The Sabermetric Revolution: Assessing the Growth of Analytics in Baseball (3 page)

Eighth, one of the movie’s more dramatic moments is at the end when Beane flies to Boston to meet with Red Sox owner John Henry at Fenway Park. Beane receives a job offer on a piece of paper from Henry. We are later informed that the offer was for $12.5 million. In the movie, Beane decides not to take the offer. In real life, Beane flew from the owners’ meeting in Arizona to meet John Henry at Henry’s winter home in Boca Raton, Florida.
Beane returned to the meetings and then flew home to Oakland to contemplate Henry’s offer (which was for $2.5 million annually over five years). The next day Beane called Henry and accepted. Champagne bottles were popped open on both ends of the call. The only issue remaining was agreeing on the compensation to the A’s from the Red Sox for the prospective loss of Beane. Henry waited for a call the next day from Beane to finalize the compensation. No call came and Henry tried in vain to reach Beane. Finally, Henry called Steve Schott (the A’s owner), who told Henry that he didn’t think that Beane was going to go to Boston. Later in the day Beane finally called Henry and confirmed that he was staying in Oakland.

Other than these minor misleading elements, the movie, albeit with some simplification and some added familial drama, is faithful to the book. However, both suffer from a fundamental misrepresentation of what happened in Oakland and its relationship to the principles of moneyball.

Lewis’s central argument is that the A’s, through the systematic application of new statistical analysis, were able to produce a winning team despite being from a small market and having an undersized budget. A collateral argument is that the underlying analysis was derived from the work of Bill James and that the broad use of the analytic techniques to evaluate players’ performance and game strategy was new to baseball. We will argue that the “new” Jamesian metrics, while not new in either conception or application, are indeed important, but they do not fully explain the A’s on-field success during the early years of the 2000s. That said, the A’s were an industry leader in adopting new metrics and identifying market inefficiencies, and some part of their strong performance under Billy Beane may be related to these factors. We will discuss the evidence of the A’s overall approach to player development and strategy in
Chapter 7
; presently, we will focus on the 2002 season, as does
Moneyball
.

In 2002, the A’s finished in first place in their division, winning over one hundred games, but were eliminated in the first round of the playoffs. Lewis describes the team’s success that season as well as the A’s strategy and analysis in preparing for the 2002 amateur draft. We discuss each in turn.

Although Lewis mentions in passing the contributions of A’s shortstop Miguel Tejada, third baseman Eric Chavez, and pitchers Barry Zito, Mark Mulder, and Tim Hudson, his principal and overriding contention is that the A’s success was tied to the incorporation of sabermetric insights. Yet Billy Beane’s player moves during 2002 did not build upon what was probably the most prominent sabermetric measure at the time, and the one emphasized by Lewis—OBP. Trading away Jeremy Giambi and Carlos Peña and getting back only John Mabry actually provided a net loss of OBP. The acquisition of David Justice and the team’s other minor moves did not compensate for this, so that the A’s OBP in 2002 was .339, compared to .345 in 2001 and .360 in 2000.
⁷
If anything, the team seemed to be moving in a counter-sabermetric direction in 2002, at least based on the OBP metric emphasized by Lewis.

An alternative explanation of the A’s 2002 success seems far more persuasive. The A’s had two of the strongest hitters in the American League: 2002 league MVP Miguel Tejada and Gold Glove–winning third baseman Eric Chavez. Tejada and Chavez each had 34 home runs, 131 and 109 RBI, and .861 and .860 OPS (a measure of offensive performance combining OBP with slugging), respectively. Both Chavez and Tejada are referred to in the book as “Mr. Swing at Everything,” meaning that neither one was patient at the plate and neither followed the fundamental sabermetric tenet of working a walk.
⁸

One of the most complex and inclusive measures of player performance is WAR (wins above replacement player), which we will discuss in
Chapter 4
.
⁹
Tejada and Chavez had a combined WAR of 9.2. Between them, they accounted for 41 percent of the team’s total WAR from its position players.
¹⁰

