Balancing Bluffs vs Balancing Strategy

[gaming_mouse]

[ QUOTE ]


One obvious example of this is the "arms race" on the turn in high limit LH games online. What developed there was a tendency after the sequence rc; kbrc; b? on two-flushed boards for people to semi-bluff raise liberally on the turn (including all flush draws, straight draws, and other weaker draws such as small pair+gutshot and the like).

The exploitive response to this that I observed was for original bettors to three-bet liberally, including hands as weak as middle pair in this sequence, because the number of semi-bluffs was just far too high compared to the number of value raises (as these players wouldn't adjust their value raise thresholds to balance).

jerrod

[/ QUOTE ]

Jerrod,

I love this post. I've been trying to think about different situations like the one above (sequences of moves that occur regularly and help define your opponent's strategy) for some time, and your approach above is what I've been looking for.

In your book (I just ordered it tonight), do you list and analyze what you consider the most common betting sequences in limit and NL holdem? If not, do you do this as a way of improving your own play? How systematic are you? It seems that one approach to the game (not necessarily the best, but an instructive one at least) would be to exhaustively list these betting sequences, how often they came up, what their frequency should be and therefore how to exploit opponent's whose frequencies deviate from the optimal ones.

Thanks for any further thoughts,
gm

[Jerrod Ankenman]

[ QUOTE ]
[ QUOTE ]


One obvious example of this is the "arms race" on the turn in high limit LH games online. What developed there was a tendency after the sequence rc; kbrc; b? on two-flushed boards for people to semi-bluff raise liberally on the turn (including all flush draws, straight draws, and other weaker draws such as small pair+gutshot and the like).

The exploitive response to this that I observed was for original bettors to three-bet liberally, including hands as weak as middle pair in this sequence, because the number of semi-bluffs was just far too high compared to the number of value raises (as these players wouldn't adjust their value raise thresholds to balance).

jerrod

[/ QUOTE ]

Jerrod,

I love this post. I've been trying to think about different situations like the one above (sequences of moves that occur regularly and help define your opponent's strategy) for some time, and your approach above is what I've been looking for.

In your book (I just ordered it tonight), do you list and analyze what you consider the most common betting sequences in limit and NL holdem?

[/ QUOTE ]

No, the book doesn't contain much specific advice on particular poker games (by design - we do consider a particular case study of NL - the BB vs an early raiser).

[ QUOTE ]
If not, do you do this as a way of improving your own play?

[/ QUOTE ]

Yes, and I recommend it to others.

[ QUOTE ]
How systematic are you?

[/ QUOTE ]

Not very, because I'm kinda lazy.

[ QUOTE ]
It seems that one approach to the game (not necessarily the best, but an instructive one at least) would be to exhaustively list these betting sequences, how often they came up, what their frequency should be and therefore how to exploit opponent's whose frequencies deviate from the optimal ones.

Thanks for any further thoughts,
gm

[/ QUOTE ]

I think that you could totally benefit from just looking at your action frequencies in the most important lines in headsup pots - ie, button raises, blind defends, check-bet, etc.

jerrod

[Deorum]

[ QUOTE ]
Theory has been defined as the opposite of practice.

[/ QUOTE ]

I don't want to hijack this thread, so please, nobody respond heavily to this.

This definition of theory has always bothered me, and I just wanted to get it off my chest after having been reminded of it here. I often hear somebody make an argument, to which somebody else will respon 'well, in theory that works, but not in reality.' This really irritates me. If something does not work in practice, it does not work in theory either. Rather, it means that there is something wrong with your theory. Something is being omitted or miscalculated.

For instance, if somebody says 'if everybody in the world did X, then Y would be the result' and another person replied, 'well, that works in theory, but not in reality' what they really mean is 'well, that works in your hypothetical but not in reality.' The words theory and hypothetical are not synonymous. A theory is an explanation of how different parts of a system work, and why those specific parts work the way they do. A hypothetical is a contrived system.

[leaponthis]

[ QUOTE ]
using Game Theory to balance an overall strategy, practically guarantees a long term win

[/ QUOTE ]

[ QUOTE ]
The whole point of a game theoretic strategy is that it can't be a losing one, can't be exploited.

[/ QUOTE ]

I definitely understand that a game theoretical approach can't be exploited. The problem is that there is no game theoretical approach. Nor can any of these math guys come up with one that is workable. And that goes for heads up play also. If there existed a workable (winning) game theoretical strategy, guys like Sklansky and Fergusen and Weideman, etc.. would win all of the heads up championships. But they don't.

