Book review: Killer Poker by the Numbers

Yesterday while I had a bunch of down time with an inoperable computer, I took the opportunity to read one of my newest acquisitions, Tony Guerrera's Killer Poker by the Numbers.

I don't do many book reviews here. The two most prominent reasons for this are (1) I have a nasty habit of starting a lot more books than I finish, and generally I think it's unfair to write a critique of a book I haven't read all the way through (though there are exceptions, in which case I state explicitly that I'm reviewing just a particular portion), and (2) my slightly OCD-ish tendencies make me inclined to write reviews that are as long as the books. I want to list everything I like and dislike, agree and disagree with. That's too much work, so I often just don't even bother starting.

But in this instance, I don't have any trouble writing a generalized, summary review. I didn't hate this book, but I also didn't get anything out of it.

The basic idea is that he walks you through a bunch of poker situations of increasing complexity. With each one, he shows you how to estimate a range of hands that your opponent(s) might have, count the number of different card combinations that could make each of those possible hands, guess at an opponent's likely response to your decision to call or raise, figure out the various possible outcomes of the hand with their associated probabilities and the size of the resulting pot you will win or lose, then put all of those elements together mathematically to reach a conclusion about which move has the highest +EV.

But none of this is new to me. As far as I can tell, there isn't a single original concept here. Yeah, it's reasonably well presented, if you've never worked through this kind of complicated problem before. But once you've done it a few times, there's not much more to learn.

All such exercises have some value in just being able to do them in detail away from the table, with no time pressures. The practical problem comes in being able to translate the kind of theoretical exercise that might take an hour or two to work through into a useful shortcut that you might actually be able to employ in the 30 seconds or so that you typically have to make a poker decision in the real world. That is the gap that I had hoped this book would bridge, an impression that was bolstered by the descriptions of it I read on amazon.com. But it was not to be. I didn't finish the book feeling any better able to apply the math at the table than I was before starting it.

The problem, obviously, comes from the inability to put any decent estimate into what a particular opponent will do in reaction to a particular move that I make. For example, when I need to figure out the probability that a player will fold to a bluff or semi-bluff, I find that I am usually unable to do any better than "more than 50%" or "less than 50%." As Guerrera acknowledges, your overall computation can't be any more accurate than the least precise factor in the equation, so if you can do no better than a fairly wild guess at such a basic parameter as whether an opponent will fold to a bluff raise, it's pretty silly to spend a whole lot of time following an outcome tree to a series of all possible outcomes.

With most poker strategy books, a situation is described, and your options are explained, along with rough estimates of how opponents might react (with different approximate likelihoods assigned to different types of players), and the outcomes that one might expect. Well, KPBTN does the same, but just replaces guesses such as "usually" and "typically" and "often" with numerical estimates, then does EV estimations.

I'm just not convinced that the result is any better than the more traditional and generalized way of discussing hands. The inability to come up with anything more accurate than ballpark guesses of specific opponents' reactions makes the resulting math, well, not a sham, exactly, but not especially useful. When I read through the scenarios, I was able to pretty quickly come up with a list of possible lines of action to take and a gut sense of which ones I thought were most profitable. After Guerrera spends pages and pages slogging through the numbers, his conclusions tended to mirror what I had already decided would be the best approaches to the hand.

In short, for an experienced player who has played enough that pattern recognition is already acting as a shortcut substitute for an explicit, step-by-step, deductive process as to what an opponent has and is likely to do, KPBTN shouldn't be a step backward, but I also think it's unlikely to be much of a step forward. If you've never worked through the hard math of a poker problem decision tree, I think it's probably worth reading this book and forcing yourself through the calculations, because it makes you think explicitly about all of the possible outcomes and their relative likelihoods. That is undoubtedly good brain exercise. I'm just dubious that it will actually improve your decision-making process the next time you have a tough situation to analyze in the heat of battle over the green felt.

Maybe it just comes down to me being more of a "feel" player than a "math" player (which may also relate to why I do so much better live than online). I'm not intimidated by probability calculations (as I hope I've adequately demonstrated in previous posts). It's just that I'm not convinced that anything beyond fairly rudimentary pot-odds math is going to be of much practical help in most poker situations.

Almost two years ago, I wrote this post, in which I included mention of what Charlie Shoten tells himself when facing a poker decision: "I am calm, confident, and clear, and I wait for my best choice to appear after considering all of my choices and the consequences of each. When my best choice appears, I act." If you can do that, I don't think you need Guerrera's book, which basically just gives you the long way of getting to the same destination.