The A’s also had three of the top ten pitchers in the American League that year. One of them, Barry Zito, won the Cy Young Award. Barry Zito’s WAR was 6.9, Tim Hudson’s was 6.6, and Mark Mulder’s was 4.4, making the combined WAR of these top three pitchers 17.9 out of 25.3 for the entire pitching staff. Nearly all of the remaining WAR for the staff was accumulated
by three pitchers: fourth starter Cory Lidle (3.3), closer Billy Koch (1.7), and right-handed submariner Chad Bradford (1.5). No other pitcher’s WAR was statistically distinguishable from zero.
¹¹
Yet while Bradford received extended treatment (an exclusive, full chapter) by Michael Lewis, the other five were mentioned only in passing, despite accounting for 90.5 percent of the total WAR of the pitching staff. The selection of these five pitchers was based on traditional scouting reports and performance measures.
¹²

Bradford does have an interesting story and a unique delivery, but as we saw above, his contribution to the A’s success in 2002 was relatively minor. Lewis goes through significant contortions to establish Bradford’s bona fides. Thus, for example, he writes: “Dropping his release point had various effects, but the most obvious was to reduce the distance between his hand, when the ball left it, and the catcher’s mitt. His 84-mile-per-hour fastball took about as much time to reach the plate as a more conventionally delivered 94-mile-per-hour one.”
¹³
Sound implausible? It is. To increase from 84 mph to 94 mph is an 11.9 percent increase in speed. To achieve this, Bradford’s hand would have to be 11.9 percent of 60 feet and 6 inches closer to home plate when he released the ball than a typical pitcher’s hand. That is, Bradford’s hand would have to be fully 7 feet and 2.4 inches closer to home plate than a typical pitcher’s hand! Lewis attributes this closer proximity (equal to a bit more than Shaquille O’Neal’s height) to the fact that Bradford throws underhand. It is not obvious why the angle of delivery should affect proximity to home plate.
¹⁴

Last, consider the A’s performance in various statistical categories relative to the league average in 2002. The A’s OBP was 2.4 percent above the American League average and its OPS was 2.1 percent above. On the pitching side, the A’s pitching staff’s ERA+ was 18 percent above average, its strike out-to-walk ratio 12 percent above and its WHIP 7.7 percent above, all suggesting a stronger contribution from pitching than hitting to the A’s success in 2002.
¹⁵
Of course, an appreciable part of the team’s pitching performance was likely a product of the team’s defense—another central aspect of success virtually ignored by Lewis.

Lewis devotes well over a chapter to discussing the A’s 2002 June amateur draft strategy and picks. Again, the underlying premise is that Beane employed saber-savvy analysis to discern the diamonds in the rough and out-smart his competition. Lewis characterizes the A’s analysis of prospects prior to the 2002 draft as a collaborative process. But, says Lewis, the 2001 draft had been a disaster, and in 2002 Beane decided that he would remove the scouts and their conventional “tools”-based evaluation from the process, and rely entirely on his statistical analysis.
¹⁶

Although Lewis sat in on some of the A’s strategic draft discussions and he extols the wisdom behind the identification of the team’s top choices, Lewis published
Moneyball
in 2003, prior to the maturation of the A’s 2002 top picks. The 2002 draft promised to be a fruitful one for the A’s, who were in the unique and highly enviable position of having seven first-round picks. If Lewis’s theory were correct, ten years later we should have witnessed the flowering of these young prospects. What is the record?