Sklansky has now stated that the strategy he (and Malmuth) have sold to the public is not a winning strategy. To beat high limits you must adjust that strategy with game theroy. He says that but where is the game theory adjustments of which he speaks? I'll tell you where they are. They are under development. Sure they are. My point about the social aspects of poker that Sklansky doesn't give much credence to are basic skills like reading your opponent, and understanding position. In poker there is such a thing as talent. Talent that allows a player to rise above his opponents and out play them. Out play - not out calculate. Sklansky disregards this in his theories because talent does not lend itself to logical analysis.

leaponthis.

[mvdgaag]

Game theory uses randomness to choose an action. This works for bluffs when your opponent doesn't see a part of your hand.
If he gets to see the river card that you choose to bluff on (like your example in the theory of poker, but now it's holdem) he knows it is more likely you are bluffing if it's a blank than if it might have made your drawing hand. So game theory is flawed here.
The same applies to strategies. Your strategy reveales itself a bit by your actions on previous streets like that rivercard revealed your bluff. I think that if you'd randomise your strategy every hand, it's very, very hard to determine that strategy for your opponents. It will probably work great for most players, but real good players might have better results by choosing their strategy by judgement of the situation/game.

[Dendrite]

Jerrod,

[ QUOTE ]

Playing optimally or in a balanced manner isn't about equalizing all your opponent's actions -- just the ones that are on the borders.


[/ QUOTE ]

What does this mean with regard to your example? It's important to make sure your opponent can't profit by 3betting all of his hands (or close) at a certain point in a sequence?

[Jerrod Ankenman]

[ QUOTE ]
Game theory uses randomness to choose an action. This works for bluffs when your opponent doesn't see a part of your hand.

[/ QUOTE ]

The "randomizing" aspect of game theory is pretty unimportant in poker compared to the "maximizing" aspect. Many people have written about randomizing their play with their watches or whatever, but this type of thing is only a very small part of playing optimally - in fact, you generally should not randomize your bluffing by selecting some random card to bluff on. Instead, you should use card removal effects to decide what the best hands to bluff on are, and use those. The only thing that really needs to be randomized are hands that require mixed strategies, and there are quite likely to be very few of those.

[ QUOTE ]
If he gets to see the river card that you choose to bluff on (like your example in the theory of poker, but now it's holdem) he knows it is more likely you are bluffing if it's a blank than if it might have made your drawing hand. So game theory is flawed here.

[/ QUOTE ]

This is quite untrue - game theoretic strategies have a mixture of bluffs and value bets on *every* river card.

[ QUOTE ]
The same applies to strategies. Your strategy reveales itself a bit by your actions on previous streets like that rivercard revealed your bluff. I think that if you'd randomise your strategy every hand, it's very, very hard to determine that strategy for your opponents. It will probably work great for most players, but real good players might have better results by choosing their strategy by judgement of the situation/game.

[/ QUOTE ]

"Randomizing strategies"(ie playing a different strategy on each hand) isn't what we're talking about at all.

jerrod

[Jerrod Ankenman]

[ QUOTE ]
Jerrod,

[ QUOTE ]

Playing optimally or in a balanced manner isn't about equalizing all your opponent's actions -- just the ones that are on the borders.


[/ QUOTE ]

What does this mean with regard to your example? It's important to make sure your opponent can't profit by 3betting all of his hands (or close) at a certain point in a sequence?

[/ QUOTE ]

Yeah. Imagine the other guy's strategy as like a series of regions (the value raise region, the value call region, the semi-bluff region region, the drawing call region, and so on). Now if there weren't card removal effects (like this were some kind of idealized game), the only hands that would be mixed strategies would be hands on the border, which are mixed to make the overall frequencies right.

In real poker there are card removal strategies and in order to balance for new public information (like the community cards to come) there is probably a little mixing for several hands around. But almost certainly many hands use pure strategic options. (this means they do one thing or another 100% of the time)

In general, when a player deviates from a strategic option that is pure, he will often lose value.

jerrod

[David Sklansky]

To give a specific example of what I was talking about, take the chapter in our no limit holdem book where we recommend limping with specifically two eights with a certain stack size. The Mathematics of Poker seems to make reference to this play and states that any small gain from limping would be negated by the fact you are not playing preflop in a consistent way and could thus be more easily read. Whether that is true depends on the game.

[jogsxyz]

[ QUOTE ]
And just like using Game Theory to bluff and call bluffs, using Game Theory to balance an overall strategy, practically guarantees a long term win. My experience tells me that the vast majority of games will be beaten for a greater amount if this approach is shelved for my more exploitive approach but time will tell.

All comments welcome.

[/ QUOTE ]

When this statement becomes false, the games will no longer be beatable. No one will be able to overcome the hefty expenses.