire range (knowing I have AA).

Now, obviously this is no particular intelligent strategy (for neither player), but as you can see, the EV of my AA now is +5bb instead of +1.5bb (and your overall EV is 0), so in a vacuum, I found a way to make limping way more profitable to raise with my AA.

That demonstrates, how strategies aren´t made of iron but always can go back and forth ... once Villain found a counterstrategy to combat our "Limp-AA"-strategy, we might be "forced" to always raise AA again, because EV of raising is higher than limping. And maybe (who knows?!) eventually we will find a Nash Equilibrium, where limping has the same EV as raising with our Aces and then - and ONLY then - we would start to play the so-called "mixed strategy" and sometimes raise and sometimes limp with our Aces - and get the "deception" you where talking about.

Got clear?

AF3 10 years, 8 months ago

  I'm focusing on the concept of GTO in an actual poker game being
played currently as opposed to an MES using available information from
that game.  You're discussing the development of a GTO strategy via a
process of MES against nemesis opponent(s). 

By definition, a game theory optimal strategy (pair) is two strategies which are maximally exploiting each other.  You converge upon the GTO strategy (pair) through maximal exploitation. 

AF3 10 years, 8 months ago

For example, even in a vacuum I'm not sure if open raising AA from SB is
higher EV than limping.

When you open-limp pre-flop in SB vs BB, the Big Blind can do the following with every hand they would have folded to a raise:

1)  Check with every hand that they would have folded to a raise.

2)  Fold every flop on which they do not have the nuts.

It should be very clear that this entitles them to a larger percentage of the pre-flop pot than if you shove AA, or even if you raise to any amount at which they would have folded for that matter (in which case they would have folded, and get zero percentage of the pre-flop pot).  This is the same reason why people believe open-limping from any other position is dominated by the strategy of raising first in. 

Even if villain checks back pre with air that
he'd fold to a raise, sometimes we end up making more money by either
letting him catch up a little bit or bluff

This seems to be a by-product of people not understanding the above.  For the formation of Small Blind vs Big Blind, there shouldn't be anything inherent about this in structure of the game.  I don't need to bluff you post-flop if I get a free-play preflop.  I've still increased my expectation with every hand by checking (if I know you have AA).  It doesn't matter if you know I always have the nuts when I bet. Then you just fold some small % of the time that you limp, and you've won less than if you would have raised pre-flop (with AA). 


Steve Paul 10 years, 8 months ago
I think you're misunderstanding the point of the "golden rule". Start from a GTO strategy pair, where by definition neither player can change their strategy to increase their EV. Then it must be the case that every hand is being played in the highest EV way because otherwise one player could make the change for a hand not being played in the highest EV way and increase their EV.

Now assume you take that GTO strat and play some idiot who 3bet shoves every hand when you open. A ton of your hands are not being played in their max EV way (in particular every hand you raise/fold) but that does not change the fact that you are playing the GTO strat.

In a GTO strat, every hand is played in the max EV way under the assumption that your opponent is playing a strategy that maximally exploits your strategy. Vs other (worse) strategies you make more money but individual hands are not necessarily playing in their highest EV way.

edit: So just because limping AA bvb does worse against a bunch of strats does not imply that it cannot be part of a GTO strategy. (FWIW I doubt limping your whole continuing range bvb is correct, but that's neither here nor there as no one can prove it either way. Generally I think these "play your whole range in one way" strats are not optimal but make it way easier for humans to play reasonably well vs the alternative of having a whole bunch of ranges to deal with.)



oblioo 10 years, 8 months ago
It should be very clear that this entitles them to a larger percentage of the pre-flop pot than if you shove AA

In a game where SB's entire range is AA and BB knows this, then yes this is true, but in practice it's not that simple. Clearly it's not correct for BB to only put more money in the pot postflop with made hands that beat AA. Also you did not address the point of being able to 3bet after limping. 

I could maybe be proven wrong but again it's not as simple as a toy game in which our entire range is AA.