Using sabermetric principles, Beane and DePodesta identified eight pitchers and twelve position players whom they would draft in a perfect world, if money were no object. The A’s signed thirteen of these twenty players, an unexpected coup that left Beane ecstatic. One of the A’s seven first-round picks was catcher Jeremy Brown. Brown is singled out by Lewis as an especially enlightened choice by Beane; indeed,
Chapter 5
is entitled “The Jeremy Brown Blue Plate Special.” How did this enlightened pick turn out? Not so well. In his professional baseball career, Brown had a total of ten major league at bats.
¹⁷

The A’s third pick in the first round was a shortstop named John McCurdy. McCurdy never played a major league game. Of Oakland’s remaining top thirteen picks in 2002, only Nick Swisher (the A’s first pick) has had a clearly above-average major league career, and, as Lewis acknowledges, Swisher was a favorite of both Beane and the scouts. Of the other ten players, only three (Joe Blanton, Mark Teahen, and John Baker) have played significantly in the majors.
¹⁸
DePodesta thought that one of the seven players who never made it to the Show, Brant Colamarino, “might be the best hitter in the country.”
¹⁹

A fair accounting of the
Moneyball
draft, with the benefit of hindsight,
reveals a mixed bag, rather than the coup that is depicted by Lewis. Using data from the sixteen amateur drafts during 1990–2005, we constructed two models for the expected return—in terms of WAR accumulated during a player’s first six major league seasons—of any of the first 220 draft slots in each year.
²⁰
In essence, the first thirteen selections the A’s made in 2002 accumulated 25.2 WAR before reaching free agency—this is an estimate of the value that they provided to the club. Our first model, which assigns a WAR of 0 to every player who never reached the major leagues, suggests that the A’s should have expected these draft slots to earn about 18.8 WAR. This implies that the A’s did better than expectations given their draft slots.
²¹
In contrast, our second model simply ignores all players who never reached the major leagues, which raises expectations. In this scenario, we compare the A’s drafted players who made the majors to the expected WAR of drafted players who made the majors on other teams from similar draft slots (23.2). By this method of accounting, the A’s did just slightly better than expected (
Table 1
).
²²

But while the A’s 2002 draft slightly surpassed expectations, the draft fails to bolster the case that statistics should supersede scouting in the evaluation of amateur players. Fully 89 percent of the WAR accumulated by these thirteen players drafted by the A’s was earned by just two players: Swisher and Blanton, both of whom were first-round picks on the basis of traditional scouting. Swisher—the son of a former major leaguer—was a no-doubt first rounder coveted by several teams. Chicago White Sox GM Kenny Williams, a former player who is notably agnostic toward sabermetrics, told Beane he was going to draft Blanton.
²³

Table 1. WAR of Players Drafted by Oakland A’s, 2002

Lewis also depicts a sharp conflict between Beane and scouting director Grady Fuson over amateur pitcher David Beck. Not surprisingly, Beane prevails and Beck is signed by the A’s. As evidence of the perspicacity of Beane’s statistical methods, Lewis gloats that Beck “went out and dominated the Arizona rookie league.”
²⁴
Unfortunately for the A’s, the rookie league was the only thing Beck ever dominated. He never pitched above single A ball and was out of baseball by 2003.

According to Lewis, Beane didn’t have “the slightest interest” in Scott Kazmir, because he was a high school player.
²⁵
Kazmir was a dominant starting pitcher for several years for the Tampa Bay Rays, with an average WAR of 4.3 during 2005–2007. While the A’s were willing to indulge in the unathletic Jeremy Brown, they were uninterested in future all-star Prince Fielder, who was “too fat even for the Oakland A’s.”
²⁶
Fielder had a pedigree that was even better than Swisher’s, as the son of former home run champ Cecil Fielder.

The A’s took the sabermetric observation that college picks had a higher success rate than high school picks (principally because the level of competition was higher in college and, hence, player statistics were somewhat more meaningful) and made it religious dogma. No matter how much the tools of a high school player impressed the scouts, Beane wanted nothing to do with them. Lewis reports that after high school pitcher Jeremy Bonderman was selected by the A’s as the twenty-sixth pick of the 2001 draft, Billy Beane threw a fit: “Billy erupted from his chair, grabbed [the phone] and hurled it right through the wall.”
²⁷
Beane traded Bonderman to Detroit during the 2002 season (in the Carlos Peña deal). Bonderman went on to have a solid eight-year career with the Tigers, averaging a WAR of 3.9 during 2004–2005, and starting Game 4 of the 2006 World Series.
²⁸