I recently had the privilege of reading the book and watching the movie, Moneyball. As a stat-headed baseball nerd, I thoroughly enjoyed both. For those unfamiliar with Moneyball, it is a book-turned-movie which also energized a new approach to baseball analytics under the same name. This new approach is presented in the book and movie as a sort of scientific, rationalist enlightenment in the world of baseball. Prior to the enlightenment, baseball managers, general managers and scouts were said to over-rely on statistics that were not particularly predictive of player value (e.g., batting average, RBIs, saves) while under-relying on more predictive stats (e.g., on-base percentage, slugging percentage, walks, WHIP), and paying too much attention to peripherals like body-type – including jawline… The Moneyball movement was about using rigorous data collection and analysis and the testing of scientific hypotheses about the role of certain baseball variables in terms of such big picture issues as game strategy and player value. As with other scientific revolutions, Moneyball led to challenging the authority of old ways of thinking, playing, trading, signing and drafting; many sacred cows were slaughtered. By my estimation, the Moneyball/Sabermetrics (i.e., advanced baseball statistics) movement has produced valuable advances in baseball playing and team building information and strategy.
This is not to discount traditional scouting. I think that the ideal path is the combination of advanced scientific and statistical analysis with expert experiential baseball knowledge. The baseball pro is invaluable in generating testable hypotheses, interpreting data, and responding when there are no numbers or no time for numbers, and so forth. The numbers approach is good because it adds a level of rigor that individual minds often lack and allows for consideration of volumes of data and variables that the individual mind would be hopeless to deal with.
Not everyone likes the Moneyball school of thought, however. As recounted by Michael Lewis in the new Afterword to Moneyball in a recent reprint of the book, many baseball insiders – baseball writers, managers, GMs, ex-ball players, etc. – aggressively scorned the book – usually without even having read it. Some, Lewis claimed, exuded pride in their having not read it and refusal to read it. Another set of detractors that do appear to have read the book exist in Sheldon and Alan Hirsch, who penned the book The Beauty of Short Hops: How Chance and Circumstance Confound the Moneyball Approach to Baseball.
Has anybody read this book? I haven’t, but am considering doing so.
The book, from the summaries I’ve read, argues that Michael Lewis’ Moneyball book (and the subsequent movie) ignore 1) the fact that, for non-moneyball reasons, the A’s had 3 of the best starting pitchers in the game at the time – Barry Zito, Tim Hudson and Mark Mulder, and that this was a massive contributing factor to the teams’ success; 2) that chance alone would allow even poor teams to put together a string of good years every now and then; and 3) that chance and circumstance (e.g., bad hops, fan interference, wind, etc.) play a big role in baseball that stats largely can’t account for.
I have to say that, based on the summaries and blurbs on the book authored by the authors, I have some strong skepticism regarding this book’s skepticism toward Moneyball.
Firstly, yes the A’s did have the three big pitchers. However, even after losing Jason Giambi (probably the second most dangerous offensive player in the game at the time), Johnny Damon and Jason Isringhausen, being forced by a tight wallet to replace them with low-rent players that the A’s statistical analyses suggested were tremendously undervalued on the free agent market, the A’s managed to outperform their previous year’s performance, prior to losing these three core players. What is more, they won 103 games (they’d won 102 the year before)! These are STUNNINGLY, EXCEPTIONALLY good win totals! Now, yes, chance, chance, chance. Sure, this team had one of the few lowest payrolls every year from 1999 to 2006 but never won fewer than 54% of their games during that period (this, by the way, is GOOD!), and had an average winning percentage near 60% (this is GREAT!), but yes, chance, chance, chance. Can’t discount chance. Plus, not all of their players were the product of Moneyball methods, and the team wasn’t nearly as poor just prior to this string of 8 consecutive strong years. The team has also done quite poorly since 2006 – though one could argue that this was due in part to their trade secrets having gotten out. So fine. But this is hardly the mother lode of my skepticism.
The Mother Lode of My Skepticism
Yes, chance and circumstance are big parts of baseball and they muddy up prediction, but that does not take away all the value of good data. Chance affects everyone and every game, but some players at the end of the day get on base more often than others (and thus get out less and score more, all else equal) – that matters. As was said repeatedly in Moneyball, Beane saw himself not as a fortune teller but as a card counter at a casino. Card counters don’t know the next card, but they can make probabilistic inferences. To the degree that a team’s management is able to identify, collect and mathematically analyze data that is keenly relevant to the winning and losing of games (or factors indirectly related to this – e.g., bullpen longevity across a 3-game series) or a players value in terms of productivity, the better off they will be, despite still being subject to chance. So long as it is used effectively and weighed in proportion to its epistemological value (that is, one does not under or over-estimate what they know), more information is good.
Can Advanced Baseball Statistics Make a Team Less Vulnerable to Chance?
It can be argued that a team run by card counters may be somewhat less susceptible to chance by virtue of having made more of the unpredictable predictable. A long time ago humanity had no idea what the weather would be like over the next day or two. Today, we still don’t know for sure, but we can make predictions that are more accurate than coin-flipping. That’s an informational advantage over people of the past that allows us to do a bit better than simply taking our chances on the weather. Good baseball data identification, collection and analysis could give the card counting team a bit more leverage over chance. There will still be bad hops, freak accidents, fan interference, pigeons in the wrong place at the wrong time, April flurries and powerful headwinds that will chance the game in ways that stats can hardly address specifically. But these factors affect every team; I’d rather be on the team with more information. What is more, the player with strong aggregate statistics is apparently playing better than other players against the broad spectrum of playing conditions and freak occurrences that take place over the course of the 162 game season.
If anyone has actually read the Hirsch’s book, or objects to Moneyball or anything stated here, please share your views.
PS: Since this is my first substantive baseball/Sabermetrics post, I’ll drop a few links to baseball blogs and sites I’ve been following so that interested readers can take a look at them and, hopefully, some of these bloggers will come here and share their views on this post. You’ll recognize immediately the team that I favour.