Nate's Theorem on the "Value Bluff"

[Nate tha\\\' Great]

[ QUOTE ]
im not sure, but are you insinuating we find a game-theory style which makes bets against every possible villain so that it is inexploitably +ev? if so, doesnt this ignore all reads which influence the decision?

[/ QUOTE ]

I've increasingly come to believe that a starting point for a poker strategy is to find a game theoretically sound strategy (rather than a "solid" strategy), and then to deviate from that as needed when you do have reads.

[jfish]

i agree, i just wasnt sure if thats where you were going with it. really great post.

[Teh1337zor]

awesome awesome post

[Victor]

more proof limit players are just smarter.

[Rotating Rabbit]

Very, very, impressive post.

[RiverFenix]

Good stuff. Think about submitting it to 2p2 mag or Bluff or something like that?

[VanVeen]

great post

only addition i would make is the same one trix suggested: the same player may be a composite of player types and thus the term "individual opponent" is somewhat misleading.

2. In order for a bet to be profitable as a "value bluff", it must be profitable either as a bluff against some individual opponents and be profitable as a value bet against other individual opponents, or profitable as a bluff against an individual opponent some of the time and profitable against that same opponent as a value bet some of the time.

[wpr101]

Nice post. I think most of us has thought about this intuitively at some point; but maybe not on these terms.

[xorbie]

vanveen is right. the whole idea of epistemic types is a hot topic in game theory these days, and it has clear applications to poker.

there are mixed strategies to be played, of course, but in an individual hand the person you are playing (incorrectly, perhaps, but so it goes) either does or does not believe you are bluffing. so generally they will either call top 10% (totally random) or top 30% of their hands, meaning you can actually "value bluff" the 15-20% percentile of your hands because the will fold better and call worse.

[Nate tha\\\' Great]

[ QUOTE ]
vanveen is right. the whole idea of epistemic types is a hot topic in game theory these days, and it has clear applications to poker.

there are mixed strategies to be played, of course, but in an individual hand the person you are playing (incorrectly, perhaps, but so it goes) either does or does not believe you are bluffing. so generally they will either call top 10% (totally random) or top 30% of their hands, meaning you can actually "value bluff" the 15-20% percentile of your hands because the will fold better and call worse.

[/ QUOTE ]

Clearly, there's a spectrum of opponent types. On the one extreme, you have opponents who are too tight and against whom you can bet profitably as a bluff; on the other extreme, you have opponents who are too loose and against whom you can bet for value. Then you have the opponents in the middle who play "just right", against whom the bet presumably loses money.

What it seems like you're saying is that any given opponent at any given time tends to be lined up toward one or another end of this spectrum. This is is how people are trained to play NLHE, after all, i.e. they're taught that there is little difference between some moderately strong made hand like top pair with a marginal kicker and a bluff-catcher when it comes to calling off a large bet. So there's some inherent tendency to call off either too loose or to thin.

Very interesting if true. I suspect that this is also an aspect that is fairly unique to NLHE. Hand values behave more continuously in something like LHE or on the pre-river streets in PLO.