The Sabermetric Revolution: Assessing the Growth of Analytics in Baseball (6 page)

But we can adopt a good rule of thumb using Lewis’s implied characterization of Paul DePodesta as a sabermetrician. At the time, DePodesta was the assistant general manager of the A’s, the number two post in most baseball operations front offices.
⁴
Under this definition, DePodesta certainly would have been the highest-ranking sabermetrician in baseball history. DePodesta was an economics major at Harvard who “always had some sort of facility with numbers.”
⁵
He certainly has the quantitative background to perform and interpret linear regression models, but, like James, lacks formal training in mathematical statistics or advanced techniques at the level of, say, hierarchical Bayesian models. This background (undergraduate social science major at an elite university with some quantitative training) is now very common among front office executives at all levels of the baseball operations hierarchy.

Using DePodesta as a guide, we have estimated the number of full-time employees working in baseball front offices whose jobs appear to be primarily analytical in nature. The results (
Table 2
) were obtained by scouring team media guides and talking to those who work in the game. Of course, it is very difficult to know precisely who is doing what, but any employee in a
baseball department whose job title contains the word “analytic” or “statistic” is included. For the purpose of consistency, top-level executives who may be former analysts themselves, like Paul DePodesta, are not included. We have chosen to make this distinction since top-level executives are not primarily working on analytics. Rather, executives like DePodesta or Theo Epstein have no doubt incorporated sabermetrics into their philosophy, but delegate the actual number-crunching to lower-level staffers. (Later, we will examine how the assimilation of sabermetrics has paved the way for young executives like Epstein.) Again, it is important to caution that the breakdown in
Table 2
is meant only to be illustrative. There may be employees without sabermetric-type titles who do analytic work and employees with sabermetric-type titles who spend only a portion of their time on such endeavors. Further, if one were to ask a baseball operations employee at any team whether they practice sabermetrics, it is likely that the answer would be affirmative. To say otherwise would be tantamount to proclaiming you are against a healthy diet. The issue we are interested in, however, is not whether someone in baseball operations looks at statistics (of course, every team does this and always has), but the nature of the statistics that are considered, how they are analyzed, and what they signify in the organization. Based on our knowledge of front offices, we believe that an initial reasonable proxy for the sabermetric orientation of a team is whether or not positions are labeled analytic or sabermetric.
⁶
Also, keep in mind that teams regularly add and subtract employees in all areas, so what we have presented is merely a snapshot in time.

While in 2002, there were only a handful of people working in analytics, our research indicates that today at least seventeen teams employ more than one person who focuses on analytics in some capacity, while another five teams employ one person full-time and four more have someone working part-time in this area. It can be no coincidence that most of these people have been hired since the publication of
Moneyball
, and, in some cases, the connection is directly causal.
⁷
Further, some if not all of the teams in the “no analytical presence” column likely have a baseball operations person who spends time either doing her own analytical work or reading and sharing the work of other statistical analysts.

Table 2. Breakdown of MLB Front Offices by Number of Full-Time Employees Working Primarily in Analytics, 2012

Of course, counting the number of employees designated for analytics does not directly reflect upon
Moneyball
, since a primary innovation of the 2002 A’s was not so much the number of employees devoted to sabermetrics, but rather the influence that sabermetrics had on their general manager. Today, sabermetrics is certainly highly influential in many front offices (Tampa, Boston, Cleveland, Baltimore, the New York Mets and the Yankees, and of course, Oakland, to name a few), but is almost certainly less influential among others (e.g., Miami and Philadelphia).
⁸
But even among those teams that employ multiple people doing analytics (e.g., Milwaukee and Cincinnati), the influence of sabermetrics on the GM’s thinking may be slight.
⁹
We have also included some employees whose primary analytical medium may be video, as opposed to statistics.
¹⁰

We feel the inclusion of video analytics is warranted, because a growing trend in front offices is to replace the traditional advance scout, who follows the team’s schedule one or two series in advance and writes detailed reports on the tendencies of their upcoming opponents, with one or more front office employees who accomplish this same task by blending analysis of both statistics and video. While this change in policy is often written off as a cost-costing measure, the true innovator in this respect was likely DePodesta himself, who started his career as an advance scout for the Indians. He recalls that his motivation for sabermetric self-education grew out of a desire to overcome what he perceived to be his own shortcomings in traditional scouting. In any event, it is largely due to the ambiguity of the relationship between counting employees and the true development and emphasis on sabermetric practice that we seek a more objective and quantitative measure of team sabermetric intensity in
Chapter 7
.

Although almost all of those working in analytics hold an undergraduate degree, a master’s degree in statistics is not uncommon, especially among those who have risen to mid-level positions such as “senior quantitative analyst”
¹¹
or “manager, baseball analytics.”
¹²
Farhan Zaidi, who earned a doctorate in behavioral economics at UC-Berkeley before becoming the director of baseball operations for the Oakland A’s (usually the number three post), is the highest-ranking of just a couple of front office employees with a Ph.D.
¹³

Among the teams more aggressive with their employment, an expanded version of the R&D model described in
Moneyball
seems to have gained some currency.
¹⁴
Under the direction of former Goldman Sachs partner Stuart Sternberg, the Tampa Bay Rays now employ at least eight people contributing to their analytics department.
¹⁵
The top-down model consists of three systems developers, two baseball operations assistants, and two junior analysts working under the director of baseball research and development James Click (a Yale graduate who got his start writing for Baseball Prospectus, an online source for sabermetric articles and statistics). The Yankees, Indians, and Red Sox have a similar, but leaner, structure in place.

