The Sabermetric Revolution: Assessing the Growth of Analytics in Baseball (5 page)

Moneyball
, in the final analysis, purports to be about how a team adopted a systematic, quantitative approach to player valuation that both reduced the team payroll and produced a winning team. It appears, however, that Lewis never convinced himself that Billy Beane was capable of working with such discipline and methodical commitment.
⁵²

Lewis makes two sets of predictions. First, he implicitly makes various predictions about the A’s selections in the 2002 draft, and also for David Beck and Jeremy Bonderman from 2001. As discussed, these have not stood the test of time.

Second, he blithely suggests easy success for J. P. Ricciardi in Toronto. Ricciardi, who also did not make it into the movie, was Alderson’s special assistant from 1996 until November 2001 when he was hired by the Blue Jays to be their GM. Lewis writes that J. P. Ricciardi is “now having a ball tearing down and rebuilding his new team along the same radical lines as the Oakland A’s.”
⁵³
He also quotes the Blue Jays then-CEO, Paul Godfrey, explaining why he hired Ricciardi: “Of all the people I’d talked to, J.P. was the only one with a business plan and the only one who told me, ‘You are spending too much money. . . . If you can stand the heat in the media, I can make you cheaper and better. It’ll take a couple of months to make you cheaper and a couple of years to make you better. But you’ll be a lot better.’ ”
⁵⁴
One of Ricciardi’s first moves was to hire Keith Law, a twenty-eight-year-old Harvard-schooled sabermetrician and writer for Baseball Prospectus. Soon thereafter Ricciardi reportedly fired twenty-five Blue Jays scouts.
⁵⁵
As we shall see, Toronto did shift toward a more sabermetric strategy under Ricciardi. The question is whether or not the new approach paid off.

During the two years prior to Ricciardi’s arrival, the Blue Jays averaged
81.5 wins and a $64.4 million payroll. During the eight years under Ricciardi, 2002–2009, the Jays averaged 80.25 wins and a $72.5 million payroll.
⁵⁶
The team did have a strong turnaround year in 2003, but that was due to an MVP year for first baseman Carlos Delgado, with an OPS of 1.019, 42 home runs, and 145 RBIs, as well as the emergence of star pitcher Roy Halladay and outfielder Vernon Wells. All three had been signed by the Blue Jays years before the arrival of Ricciardi. If there was some simple magical formula behind moneyball, as Lewis insinuated, then it eluded the Toronto Blue Jays.

Lewis begins
Moneyball
by saying he fell in love with a story and goes on to copiously cite statistics about financial and on-field inequality in baseball. He implicitly mocks MLB Commissioner Bud Selig’s comment that the A’s success was an aberration. Rather, for Lewis, it is a David and Goliath story, and Billy Beane together with moneyball slay the giant. Moreover, Lewis hints that with moneyball principles and mentally sharp GMs, baseball’s problem of competitive imbalance can be remedied. Listen to Lewis’s bold characterization of what Beane believed: “The market for baseball players was so inefficient, and the general grasp of sound baseball strategy so weak, that superior management could still run circles around taller piles of cash.”
⁵⁷

Even if the inefficiencies in the player market were so pronounced and the grasp of game strategy sufficiently undeveloped, and even if Billy Beane took advantage of this situation so that the small market A’s were successful, it would still be a dubious proposition that sabermetric smarts could promote better competitive balance. The reason is that Lewis would have made the baseball world aware of the magic formula and the big market teams could now exploit it as well as the small market teams.
⁵⁸
But, as we have seen, the available evidence does not support this premise.

Further, Lewis misapprehends the nature of imbalance in baseball. Not only does he misrepresent some relevant facts, but he seems to presume that, absent sabermetrics, high payrolls necessarily lead to winning teams.
⁵⁹
In fact, the correlation between team payroll and win percentage in baseball over the past twenty years is far from determinative. As we shall elaborate in
Chapter 6
,
the variance in win percentage explained by the variance in payroll generally has ranged between 10 and 35 percent, depending on the year. This, of course, means that between 65 and 90 percent of the variance in win percentage is explained by factors other than payroll (such as player performance, player injuries, team chemistry, intelligent management, or luck). That said, although high payroll does not guarantee success, it certainly helps.

The most common measurement of competitive balance in baseball is the standard deviation of win percentages across teams. By this metric, there is little empirical support for the claim that the spread of sabermetric knowledge after 2000 promoted greater balance. Using the ratio of standard deviation to idealized standard deviation,
⁶⁰
we find that the level of competitive balance in baseball was the same during 1980–1989 and 1990–1999 with a ratio of 1.7. In what Lewis would have us believe to be the first sabermetric decade, 2000–2009, the ratio actually jumped to 1.86, suggesting less balance.

If we look a little further into the dynamic of the period, we can discern an alternative explanation of the pattern. Consider the period 1996 through 2009, broken down by the underlying three collective bargaining agreements (CBAs), 1996–2002, 2003–2006, and 2007–2009. The competitive balance ratio went from 1.89 during 1996–2002, to 1.90 during 2003–06, to 1.68 during 2007–2009. Over the course of these three CBAs, revenue sharing from baseball’s rich to poor teams increased from an annual average of $107.1 million, to $278.7 million, to $396.3 million. That is, despite the almost tripling of revenue sharing between the first two CBAs, the measurement of imbalance actually increased (albeit marginally). Only with the third CBA, after a further, more modest increase in revenue sharing, did the measure of imbalance decrease.

