It has long been known that scrub jays, cousins of the crow and the raven, both hide and steal food from each other. Researchers recently demonstrated, however, that jays first learn to hide food, not by being victims of a theft, but by stealing themselves. That is, it seems not to occur to the jay that another bird might try to steal its food until he himself steals from another bird:
While experienced thieves engage in high levels of re-caching after being observed hiding food, birds that had never stolen another bird’s cache move few, if any, items. It would appear that it takes a thief to know a thief. This suggests that scrub-jays use their past experiences of, say, having been a thief, to predict what another individual, in this case the potential pilferer, might do.
Discovering the ability to steal leads to an epiphany: “If I can steal from him, he can steal from me.” This in turn leads the jay to take counter-measures in the future, adapting his food storing strategies to account for the possibility of theft.
What does this have to do with poker? I’ve heard it called the “Mirror Theory”: your opponent will expect you to play the way that he plays. If he slowplays the nuts, then he will expect you to do the same and may pay you off if you play a big hand fast.
Likewise, a player who only bets the flop when he hits it will probably fold to far too many of your continuation bets. Presumably, if he were aware of the concept of a continuation bet, he would employ it himself. The fact that he does not suggests that it has never occurred to him and thus that he will not have a strategy for combating such a play from you.
Pinpoint parallels and similarities between people and animals
http://www.youtube.com/watch?v=pSm7BcQHWXk
Observation data:”player who only bets the flop when he hits”
Generalization and imagination: “it will probably fold to far too many of your continuation bets”
you touch the aspect how and why is relatively easy to build profile for your opponent.
The ability to profile your opponent is the key to make exploitative strategy.
We build such player profile by observation and generalization and imaginations too.
The profile is sum of tendencies and generalizations and imaginations.
Many times profile contains more generalizations and imagination than real observed facts.
It is still great tool to make $$$.
Why?
Because sometimes we can generalize across wide range of players and assumptions holds true.
Yeah the way that people play when they claim to have “no read” on a given player, and the way that you should play if you genuinely had no read on a player are vastly different. You make all sorts of assumptions based on what you have observed about other people who play your stake or just about randoms in general, and while these are far from precise, you are usually better off playing an exploitive strategy based on them than trying to play GTO.
This is often a good spot to make an “image” play, as an investment towards future winnings, rather than relying on a stereotype that is often so general as to be useless anyway. If you’re basically pulling a guess out of your posterior, err on the side of sending the message you want to send.
Also, someone should probably mention that you often can and should employ Bayes-like analysis in these situations. I think there are a couple examples in The Mathematics of Poker. I’m too lazy to go find my copy and look and it would be too long-winded to explain here anyway. But it’s worth taking the time to understand, because it’s often the case that your assumptions about the probability of an opponent fitting a particular type based on his betting will be way off unless you understand conditional probability and apply it correctly.
You have quiet different approach to poker than me.
Your approach seem to be very scientific and strictly logical.
My approach is very different.
You wrote “stereotype that is often so general as to be useless anyway”.
LOL.This is axiom – a self evident truth that require not proof. I am not able to respond.
you suggest me Bayes-like analysis.Thanks for advice.
I have to tell you I am not a fan to apply Bayesian interpretation of probability to make generalizations, assumptions,decision in poker.
When I make decision I just use two things: logic and intuition or I try to find some consensus between my conscious and my subconscious.
So recently playing high stakes I came to conclusion I just use just more my intuition than str8 logic.Or definietly I use some MIX.
Example:
Maybe you know subconscious is the center of emotion.Well.
Sometimes I imagine I sense some emotions of my opponent when I catch him bluffing allin.
Maybe the reason is size of his bet or “timing tell”.Maybe my imagination -I imagine he is a maniac or hopeless bluffer?.
Yes I do profilnig.Yes I do stereotyping.Yes I do “unscientific” generalization.
By the way subconscious is the center of imagination too.
I’m using “stereotype” in the technical sense. That sort of stereotype can be highly accurate. For example, the stereotype of “calling station” is highly accurate for nearly all calling stations.
The stereotypes that I find too broad to be of much use are ones such as “average small-stakes NLH player”, “well-dressed guy with a lot of jewelry”, “Asian”, etc.
Also, I have nothing against playing intuitively. I think some of the best players are mostly intuitive players – Jennifer Harmon, Daniel N, Freddy Deeb.
At the other extreme are successful players with no observable intuitive ability – Ferguson and Bloch are always the first two names that pop into my head when I think of that type. If Chris or Andy has ever made a play based on gut feeling, I haven’t heard about it.
