The one thing good leaders have in common is a willingness to let new evidence change their views.—The Economist, citing a study by the executive search firm Korn/Ferry
In the spring of 2003 author Michael Lewis created a sensation in the baseball community with the release of Moneyball, a book that relates the story of how Oakland A’s general manager Billy Beane kept his undercapitalized team competitive. A onetime financial trader turned writer who had unprecedented access to Beane’s activities, Lewis identified two related causes for the A’s success: Beane understood the concept of market inefficiencies and the analogous benefit of finding undervalued players, and Beane believed that these players could be better identified using statistical and analytical techniques than by traditional scouting methods.
The second issue caused a fair bit of controversy. Lewis is a terrific writer, and much of the charm of his book came from the dichotomy he drew between the old-school scouts and the newfangled statistical analysts. As Lewis put it, “Billy had his own idea about where to find future Major League Baseball players: inside [assistant general manager Paul DePodesta’s] computer. He flirted with the idea of firing all the scouts and just drafting kids straight from Paul’s laptop.”1 The two most prominent statistical insights of the A’s, as highlighted by Lewis, were that most organizations undervalued hitters with a high on-base percentage who did not otherwise stand out and that teams undervalued college players when compared to high school players in the amateur draft.
Baseball statistical analysis had been evolving and developing for roughly fifty years and had begun to find an audience with the writings of Bill James starting in the late 1970s, but this audience mainly consisted of independent researchers and a particular type of fan. Sabermetrics, a word coined by James, did not prescribe a set of formulas and answers, as its critics might have thought. It is a process, a philosophy that teams should make decisions based on evidence and data. This was not a new idea—scouts had been using radar guns and stopwatches for decades rather than merely trusting their eyes—but sabermetrics suggested that baseball’s vast statistical record could tell a team which players were actually helping the team score or prevent runs, which strategies would increase the team’s chances of winning, which Minor Leaguers were likely to be good Major Leaguers, and more. Much more, in fact.
By the 1990s sabermetrics had begun to creep into some of the more progressive baseball front offices. For example, Yankee general manager Gene Michael stressed several key statistical areas, including on-base percentage, when he rebuilt the Yankees early in the decade.2 More analytically, Rockies general manager Dan O’Dowd and Major League administrator Thad Levine were making sophisticated mathematical evaluations of the effects of their high-altitude Coors Field in 1999.3 But most teams, before the publication of Moneyball, kept their analytical efforts out of the public eye. Not surprisingly, Lewis’s portrayal of a general manager who seemed to be rejecting one hundred years of supposedly hidebound traditionalist scouting in favor of novel statistical methods created a rift between the proponents of traditional scouting and statistical analysis.
The escalation of data-driven analytics in the evaluation of players and game strategies created a new opportunity for baseball teams to gain an advantage over their peers. The team most associated with analytics after the A’s was the Boston Red Sox, whose principal owner John Henry made a fortune as a commodities trader by taking the emotional element out of trading decisions. He hired James as a consultant and a young Yale graduate, Theo Epstein, as general manager. The Red Sox success—they won the World Series in 2004 and 2007—along with improvement in other analytically associated teams like the Cleveland Indians, led to a more widespread acceptance that analytics could offer important insights into both the understanding of a player’s value and how that player’s value was likely to evolve.
Pat Gillick, who had earned his formidable stripes as a scout, was not an early adopter of statistical analysis. For people like Gillick, Lewis’s unflattering portrayal of traditional scouts poisoned even his more compelling statistical arguments. Because Gillick’s Mariners were also competing directly with the A’s in the AL West, Gillick managed to get himself into something of a feud with Beane and Lewis.
