The Singularity of Bluffing - Page 3

On November 02 2007 18:13 twotimesopt wrote: Show nested quote + Your bluff was bad because he called. We made a mistake thinking he would fold JT when he wouldn’t and that is our fault. i disagree. i think it's important to note that we are bluffing against a RANGE of hands, not just a particular hand. and I still think that we are ALWAYS trying to maximize EV with every check, bet, or fold in poker the way i see it, a bluff consists of two parts: (1) an approximation of the villain's range (2) the conclusion that, given this approximation, a bluff causes a fold often enough to be profitable If (1) or (2) is flawed, then it was a bad bluff. If they are both true, then it was a good bluff. Notice this assessment is completely results-blind. I hate disagreeing with you PU b/c you're clearly the man. You seem to be advocating a viewpoint that is very contrary to mine, and I'd be very eager to hear any thoughts from you on this. cheers.

kevin although I agree with what you said. You should also consider other bits of information such as the current game conditions and what has happened to villain recently that could lead him to call a bluff.

Although in general i think thinking of ranges and figuring out what % of his range he'll fold I think at the higher levels you gotta consider the other factors just as heavily or even more so.

On November 02 2007 17:49 casinocasino wrote: I dont get this article

Interessting article.

I would tend to play more "G buck oriented" with my bluff, kind of what like floofy or jman described perfectly, but i think that both views are correct.

Lets me explain it a little deeper : Actually what G buck teach us is that for exemple a 3 barrel bluff with JT could be theorically "sound" if u have a range that make it +"G-buck EV" against the calling hand of your opponent.

However, if u are somehow sure that someone IS calling with a hand that make it +Gbuck to bluff but you know that metagame doesnt matter or he is tiltling etc. it is correct as well to "not bluff" if you are 100% sure he will call here with his "inferior g-buck hand" but still "superior EV hand" for him

The two different perspective arise from lenght of play issue i think, like if u are playing someone currently on tilt or who wont play you enought so that metagame matters, the PoorUser way of thinking IS the correct one. However, if u are playing a good reg who notices your range and such and isnt tilting or anything and playing in a metagame way, the GBuck way of thinking is the correct one.

At least that is what i think :-)

And anyway against fishes and bad players the meta game issue kind of "never" matter, so the PoorUser way of thinking should be better the majority of time, except when you are playing everyday the same good regs. that adapt to metagame and such

But in a nutshell none of these "views" are incorrect, i mean both way of playing and thinking still are +EV and good, PoorUser view is "direct +ev" and g-buck is "long term +ev"

nh

The apparent contradiction between G-bucks and PU thinking is because when thinking about G-bucks we assume villain will fold TJ here 100% (because this is norally folding range).

But say he actually folds it 85% of the time. Now when we bluff 100% of the time we are +Ev.
But if we can determine the times he'll call (as mentioned: he has folded to big riverbets recently, tilted... whatever else read that makes us careful) we will not shoot the 3rd barrel in these situations and thats more Ev+ in short and in long run than just automatically going for it everytime.

On November 02 2007 23:19 sawseech wrote: the goal of bluffing is to cause your opponent to become indifferent to calling with hands that they otherwise would not, so as to improve the EV of your overall range in a specific spot the actual success of a bluff is meaningless; it's important to gauge ranges with some accuracy but even more important to identify points of indifference so that you can estimate where his thresholds lie so that you can determine how often you need to bluff in order to strengthen the result of your overall range basically bluff a station with very low (but nonzero) frequency and bluff nits all day and hover in between for the rest assuming that a player plays a reasonably strong range in any given circumstance ideally the longterm results of your bluff would mirror the results of your entire range in any specific circumstance

props

Caveat: I'm no expert obv and this is only my interpretation of what the article meant to me.

So I read it a couple times first and I see where one can find contradiction with Jman g-bucks concept, and think that's independent from the concept PoorUser intends to highlight.

Again, I'm only speculating to what PoorUser meant, but the people pointing out that if a villain calls with 15% of his range then a bluff can still be correct are not invalidating PoorUser's point either. The problem is not with that logic. The point is that with a closer look, we may very well be able to filter out that 15% of the time a particular decides to make that call and avoid it.

Even though we have correctly identified the villain's range of holdings, we still are prone to miscalculating what % of his range he's calling with. As pooruser said, he's likely to fold Jxx to a bet. And furthermore, given other available information we might have the clues from what we know about the villain, our image, and the recent gameflow that the likelihood of him calling with lighter than usual may be abnormally high.

I think in retrospect PoorUSer realized that this villain was likely ALWAYS calling here with JTo almost always in this spot and maybe with much worse such as 33. It's not like the villain said, I should call with this weak holding X% of the time and then used some RNG to decide what he would do in this spot. Even your more sophisticated villains who know they can't be always doing X or Y in a certain spot are probably randomizing their actions based off of gameflow considerations - but in general most villains without history are going with something 100% of the time given context and the uncertainty in our equation comes from (failing to observe) that very context.

The revelation is that PU did a poor and biased job of estimating the villain's call/fold frequencies despite an accurate estimation of his range of holdings because he failed to consider some other possibly trivial things. I'm guessing PU felt that this PARTICULAR villain was calling in PARTICULAR spot at this PARTICULAR moment 100% of the time with JTo and maybe worse like 33 and he could have come to that conclusion before making the bet based on info he had up to that point. It is this revelation that makes this SINGULAR bluff incorrect.

Blindly betting here always may be correct, but PoorUser wasn't blind, he had more info about this villain and about his perceived image and the recent gameflow and chose to ignore it. It was in the context of this extra info that the villain was actually more likley to call with holdings as weak as Jx and thus a good bet (ignorant of this extra info) thereby became a 100% incorrect one.

I believe PoorUser felt that he had the info to conditionally constrain this spot more thoroughly and conclude that in this unique spot the liklihood of a call was much greater than the originally predicted 15% and possibly very close to 100%.

What I take from PoorUser's article is that an overwhelming majority of my villains are either ALWAYS calling or ALWAYS folding given all variables up to that point. We can be lazy and say an unknown villain calls a bet on this board only 15% of the time - so a bluff is profitable. But if we look closer, sometimes we might realize we have the info to avoid a lot of the situations that fall in the 15% of the time he's very highly likely to call us light. That is, if we consider that this particular villain has displayed characteristics X,Y,Z we'll conclude he's more likley to call light (and maybe if we consider that he saw us do A, B, C earlier in the session we'll be even more convinced). That's awfully complex, but most of the time it's rather simple. For instance, maybe in PU's example, he saw this villain make a very loose call earlier, his pahud stats are donkish, and he just got caught bluffing last revolution.

Meh... goodluck if anyone can interpret what I'm trying to say.

pocakey,

i was one of the first ones to disagree w/ poorusers stance, but after you clarified it i agree 100%. i think u nailed it right on the head, gw.