I liked this book, a lot. I've read a fair number of popular science/statistics/math analysis books, and the ones make an impression are those that fo...moreI liked this book, a lot. I've read a fair number of popular science/statistics/math analysis books, and the ones make an impression are those that focus on human stories. I hate to admit it, but as intellectually compelling The Wages of Wins is, Moneyball is a much better book, from a literary perspective.
As Michael Lewis confirms in an afterword, I find it incredible that had Billy Beane, the general manager of the Oakland Athletics organization in Major League Baseball, had any part in vetting the book, that he'd actually let it be released (Beane had no such power). The details that are in the book made me cringe, revealing such an intimate portrait of a still living figure. However, they serve a purpose, creating a parallel where the downfall of Beane as a baseball player reflects the ascension of statistics in talent evaluation in baseball.
I'm quite late to enter into the statistics vs. scouting arguments that have been made, beginning 6 years ago with the publication of the first edition of this book. And Lewis already noted how everyone had missed the point. It isn't that statistics is better than vision. It is that no one in power had the inclination to test, systematically, whether the current and traditional way of doing things was really the best way. Forget about testing any new system. I am not ignoring the free thinking managers and players who sometimes realized that there may be alternatives. The point is that the people bankrolling the team, signing checks, and are otherwise good business men, seem to be so disinterested in bringing applying their acumen into making something of their investment.
Lewis writes several times that some of the main characters do not give a damn about the numbers. Instead, they have a question, and they need to answer it; it so happens that statistical analysis is a pretty good tool for the questions they ask. Scratch that, actually, statistics can be used to good effect on just about anything; the trick and talent of analysis to phrase the question in a way that numbers can answer.
The book does detail some of the history and rationale behind on-base percentage, slugging, pitching stats, and fielding, but it creates dramatic tension - and clarity in presentation - by pitting scouts vs computer geeks. I think most people would agree that both have a place in the money game of signing ball players (and not just baseball, mind you.) The question is, how would you go about weighting stats and visual information?
I present one small detail from the book, that hasn't been discussed too often, I think, that speaks to this point. The numbers man comes up with a college player whose stats surpassed some threshold; he asks one of the scouts to go an see him. The scout hems and haws and shirked his duty, and so the kid was never looked at or interviewed. Somehow, during the annual draft, this kid was selected with a pick as some sort of olive branch from the scouts' side to the geeks' side (Apparently, someone actually made the call without consulting with Beane. This portrayal may be inaccurate, as Lewis is sympathetic to the geeks.)
Regardless, the point is that there are always other considerations. Some of them have to do with how much the kid wants to be paid, whether the prospect wants to play or go to college, whether he has injuries, what his competition may have been like, etc. Anyone who works with data knows that not all the points fall into a straight line. Sometimes, judgment is needed to find out how discrepancies arise. In other words, observations may not have been made correctly or just plain missed.
Thus the most important point in the book is not that Beane fired his scouts and brought in geeks; it was just that when faced with some observations to the contrary, scouts could not adjust. Again, one must consider that Lewis doesn't really defend the scouts, but at the time Lewis observed the A's, the scouts gave any number of reasons defending their guy. Numbers were used, but the scouts never really converted their opinions into a rigorously tested analysis. So the guy runs a fast 40 yard dash. So the guy is 6 ft tall and a muscular 220 pounds (that automatically inspires visions of a 50 home run hitter.) The scouts might also rely on some ranking that scores the man's hitting, throwing, fielding, and speed.
All this is fine, but you just think scouts would be more interested in figuring out how well their observations can be tied to player outcomes. Why is it that great prospects fail in the majors? Why is it that some unknowns can achieve so much? What is it that the scouts missed? More than anything, it is this inability to adjust to new information that Moneyball describes.
For instance, there are some numbers, such as "saves" and earned-run average, that aims to categorizes a pitcher's performance. However, no one ever provided an argument for why that number should be constructed the way it is and whether it in fact lends anything useful in analysis. In contrast, the idea of analyzing a pitcher's performance might just be simply decomposing his stats into walks, strikeouts, and home runs (but not hits he gives up!), which at least has the advantage of removing the effects of the pitcher's teammates on the field (i.e. if you have good defense, then you might give up fewer hits although other batter's make contact with you more often than other pitchers.) Neither stat could be placed into context until some sort of analysis was run, and Lewis points out that very few baseball people seem inclined to even begin testing these things. Just because you can count something doesn't mean it is useful. Again, the trick is to find out what your question is, find a tool that purports to answer it, and then re-assess your conclusions as you analyze whether that tool is the best one for the job.
I said at the beginning that because Lewis focuses on Billy Beane's story, this book became a good, literary read and not just an informative one. Billy Beane seems to be rather emotionally engaged in his job, and that made for some fireworks (yes, I admit that does make me a voyeur, delighting in the minutiae of some other person's job.) Likewise, the trajectory of Beane as a player and his self-analysis were also quite interesting. Specifically, one cannot help but think that Beane took delight in getting rid of the scouts as a result of his blaming his failure as a baseball player on the scouts. The system elevated Beane, it seems, because he was handsome, and he had athletic gifts. The scouts ignored data from Beane's statistics, although neither visual scouting or stats could have predicted the mental issues that led to Beane's failure to do well playing baseball. Although Lewis didn't present Beane's stats (and Moneyball analysis), a player like Beane is the opposite of the type of players that numerical analysis tends to favor. Lewis himself noted that Beane seemed to have made it a mission to find players that are anti-Beanes. The story uses Beane and his relationship with the expectations made of him to good effect.
Numbers do not predict with 100% certainty; anyone who lists the failures of prospects selected within a numerical framework misses the point. The point is that no one ever bothered to compare the old way of doing things with alternatives. As Beane realized, using stats driven talent evaluation certainly couldn't lead to a worse outcome.(less)