“I thought the way the writer treated people in the book was in poor taste,” Gillick said. “That’s why I wouldn’t buy it. I’m very, very disappointed that anybody would take shots at people as were taken. They took a shot at one of my guys right here, Roger [Jongewaard, the Mariners’ longtime vice president of scouting and player development, who originally scouted and signed Beane].” Gillick did not let Beane off the hook, either. When asked if Beane’s interpretation might have been misrepresented, Gillick replied, “He was not misquoted for 200-and-some pages.” To Beane’s credit he did not run from his comments: “I never said I was misquoted. I don’t feel the need to pacify others. Again, my energies, as they always have been, are on trying to compete, with the lowest payroll in the division.”4
“You have to give them credit, but the test is going to be how they maintain it,” Gillick said. “It’s difficult, with that payroll, to maintain. We’ll have to wait and see. Initially, they’ve gotten it done, but once [shortstop Miguel] Tejada and [third baseman Eric] Chavez are eligible (for free agency), if they can’t pay them, you’ll probably see a decline in their won-loss record.”5
Lewis thought that Beane’s advantage would dissipate for a different reason: because other teams were going to start mimicking his strategies. “He [Beane] may feel pretty happy with himself now, because his team reflects inefficiencies exploited in the past, and looks pretty damned good. He might even get through this whole year without having to use the trade deadline, one of his favorite things. But two, three years down the road, he has problems.” More specifically, Lewis concluded, “I don’t think [Beane] gave away all his secrets. They still have secrets; but I don’t think they have secrets from the Red Sox. They certainly don’t have secrets from the Dodgers. But the Dodgers may have secrets from the A’s, because I don’t think Paul [DePodesta, whom the Dodgers hired away as general manager] coughed up everything he knew to Billy.”6
Beane, too, understood this predicament: “Time will tell. Listen, at some point, yeah, there will be a dip in the performance level. You don’t have to be Nostradamus to predict that. But you know what? That’s the case with every sports franchise. It doesn’t take a genius to figure that out.”7
Gillick also added his thoughts on some of Beane’s specific theories as translated by Lewis:
I think from the Double-A level up, statistical information is more pertinent than from Single-A down. At least when you get to Double-A, Triple-A and the Major Leagues, you have something to compare the statistics against. I’m not sure what the level of competition is at Class A, rookie or amateur level. I don’t want to limit myself in one area. If we think the best player is from college, we’ll take him; if we think the best player is a high-school player, we’ll select that player. If we think we want to get a player from China, Japan or the Dominican Republic, those are all areas you have to investigate. Why limit yourself to one area? Why say you have to draft college guys and you have to fit this criteria? I think you’re limiting yourself and not looking at the big picture. Baseball is full of exceptions and opinions. . . . They have a theory what they do, but I think what they’re doing is limiting themselves, maybe because of economics. They think high-school kids are too much of a longshot, too much uncertainty. But the old saying is, if you want to hit it big, you’d better take a risk.8
For Gillick, this was another side of his “many rivers” philosophy of finding talent: be prepared to use all the information you have available to you.
But what Beane, through Lewis, was saying was not that high school players should not be drafted, but that they were currently overdrafted, and college players taken at the same draft position were a better risk. If all teams made this determination at once and shifted their strategies, high school players could suddenly become underdrafted, and a market-driven team would then shift its focus to high schools.
As with much of Lewis’s book, this point was not sufficiently understood in the debates that ensued.
In the decade after the publication of Moneyball, analytics has gradually become much more sophisticated and ubiquitous. “We go after players we feel have a positive residual, guys we like better than everyone else,” explained A’s director of baseball operations Farhan Zaidi in 2013. However, Zaidi further points out that it is important to recognize “what data is commoditized, and what data really gives you a competitive advantage. Knowing that—knowing when you’re using data that other teams have access to, versus data that is legitimately proprietary—is an important point to be able to recognize.”9
Statistical analysis can also be used to evaluate in-game situations. The mountains of data rapidly becoming available allow comprehensive analysis of on-field events such as batter-pitcher matchups, strategic decisions such as bunting, and defensive positioning. As the front offices in some of the more statistically savvy organizations better understand these relationships, there has been a natural tendency to impose some of this knowledge on the manager. Not surprisingly, this has occasionally upset the barrier that has been observed between the front office and the field staff for nearly a century. “You’re the manager and you’re going to get no interference or second-guessing from me,” Yankees general manager Ed Barrow told manager Miller Huggins in the 1920s. “Your job is to win, and my job is to see that you have the players to win with.”10
Analytics has changed this relationship; the front office now has information that might contradict what a manager ordinarily would want to do. As one writer recently observed, “Teams don’t want a seasoned, master tactician anymore so much as they want a manager with a small ego and an open mind. At the root of this change is the proliferation of statistical analysis, which can make decisions for managers if they’re willing to embrace it.”11 Lewis described Beane’s preferred approach in Moneyball: “Beane ran the whole show. He wasn’t just making the trades and supervising scouts and getting his name in the papers and whatever else a GM did. He was deciding whether to bunt or steal; who played and who sat; who hit in which spot in the lineup; how the bullpen was used; even the manager’s subtle psychological tactics.”12
This was clearly the extreme, but other teams that put more emphasis on the statistical side also made in-game suggestions. In Boston, “in those first years they would have guys who would send me lineups,” manager Terry Francona later complained. For example, “they would tell me not to hit David Ortiz against Scott Kazmir because chances are David’s going to have a rough night. Well, I’m not sitting David. He’s got a chance to be MVP, and you want me to start Doug Mirabelli at DH because Doug has better numbers against this guy?” After winning the 2004 World Series Francona had earned enough standing to push back. Nevertheless, his statistically inclined owner continued to pepper him with advice gleaned from an analysis of the team’s statistical research. Henry would often send late night emails, “usually asking why certain decisions were made that ran contrary to the imposing database maintained by the baseball operations staff.”13 (We discuss the Red Sox at greater length in a later chapter.)
