Quote:
Originally Posted by shysti
I mean truly post flop play has to be learned from experience but he's written a book about it, so how do you guys feel about that?
I believe that the reason most people feel that great post flop has to be learned from experience is due to the nature of the problem.
The correct (best long term EV) course of action (fold, call, raise (and by how much)) are the output of a function driven off a large number of dependent variables. These types of problems are often solved through a set of techniques called regression analysis. Regression analysis involves looking at a large sample of the values for these variable and the correct corresponding outputs. When done properly, a model is developed that can be used to predict the appropriate outputs for other input values.
Fortunately for humans, our brains are built around neural networks. Neural networks, both biological and artificial, are particularly adept at solving problems that require regression analysis. This is one of the reasons that artifical neural networks and human brains 'learn' through experience. Each 'experience' provides another sample allowing us to refine our model.
This has an upside, through regression analysis humans effortless solve enormously complex equations. For instance, how much force should I be applying to the accelerator and break pedals as I drive. Take a moment and consider the ludicrous complexity of that equation and then compare that with the fact that most of you drive thousands of miles a year without incident.
This has a downside, from a conscious standpoint we are often completely unaware of how we solve the problem; from the standpoint of what variables we take into account, what weight we apply to these variables and how we believe these variables interact.
Because of the upside, some people are very good at solving the poker postflop play problem. Because of the downside, those same people are often in no position to explain why their play was correct (even though it was.)
As we are often unaware of the nature of the model we have created, we sometimes begin to believe that there is no model; we 'magically' feel the right answer due to our 'experience'.
But it is a model and can be expressed in a readable notation.
Authors like Harrington try to relate the primary driving variables, their general weights and their general inter-relationships. They then wisely put in a caveat that any given situation may not follow their simplified model. Because the All-In re-raise pre-flop from the super tight guy DOES have a different meaning if we see an ambulance outside the building, an EMT has just walked up and whispered something into the player's ear, upon which his face went white as a sheet, and he threw a panic stricken glance at the ambulance right before he pushed. Your neural net knows this, and has a variable for 'opponent is experiencing horrific personal tragedy' and has applied an enormous weight to that variable. But in practice, that variable rarely comes into play so it would be pointless for a poker book to mention it.
But for a lot of people whose internal model is not yet sufficiently complex, these guides can allow them to deductively arrive at a reasonable answer. All the while, most would agree that later in their career after they have sufficient experience, the player will likely intuitively arrive at a better answer.