AF3 10 years, 8 months ago

In a game where SB's entire range is AA and BB knows this, then yes
this is true, but in practice it's not that simple. Clearly it's not
correct for BB to only put more money in the pot postflop with made
hands that beat AA. Also you did not address the point of being able to
3bet after limping.
 

I thought I did: 

Is the idea that they are somehow forced to raise a certain percentage
when we limp, or else we have improved the expected value of our weakest
holdings by limping?  

For some reason everybody is ignoring that part of the post, lol.  

I'm not arguing one way or the other, by the way, I'm just trying to understand this. 

oblioo 10 years, 8 months ago

Oh sorry, yes that's part of it. You may be confused because, as others have alluded to, you are not considering your whole range and how one aspect of your opponent's strategy is that she/he never knows exactly what hand you have.

AF3 10 years, 8 months ago





stevejpa


I think you're misunderstanding the point of the "golden rule".

You made a lot of good points, and I do think I had some confusion regarding individual hands being played in their most +EV manner.  Thanks for clearing that up. 

I don't think I misunderstood what's going on, though: 

Is the idea that they are somehow forced to raise a certain percentage
when we limp, or else we have improved the expected value of our weakest
holdings by limping? 


This seems to follow from what you're saying, right?
AF3 10 years, 8 months ago

mightyjim:

IE there are some hands that as UTG raises in 6max NLHE are basically
breakeven. But having them as part of our range adds equity to our other
hands by allowing us to value bet and bluff certain board textures that
would otherwise be excluded by our range.  More importantly they allow
us to meet minimum defense frequencies so we aren't just exploitatively
playing too tight.

This seems more important as we get deeper, as the example of 10 BB starting stacks would show.  I used to agree with the pre-flop "minimum defense" thing as applied to opening from a non-blind position, but Sauce has kind of convinced of the opposite.  We open all hands that are +EV and fold all hands which aren't.  I guess the idea behind "Shania" is that it's impossible to consider the EV of any specific holding (except for very specific instances like where we have the nuts on the river)? 

Also, it seems like we don't consider "EV", but rather we try to find "the minimum EV" of every hand by looking at dominating strategies, if that makes sense.  In other words, we don't say "XX is worth this much", we say that "XX should be worth at least this much, or it's likely we're dominated."

AF3 10 years, 8 months ago
mightyjim:  I do not believe that a GTO strategy has to be achieved by a process of
alternating maximal exploitation until we reached an equilibrium,
although I do believe it COULD be achieved via those means.


I want to say "of course it does", with regards to your first sentence.  Suppose you form a strategy pair through some other method besides alternating maximal exploitation.  Then one of two things has happened:

1)  You are at equilibrium, in which case you have two strategies which are maximally exploiting each other.  You have just achieved a GTO strategy pair by alternating maximal exploitation by alternating zero times.

2)  You are not at equilibrium.  Then, by definition, somebody has incentive to change.  They will never change to a strategy which lowers their expectation, so whichever strategy they change to will be closer to the maximal exploitative strategy than the previous one.  In this case, it is clear that by iterating this process, we should converge to equilibrium (eventually). 

Because these are the only two cases (by the law of the excluded middle), then every way of converging to "GTO" should be equivalent to the process of alternating maximal exploitation. 

AF3 10 years, 8 months ago

I don't want to get into a big discussion about semantics but the way I
understand it is that technically a single GTO strategy is the MES vs.
whatever strategy you're playing against. It's the optimal strategy
because it makes you the most money, and exploitation is often involved.

The term "GTO strategy is just referring to a (single) strategy which is part of the equilibrium.  You're talking about something different:

the MES vs.
whatever strategy you're playing against.

This is arguably more important, and it actually has it's own name.  It's called the "nemesis" strategy, and it's distinct (but related) to "GTO" strategy (pairs).


AF3 10 years, 8 months ago

You may be confused because, as others have alluded to, you are not
considering your whole range and how one aspect of your opponent's
strategy is that she/he never knows exactly what hand you have.


Essentially, this whole thing is about how to construct a range. 

There seem to be two approaches. 

The first approach, as often endorsed by Matthew Janda, is that the range which we end up in any given situation is a by-product of taking the most +EV line with every combination.  We do not "balance for the sake of balance". 