Over the next several years many teams employed a staff of analysts, and each club had to reach its own internal accord over its primacy. Some depended on it more than others, but nearly all recognized that scouting and sabermetrics were not mutually exclusive and that some level of analytics was valuable. The most statistically inclined teams often looked to hire managers who embraced the new techniques, perhaps illustrated most notably by Joe Maddon in Tampa Bay.
By 2014 every big-league team had wealthy ownership, smart professionals in the front office, and well-organized player-development systems. The new stadium binge that saw nearly every team in baseball get a new or remodeled ballpark, and its associated increase in revenues, had run its course. Toronto, Baltimore, and Cleveland, which won in the 1990s with smart organizations and new stadium revenues, can no longer boast either advantage over their competitors.
In the aftermath of Moneyball, most teams have staff dedicated to analytics looking for strategic advantages over their competition. An indicative, but far from comprehensive, survey of various teams and technologies testifies to the sophistication, variety, and ubiquitousness of the advanced analytical techniques and concepts. The New York Mets are one of seventeen organizations (as of 2013) that employ TrackMan, a ball-tracking technology that uses Doppler radar to track the baseball. Such 3-D tracking systems that can capture the location of the ball and players at near-continuous intervals are now scheduled for installation in all the Major League parks by 2015. “A guy could be throwing 90 miles per hour with 7 feet of extension, and he gets the ball to home plate quicker than a guy throwing harder that doesn’t release the ball as close to home plate, essentially redefining velocity,” said Josh Orenstein, the company’s director of baseball operations and analysis.14 Another baseball executive noted that his club was using the technology in scouting, Minor League instruction, and the Major Leagues.15 Sandy Alderson, the Mets’ GM and the man who hired Beane in Oakland, agrees: “This is quite simply going to add immeasurably to the amount of information that’s available. To the extent that things become more granular, then we make fewer inferences as to what actually is going on. The critical thing is to be able to use the data in such a way that ultimately it can be used either in terms of player evaluation or even player education or instruction.”16 Such data-intensive video requires high-speed computers and new analytical methods to study the “big-data” output from this and other initiatives. One team, currently unidentified, purchased a Cray supercomputer, reported to cost a minimum five hundred thousand dollars, for analyzing the forthcoming data explosion.17
The Tampa Bay Rays, owned and managed by former Wall Street executives and investors, assembled a front-office staff tasked, in part, with uncovering market inefficiencies and unearthing new insights for finding, improving, and utilizing baseball players. As detailed in a recent book by Jonah Keri, the Rays’ staff includes James Click, who created a sophisticated database to track all available information on each player; physicist and mathematician Josh Kalk to tease insights out of the PITCHf/x data (a new technology discussed in greater detail in the next chapter); and sports psychologist Dr. John Eliot to better understand the mental side of player-performance fundamentals and the nebulous concept of clubhouse chemistry. Some of their findings, notably in the area of defensive positioning and the results of batter-pitcher matchups, have become integrated into the team’s on-field strategies.18
In 2012 the Orioles hired longtime big-league pitching coach Rick Peterson as director of pitching development, focusing on biomechanical studies of pitchers’ motions. He has long been a proponent of understanding pitching mechanics and recently teamed with famed sports orthopedic surgeon Dr. James Andrews to better understand them. “It’s based on Dr. Andrews research at ASMI (American Sports Medicine Institute),” says Peterson, “to get an MRI [magnetic resonance imaging] of the pitching delivery to make sure that the measurements in that delivery are falling into normative range to optimize performance and reduce the risk of injury.”19
The struggling Houston Astros hired Jeff Luhnow out of the Cardinals’ organization after the 2011 season, and he brought along others from St. Louis. He had video coordinator Jim Summers analyzing the vast amount of available video information. The team uses data generated by Inside Edge and runs it through a program made by Sydex Sports. For example, the Astros compile every pitch for the past few years thrown by opposing pitchers and scrutinize the data with the team’s hitters. “The right-handed batters will say ‘OK,’ let me see this period of time against right-handed batters,” Summers explained. “Now we’re going on proper preparation to see the ball. We know what’s coming. We know that on a 2–1 count, 37 percent of the time a pitcher is going to throw a fastball in a certain area. We can prepare for that. The statistics and video have broken it down.”20
