Sunday, July 02, 2006

Part II of a Review of Small Stakes Hold'em by Miller, Sklansky, and Malmuth.

The authors display a fundemental misconception about what's important in poker right at the beginning, in a section titled "Where the Money Comes From". (page 16-17.)

They correctly identify explotation of your opponents errors as the important source of expected winnings in poker. They get that right. But, they apparantly don't really understand it because they don't follow through.

Here's what they say.

"Money comes from exploiting mistakes. When your opponents make mistakes, you can make money" .....

Then in the next paragraph they say:

"People learn best through practice and feedback"....

So far, so good.

Then the next paragraph they say:

"This is the normal human process for learning. Unfortunately, it does not work at all for poker" (emphasis in original).

That's the point they dropped the ball. And I think they dropped the ball there in a major way.

They clarify what they're talking about in the next parapgraph where they say:

"The correct way to play poker is to understand it theoretically, and make sure you make the correct play, regardless of the results."

Okay, they're saying don't be results oriented, and we all understand that and we all agree with that.

But the most important part of playing winning poker is the identification of your opponent mistakes, not in determining an optimal explotation. Even if you don't always do the perfectly correct thing, if your opponents are making mistakes and you can identify those mistakes well enough to just make marginally correct response then you'll do alright.

Their error is in the emphasis of "does not work at all for poker.

The best way to learn to identify opponent mistakes is by practice and feedback and observation. It works very well as a learning mechinism for poker. It doesn't work for understanding the randomness of the cards that will fall, and it doesn't work as well as other methods for determing many nuances of optimal stratagy. But it works quite well for recognizing which opponents bluff to much and which opponents don't.

Experience and observation of the reactions to events of other players is key to playing winning poker. For the authors to dismiss that mode of learning right at the git go does not bode well for the rest of the book. It suggests there will be some major deficiencies in the analysis that doesn't really take full account of the range of responses that your opponents will make.

Stratagic thinking is the recognition that the opponent shoots back. The best way to learn about that is the method they dismiss as not useful at all.

We'll get a little deeper into the book in a later post.

Feel free to tell me where you think I'm missing the authors point in my observations.

Labels: ,


Blogger Chris said...

This comment has been removed by a blog administrator.

11:05 AM  
Blogger WillInNewHaven said...

I thought they made an important point about not learning poker by feedback. Short-term results can be so misleading. Learning the principles has to be more important than learning from a few trials.

On the other hand, I can see your thinking here. SOME things about poker can only be learned by this method.

4:18 AM  
Blogger Gary Carson said...

Thanks for the comment.

If they'd have said that some things are best learned by analysis rather than observation I'd have agreed. If they'd have said most things I wouldn't disagree.

But that's not what they said. They said nothing about poker is best learned by observation and that's just wrong. Flat out wrong.

Mason used to claim that 2+2 books never had mistakes. He backed down from that silly claim after a couple of years and more recenlty says that 2+2 books never make a mathematical mistake.

Well, this is a mathematical mistake.

It's not mathematical in the sense of 3+5 but it's mathematical in the sense of the proper role of mathematics in problem solving and the proper tools of mathematics to use.

There are many tools of mathematics which can be used in analysis of poker situations. the two being talked about here are 1. Simple probability models about card outcomes and 2. Statistical analysis of player behavior.

Number 1 does not require any observation -- it's strictly based on theory assuming random shuffles and independent events.

Number 2 does require observation.

Observation is of critical importantce and player reaction to your bets is part of the results of your bet. The only way to estimate the distribution of player reactions is from observation, you can't do it with simple probablity models, or game theory, or other such non-observatinal methods of analysis.

The passage we're talking about has Sklansky and Miller saying I'm wrong.

The idea that observation isn't important leads many players to making huge mistakes. You see that kind of mistake all the time on with questions like "how to play AK preflop", "how to play a middle pair", "how to play a flush draw", etc.

The questions are always about the cards, as if that's all that matters.

People get the idea that all that matters is the cards and that there's a single optimal way to play each combination of cards from passages like this one in Sklansky and Miller.

Short term results in can be very effective in learning which players bluff, which players never bluff, which players like to trap with big hands, etc. You don't have to watch a player very long to peg many of his playing habits. Not short term monetary results, but short term results in players reaction to your action.

6:43 AM  
Blogger WillInNewHaven said...

I think that their goals here were twofold.

First of all, they want to give people a reason to read this book, poker books in general. If the best way to learn poker is trial and error, poker books are a waste of time and money. That goal coincides with the best interests of the reader. A theoretical foundation will take you farther than reacting to each result as it happens.

It is also likely that they would rather readers not discard their concepts if they have some losing sessions as they apply them. Of course, the book is already paid for /ka-CHING/ but four has an interest in keeping people happy with their stable of authors.

10:20 AM  
Blogger Gary Carson said...

"First of all, they want to give people a reason to read this book, poker books in general. If the best way to learn poker is trial and error, poker books are a waste of time and money. "

You're probably right about that being their goal and about that being the way Miller, Sklansky, and Malmuth think about things.

But I don't think the statement "If the best way to learn poker is trial and error, poker books are a waste of time and money. " is accurate at all. The ability to make accurate observations and to analayze those observations in a way that draws useful conclusions is very much a skill that can be learned from books.

I think they and I have a difference of opinion about how to determine a useful way to go about mathematical modeling. The basic conflict is between doing an accurate model which you can only get an approximate solution for versus doing an approximate model for which you can get an exact solution.

They tend to opt for doing an approximate model for which you can get an exact solution. That's not my prefered approach, but there's nothing inherently wrong with that approach until you start spouting off that it's the only approach that's worth taking. That's where we differ.

They're saying "my way is the only way". I'm saying "their way is an okay way, but probably not the best way".

I'm going to explore the idea of exact models with approximate solutions versus approximate models with exact solutions a little deeper in a post on I think it's an important concept.

10:39 AM  
Blogger DMW said...

I reread this section Gary. I think it is clear they are not saying "don't observe your opponents"

The chapter limits their context to players who feel they are consistently outdrawn by loose opponents. They don't talk about observing opponents tendencies there.

10:03 PM  
Anonymous Anonymous said...

I have been looking for sites like this for a long time. Thank you! » »

1:27 PM  

Post a Comment

Links to this post:

Create a Link

<< Home