However, I think regardless of your style, you can get better if you at least understand the general principles behind Bayes Theorem, even if the math eludes you. The whole point is that sometimes your instincts about players are wrong, and Bayes is a useful tool for identifying where your instincts are costing you money.
“The whole point is that sometimes your instincts about players are wrong, and Bayes is a useful tool for identifying where your instincts are costing you money.”
Lyn I give you the credit.
Maybe it is good idea to revisit subject
Bayasian analysis is interesting subject.
Maybe you know something I do not know.
Or I know something you do not know.
I am consultant software developer.
The first time I heard Bayes and poker was early 2006 in NYC.
There were positions in project on poker bot where experience with Bayesian approaches and conditional probabilities was requied.
I assume there are currently Bayesian poker bots doing OK or losing money in midstakes NL.
I assume because there are plenty programming assigments to tweak such Bayesian bots because they started losing $$$.
This is very important imo. In the 200NL games I play in, calling an unknown’s river raise/check raise just because I’m somewhere near the top of my range given how I played the hand, is gonna be a huge mistake, because most players just don’t bluff raise on the river very often at these stakes.
The trick is to make sure there’s a valid basis for your “generalizations” and “imagination”.
For example, I don’t think this holds up well to empirical observation:
It is certainly true of some players, especially novices and certain types of rocks. But there’s a rather large category of players – loose-passives – for whom it is simply not true at all.
“The trick is to make sure there’s a valid basis for your “generalizations” and “imagination”.”
LOL.I am confused you saying quiet opposite to what Bayesian analisis and aproach is about.
You do NOT need a valid basis.
Bayesian analisis is about making “subjective” assumptions and generalization right of bat just after one trial.
This is the “big” difference between Bayeasian aproach and the rest (classical,etc).
In classic statistics you make no assumptions,you need a valid basis -lot of trials to deduce anything,
You do not have valid basis until many many trials occur, your data will be statistically insignificant.
You are sceptical so take our example and apply Bayesian analysis yourself.
“player who only bets the flop when he hits it will probably fold to far too many of your continuation bets”
I quess you need to modify a little the sentence to have probalities to make calculation:
Example
A Bayeasian bot sits down at HU table and make Bayeasian analisis for his opponnet.
The first hand his opponent folds to ths bot continuation bet.
Observation: He only bets the flop when he hits it.
For any bayeasian bot this information has valid basis instantly the bot make instantly guesses about the player profile:
Bot guess: There is (90%) chance that player(greenrock) who only bets the flop(30%) when he hits it will probably fold to (far too many) 80% of your continuation bets
And caluculates Bayes conditional probability.
Sorry First he calculates conditional probality and later he make guess.
Before calculates the bots needs to specify or calculate other probabilities:
-callculatee the probability that the opponent will fold regardless of being a “greenrock” or “notgreenrock”
-probability if he is a “green rock”, he will fold 80% to a continuation bet
I don’t think humans assume that other people behave the way they do. I think human beings form and hold on to stereotypical notions of “normal” or “standard” behavior for perceived groups (one such group being “poker players”).
To take your slowplay example, if your opponent doesn’t think you would slowplay a set, it’s not because your opponent never slowplays a set in that situation. It’s because he thinks the “standard” or “normal” play is to slowplay, and he has pegged you as a “normal” player for that game and those stakes. He probably has more granular categories than this – e.g., “normal good player”, “normal bad player”, “crazy Asian gambooler”, “retirement rock”, “drunken tourist”, etc. – but the point is that he’s ascribing to you behaviors that he associates with the group he has put you in. I don’t think it has anything at all to do with how he would play. If anything, I think most players imagine themselves to be better than most of their opponents and capable of making superior plays than “normal”.
Of course, he might well believe that slowplaying is the correct play, and therefore he would always do it himself. But that’s coincidental to his assuming you would slowplay.
Fair enough. I think the latter part at least holds true though: a guy can’t account for the possibility that you are doing something if it’s never occurred to him to do it. Thus a player who never turns a pair into a bluff will generally play in such a way that he is exploitable by players who will bluff with showdown value. Like the scrub-jay stealing food, it doesn’t occur to him that this is something he needs to protect against.
I pretty much agree on that, although we could quibble about what “occurs to him” means. I think human beings, unlike scrub jays, are occasionally smart enough to reason what their opponent is up to even if they had no prior awareness of the play and have not yet seen the cards. But I guess at that point, one could argue that it has just “occurred to them.”