In Chicago new Cubs president Theo Epstein teamed with Bloomberg Sports to create a first-class player-evaluation system. “The management and analysis of data,” said Epstein, “whether it be scouting reports, statistics, medical information or video, is a critical component of our operation. We look forward to developing a customized program that utilizes the most advanced and efficient technology available in the marketplace today to facilitate quicker, easier and more accurate access to all the sources of information we use to make baseball decisions.”21
The Tigers hired a performance-enhancement instructor, as of 2011 one of only a handful of teams to employ such a specialist. In this role Brian Peterson, who holds a master’s degree in counseling psychology, tries “to help all of the players, in the entire organization, be clear of mind while they’re going about their business.” Peterson might be approached by a hitter who says, “Geez, I’m 2-for-50; what am I going to do?” As Peterson sees it, “The majority of time there is some sort of personal issue that is creating some type of emotion, and their coping mechanisms maybe aren’t quite as good as they could be. I simply try to help them by giving them some good coping mechanisms and some good direction so they can be clear of mind while they’re going about their business.”22
In 2013 the Pittsburgh Pirates made the playoffs after twenty consecutive years of below-.500 finishes, one of the longest runs of futility in American sports history. Some of their turnaround is due to simply finally succeeding with their long run of high draft choices. But the Pirates actively secured this young talent by spending more money on amateur signing bonuses than any other team in baseball between 2007 and 2011.23 The team also started applying thoughtful analytics to uncover possible competitive advantages. After studying the key elements of successful defense, the team began using more targeted defensive shifts and getting their pitchers to throw more pitches that could induce ground balls, an approach that required a buy-in from manager Clint Hurdle.24 (In fact, a baseball executive told us in 2014 that convincing pitchers of the benefits of the new defenses was often a challenge.) Under their new program the Pirates improved their defensive efficiency from last in the league in 2010 to fourth in 2013 and surrendered the second-fewest runs.
Nationals general manager Mike Rizzo, the son of a baseball scout, expanded his organization’s analytical capabilities. The widely second-guessed decision to limit pitching phenom Stephen Strasburg’s innings in 2012, his first year after coming off Tommy John surgery, was the result of extensive study of previous cases. Another medical study led the Nationals to bring in a doctor to evaluate and monitor blood nutrients in the players.25
Rizzo also worked to integrate the scouting and analytics departments. During the 2012–13 offseason the Nationals were “looking heavily at the left-handed relief market,” director of baseball operations Adam Cromie told the Washington Post. One of the team’s scouts framed the question: “What is the profile of a left-handed starter who has success after moving into a relief role?” To help answer the question Cromie and Samuel Mondry-Cohen analyzed both statistical information and scout-derived evaluations. “That’s something that was very much driven by a scout,” Cromie said. “A scout came to us and said, ‘Look, this is kind of to me what a starter who fits into a relief role looks like. What does it look like to you? And how can we put those two things together and come up with a list of players who fit into that profile?’ It was a very interesting project, and I would say the results are very interesting, too.”26
In the decade after Moneyball most teams struggled with how deeply and broadly to embrace analytics. The debate in the immediate aftermath of the book was between those who supported the traditional scouting model and those who thought, as Beane did in Lewis’s book, that sabermetrics could dramatically reduce the need for scouts. The smartest and most successful teams, as it turned out, grew their analytics staff to provide information that could enhance and augment what their scouts were telling them and that, in the ideal environment, the scouts and analytics staffs could work together and learn from each other. Scouting was not going away. Nor was analytics.