However, I have also heard him say numerous times that he thinks a "GTO" strategy would involve several mixings (bet sizing, range construction, etc.)  He may have reversed his stance, on the first point, but I don't think he has. 

The second approach is "shania".  That is constructing ranges for the sake of balance.

There certainly seem to be aspects of each of these approaches which merge. 

I'm trying to understand when the two approaches merge and when they don't. 

For example, in the first approach, when you can argue that the EV of two strategic options are the same, you can play a mixed strategy.  Otherwise, your strategy is clearly dominated.  

The problem that I have with constructing ranges using "shania" is that it basically runs the risk of protecting a bad strategy.  You see all this time, if you think of somebody who goes "well...I want to value bet [insert hand they shouldn't value bet], so I'll obviously need to be bluffing more." 

You still need the first approach before you use "shania".  It seems that "shania" is a nice litmus test of your strategy, but it seems that it's a very weak approach.  (I mean "weak" in the logical sense here).

Steve Paul 10 years, 8 months ago

The first approach, as often endorsed by Matthew Janda, is that the range which we end up in any given situation is a by-product of taking the most +EV line with every combination.  We do not "balance for the sake of balance".  

However, I have also heard him say numerous times that he thinks a "GTO" strategy would involve several mixings (bet sizing, range construction, etc.)  He may have reversed his stance, on the first point, but I don't think he has.

These aren't necessarily contradictory stances. There are a lot of situations where hands are indifferent to multiple actions at equilibrium* and in those cases you will have the same hand mixed among those actions. 
*This is at least true for some slightly simplified flop/turn/river games as shown in Will Tipton's 2nd book.

Is the idea that they are somehow forced to raise a certain percentage 
when we limp, or else we have improved the expected value of our weakest
holdings by limping?  

Yes. If bb raises our sb limps less than an equilibrium strat does then he probably reduces the EV of our strong hands but increases the EV of our weak hands more than enough to compensate. We know it's at least enough to compensate (assuming there's some SB limping at GTO), because if it wasn't then the GTO strat wouldn't be the GTO strat!


Santaur 10 years, 8 months ago

I can't read this thread cause its insane, but the answer to your question in short is...


Is the idea that they are somehow forced to raise a certain percentage when we limp, or else we have improved the expected value of our weakest holdings by limping? 

Yes. They aren't "forced" but they'll raise their EV signficantly by raising a wide range preflop. In which than you can obviously 3-bet AA... which possibly might make AA higher EV as a limp than raise preflop. 

arizonabay 10 years, 8 months ago

sorry for contributing to the insanity of this thread and thank you for responding to the OP's question which was something I was genuinely interested in. So much so in fact that I entered into a conversation that I knew I wasn't especially qualified to enter.

AF3 10 years, 8 months ago

Yes. They aren't "forced" but they'll raise their EV signficantly by
raising a wide range preflop. In which than you can obviously 3-bet
AA... which possibly might make AA higher EV as a limp than raise
preflop. 

Great, thanks.  That's what I thought. 

"Forced" = They must do this to maximize their expectation in the game.  

I use "forced" to reflect the idea that if they do not do this, then the expectation they are sacrificing should show up somewhere else, and there-by raise the expectation of everybody else at the table.  In this case, it would raise our own EV of calling with hands that we would otherwise have folded.

Is that correct?

AF3 10 years, 8 months ago

These aren't necessarily contradictory stances. There are a lot of
situations where hands are indifferent to multiple actions at
equilibrium* and in those cases you will have the same hand mixed among
those actions. 
*This is at least true for some slightly simplified flop/turn/river games as shown in Will Tipton's 2nd book.


I don't think they're contradictory at all.  What I was suggesting is that one approach seems stronger than the other. 

AF3 10 years, 8 months ago

Santaur: they'll raise their EV signficantly by raising a
wide range preflop. In which than you can obviously 3-bet AA... which
possibly might make AA higher EV as a limp than raise preflop. 

This would imply that at the equilibrium, we could at least conclude that AA is higher EV to limp-shove than open-shove. 