What changed between the second and third CBAs? The incentive structure of revenue sharing shifted dramatically. During the first two CBAs, the marginal tax rates on local team revenue were actually higher on low revenue teams and lower on high revenue teams. This meant that, in addition to normal market factors, the low revenue teams had even less incentive to increase revenue than the high revenue teams. Between 2002 and 2006, every time a low revenue team increased revenue by a dollar, they lost approximately 48 cents in revenue sharing. In contrast, for every extra dollar of revenue that a
high revenue team generated, it paid 39 cents into the revenue sharing pot. Thus, the underlying incentives were perverse and moved the system in the wrong direction. Only with the new CBA in 2007 was this inverted incentive system corrected, and only after 2007 was there payroll compression across the teams. The coefficient of variation of team payrolls, after rising from .397 to .435 between the 1996–2002 CBA and the 2003–2006 CBA, fell to .405 during 2007–2009.
⁶¹
The turnaround in payroll disparity, in turn, helped to promote the improvement in competitive balance.

While it is true that small market teams that effectively practice sabermetrics have a chance to make up some of the lost ground from their revenue inferiority, large market teams also have the ability to practice sabermetrics. It is also true, as we shall see, that sabermetrics, sagaciously applied, can yield a healthy bang for the buck, and that small market teams have relatively more equal access to this resource than they do to the free agent market. Sabermetrics, then, might provide some impetus toward greater competitive balance, but the early and easy insights of baseball analytics have been exhausted. The challenges that lie ahead are formidable, so it would be prudent to expect no easy fixes. We will return to discuss related aspects of the competitive balance dynamic in
Chapter 6
.

While the book and the movie
Moneyball
can each be recognized as entertaining, they leave a distorted picture of the baseball industry and the impact of sabermetrics.

In our view, the explosion of sabermetrics in baseball (both outside and within team front offices) was a product of several forces. There was an intellectual trove of sabermetric analysis in the 1970s, waiting to be exploited. Bill James’s annual
Baseball Abstracts
helped to call attention to this work, as did some of the earlier on-field practitioners. The advent of free agency, in the context of industry-wide revenue growth, in baseball increased the average major league baseball salary from $51,501 in 1976, to $1.1 million in 1992, to approximately $3.3 million today. This explosion in player compensation naturally led front offices to seek more information on the best ways to evaluate
and to exploit player talent. Importantly, the mass development of the computer, the desktop and laptop, the iPad, the smartphone and the Internet have facilitated both the gathering and processing of statistics and the emergence of myriad statistical services, and have improved accessibility to baseball data and its analysis. Finally, for all of its deficiencies,
Moneyball
did tell a good story with some underlying validity, and it reinforced the objective forces and the momentum that were already in play.

Some have grumbled that all this data is turning baseball away from a game of instincts and emotions and into a game of boring mathematical precision. Veteran sportswriter Stanley Frank published an article in
Sports Illustrated
where he groused: “The greatest menace to big-time sports today is neither the shrinking gate nor TV. . . . It is a nonsense of numbers [and] the stupefying emphasis on meaningless statistics which is draining the color from competition.” Frank was writing in 1958. And Jim Murray, one of Bill James’s favorite columnists, wrote in 1961: “The game of statistics has begun to run away with the game of baseball. I mean, it’s not a sport anymore, it’s a multiplication table with baselines.”
⁶²

Yet, despite the statistical onslaught, baseball has only grown in popularity over the years. Statistical analysis is here to stay. The question for baseball front offices is how to most effectively combine statistical with traditional analysis. We hope to throw some light on this question in the chapters that follow.

We have called into question many of the assertions made in
Moneyball
, and examined their veracity with the benefit of hindsight. Nevertheless, the impact that
Moneyball
has had on the baseball industry is seismic, and undeniable. The book has been massively influential within front offices from coast to coast, and has been an important catalyst for the explosion of data and analytics currently roiling the larger sports world. In this chapter, we will examine the current state of analytics in baseball, and illustrate the role that
Moneyball
has played in bringing us to this point.

Lewis makes it clear that in 2002, the A’s were, if not the only team in baseball using statistical analysis to motivate their decision-making, certainly the most aggressive. Lewis asserts that “you could count on one hand the number of ‘sabermetricians’ inside of baseball.”
¹
As we argued previously, this is a loose interpretation of reality, since pioneers like Craig Wright and Eric Walker would have certainly qualified under that distinction in years past. Nevertheless, if Lewis is correct that there were at most five sabermetricians working inside baseball in 2002, how many are there now?

Unfortunately, a precise answer to this question is probably impossible, but in what follows, we derive an estimate. The first hurdle is to define who,
in fact, qualifies as a sabermetrician. What type of training is required? What job functions could she perform?

At this point, no accredited school or university offers a degree in sabermetrics or sports analytics. Moreover, one of the defining characteristics of the sabermetric movement has been the accessibility of the research conducted by those who are
not
writing for an academic audience. Most notably, Bill James’s work surged in popularity due to the quality, passion, and insightfulness of his writing, not his emphasis on careful estimation and revelation of his standard errors. (James succeeded because of, rather than in spite of, his lack of formal training in statistics, and Lewis concedes that he “sometimes did violence to the laws of statistics.”)
²

Conversely, some of the most statistically sophisticated work being done in sabermetrics today, such as Shane Jensen’s work on estimating fielding ability,
³
has failed to gain widespread acceptance within the community, but particularly within front offices, because you need something close to a Ph.D. in statistics to really understand it.