The attraction of working for a major league front office is obvious, but the depth of the pool of candidates may surprise some. Many candidates come from Lewis’s previous industry, finance, which certainly offers a broad
overlap in skill set but retains most of its talent by offering vastly more lucrative monetary rewards than baseball. Lewis describes DePodesta as fitting the finance industry mold, “but the market for baseball players, in Paul’s [DePodesta’s] view, was far more interesting than anything Wall Street offered.”
¹⁶
Sig Mejdal, now director of decision sciences for the Houston Astros, was a NASA engineer before descending on the winter meetings in search of a job. In the opposite direction, Nate Silver honed his skills in predictive analytics by building a projection system for Baseball Prospectus before becoming one of the world’s best-known political forecasters for
FiveThirtyEight.com
, and now the
New York Times
.

To develop what is one of the more compelling and accessible themes in
Moneyball
, Lewis goes to great lengths to characterize the philosophy of Billy Beane, DePodesta, and especially Bill James, as being antithetical to the conventional wisdom that controlled baseball. In Lewis’s view, although Beane and DePodesta were working within the sport, what they were doing was unusual, and in some respects unprecedented. James was a more typical outsider—a vocal minority preaching a gospel that fell mostly on deaf front office ears hiding behind “very effective walls [that] keep out everything.”
¹⁷
Lewis casts the conflict within the baseball industry as being between the “superior management” effected by Beane (motivated by the insights of James) and the Luddite conventional wisdom embodied by his scouts.
¹⁸
According to Lewis, Bill James believed that he is “right and the world is wrong,” and bemoaned the prevailing “anti-intellectual resentment.”
¹⁹
Thus, the adherents of sabermetrics, both inside and outside the baseball industry, were easily identified misfits relegated to the fringes of the established power structure.

In this respect, a great deal has changed since the publication of
Moneyball
, as believers in sabermetrics have been rapidly assimilated into the baseball industry. An understanding of baseball analytics was not merely a plus, but a requirement for the GM jobs that went to Theo Epstein in Boston and Andrew Friedman in Tampa Bay. In 2003 the St. Louis Cardinals hired Jeff Luhnow, a true baseball outsider with an MBA from Northwestern,
and tasked him with overseeing a team of statistical consultants. By the time Luhnow left the Cardinals to become the GM of the Houston Astros, he was supervising the Cardinals entire scouting and player development system and was touted for his “scouting acumen.”
²⁰
DePodesta now serves in a similar role with the Mets, overseeing amateur scouting and player development, but spends much of his time on the road, personally scouting amateur players. Keith Law was one of the few analysts working in baseball before
Moneyball
, having been hired by Toronto GM (and former A’s director of player development) J. P. Ricciardi. But since leaving the Blue Jays in 2007, Law has made a living writing for
ESPN.com
, drawing most of his material from scouting amateur and minor league prospects.

The surest sign of the assimilation of sabermetrics into the baseball industry is that it is no longer clear whose bread is buttered by statistics, and whose by traditional scouting. Carlos Gomez, currently the director of international scouting for the Arizona Diamondbacks, was a former minor league pitcher. But his path to the front office was paved writing articles for Baseball Prospectus on pitching mechanics.
²¹
Adam Fisher of the Mets, another Harvard graduate, has filled a variety of roles within the baseball operations department, including working in an analytical capacity in the front office, as a professional scout in the field, and as an advance scout combining video and statistical analysis. Galen Carr left Salomon Smith Barney in 2000 to join the Red Sox front office, but now serves as a professional scout based out of his home in Burlington, Vermont.
²²
These are just three more examples of people working within the game who have happily married the analytical mindset outlined in
Moneyball
as “heretical” with an understanding of traditional scouting set up in the book as adversarial.
²³
As we discussed in
Chapter 1
, the strict dichotomy between the two approaches depicted in
Moneyball
was always an inaccurate hyperbole.
²⁴

By 1988, Bill James had stopped publishing his annual
Abstract
, and yet by the time
Moneyball
was published in 2003, sabermetrics had already taken root among a few front offices and many more fans. Specifically, ground-breaking
theories like Vörös McCracken’s Defense Independent Pitching Statistics were communicated to Paul DePodesta via Baseball Prospectus.
²⁵
The void left by James was filled in print form by two SABR publications: the
Baseball Research Journal
; and
By the Numbers
, the newsletter of the SABR Statistical Analysis committee. But by the late 1990s, what is true today had already become obvious: the Internet had an unparalleled ability to disseminate ideas and information widely and quickly. Perhaps most important for a burgeoning, data-heavy field like sabermetrics, one could receive nearly instant feedback on statistical analysis from virtually anyone in the world who was interested. In this incubator, sabermetrics grew rapidly.

Critical to this explosion were two open sources of baseball data: the LahmanDB, a database curated by journalist Sean Lahman that contained statistics for every major league player in each season since 1871; and Retrosheet, a collection of archived play-by-play accounts maintained by biology professor Dave Smith. The former is small (on the order of several megabytes), but is well-packaged, easy to understand, and contains as much information as
Total Baseball
or any other desktop reference. While a math professor at Saint Joseph’s University, Sean Forman created a web front-end to the LahmanDB that became the hugely popular
Baseball-Reference.com
site. The website allowed anyone to quickly answer questions that would otherwise have required an extensive baseball card collection (e.g., in how many seasons did Mickey Mantle hit forty or more home runs?). The database, however, allowed sabermetricians to compute any statistic they could think of for any (and all) players in major league history and quickly sort the results. To illustrate the order of magnitude difference in computational ability at play here, imagine answering the question, “How many players with at least 1,000 hits have more home runs than doubles in their career?” There are 1,217 players with 1,000 or more career hits, so to count the ones who qualify using a website would take at least 1,217 clicks. But this question can be answered in seconds through a well-written query to the LahmanDB.
²⁶