Is it then valid to conclude that it also must be higher to limp than to raise for normal sizing?

The argument would be a kind of transitive thing between our various strategic options, using the "fact" that the EV of open raising (to 3BB, for example) must be greater than the EV of open-shoving, so we would have:

EV of raising to 3BB < EV of open shoving

EV of open shoving < EV of limping 

==> EV of raising to 3BB < EV of limping

Edit:  As stevejpa pointed out, this equation is invalid.  

"Your first line should say EV of raising to 3BB > EV of open shoving
given your previous statement and so you can conclude nothing."



Steve Paul 10 years, 8 months ago
Your first line should say EV of raising to 3BB > EV of open shoving given your previous statement and so you can conclude nothing.


Nick Howard 10 years, 8 months ago
For example, AA is always worth at least + 1.5 BB's to shove AA pre-flop, and so anything we do with AA pre-flop must be greater than or equal to 1.5 BB's in EV.  This should imply that we never limp AA pre-flop, since Villain can simply improve their expected value against our Aces by checking any two cards.

basically +1 to santaur.  i think you're confusing yourself by thinking that if he exploits AA with a subsection of his range (his PF checks), it means AA is getting exploited on the whole in the limp line.  but he could still be raising enough pre to make the EV of AA > or = to the RFI line


AF3 10 years, 8 months ago

basically +1 to santaur.  i think you're confusing yourself by thinking
that if he exploits AA with a subsection of his range (his PF checks),
it means AA is getting exploited on the whole in the limp line.  but he
could still be raising enough pre to make the EV of AA > or = to the
RFI line

Well, that was kind of the whole question--

He certainly can exploit AA with a subsection of his range, but the question was can that subsection actually be his entire range?

The answer seems to be "no" if Villain is trying to maximize their expectation. 

AF3 10 years, 8 months ago

You're right, though.

I've been playing so much Fast-Fold poker without reviewing my hands, due to this promotion on the site I'm playing on, that it's definitely messed with my head. 

Nick Howard 10 years, 8 months ago

i dunno it just seems intuitive to me that the the EV of overpairs would have to be higher as a limp 3-bet rather than an RFI once you install a substantial (& predominantly weak) SB calling range in that formation, assuming button raises more and more aggressively the more capped our limping range is


AF3 10 years, 8 months ago

I think you're right. I think it's actually kind of easy to argue (mathematically) that it dominates raising.

Steve Paul 10 years, 8 months ago

It does not seem easy or intuitive to me...want to walk me through the argument? 

To address Nick's point, if BB is putting in substantially more money vs a limp than a raise then of course we should limp our strong hands, but we should probably also raise our weak but still strong enough to not fold hands, which would incentivize bb to adjust. It's possible that limping big hands is a good exploitative adjustment in current games but I don't quite follow the reasoning why it must be better than raising vs good opponents. (edit: I'm not saying it's not best to limp, I'm saying I have no idea but I don't think anyone's presented a conclusive argument either way)


Nick Howard 10 years, 8 months ago

im not arguing for or against the limping strategy.  i'm just saying that if you do play the limping strategy, it intuitively makes sense that starting off totally capped on the limp would have BB going ape-shit raising and the SB would recounter by limp-3betting extremely linear/aggro, and it'd trend toward an equilibrium where i'm guessing AA would always be most +EV as limp 3bets after limping.    

Comparing that (^) EV of AA to the RFI EV of AA would require you to solve the whole game tree in both lines, and i don't think that part is intuitive at all 

I was unclear b/c the way i initially phrased it sounded like i was saying limping AA initially dominates RFI'ing AA.  when what i really meant was once you're already playing a predominantly weak limping strategy, limping AA dominates RFI'ing AA.    



AF3 10 years, 7 months ago

I've thought about where I was making errors in my thinking when I wrote the original post, and I think that's I was confusing the idea of game theory dictating that we "always play the highest EV strategy" with the statement that "we always play each hand in the most +EV manner", where the second statement should actually be quite meaningless on it's own, except for a few situations. 

Be the first to add a comment

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