The MVP Machine

by Ben Lindbergh

  In each of these cases, and many more, a player made the choice to use new methods and technologies to systematically address his deficiencies. Sometimes it was a mechanical adjustment that unlocked the latent power in a swing. Sometimes it was a more strongly instilled sense of the strike zone. Sometimes it was training to add velocity a pitcher didn’t know his body was capable of producing. Sometimes it was a pitch designed from scratch or promoted from secondary status to a more prominent role. And sometimes it was a modified mindset or meal plan or gym regimen. These overhauls are happening in hundreds of places across the sport, from professional clubhouses, bullpens, and batting cages, to colleges, high schools, and international leagues, to the independent petri dishes where this drive to reconceptualize talent began: boundary-breaking facilities outside of the professional game. Curious, scuffling players linked up with little-known coaching iconoclasts to spark a revolution. Now some savvy MLB teams are taking their insights to scale and lapping the rest of the league.

  Veterans who’ve looked lost are reclaiming careers, while an emerging generation of information-friendly players is seeking out stats from the get-go, fueling a youth movement in the majors and contributing to a constantly increasing level of play. “During the ’80s and ’90s it was steroids,” says Seattle Mariners director of player development Andy McKay. “And now it can be new information.”

  Mainstream baseball commentators haven’t quite figured out how to talk about this new era in baseball development. On almost every broadcast during the 2018 playoffs, national commentators fretted about jargon like “launch angle” and “spin rate,” lamenting the game’s new scientific focus. But though the language is new, these terms don’t describe new phenomena: Babe Ruth’s batted balls had a launch angle, and Bob Feller’s fastball had a spin rate. In earlier eras, there was just no way to track them. Today’s technology tracks everything, allowing progressive players to dissect their performance with unprecedented depth. The better they understand their current technique, the easier to analyze how it could be better.

  Not every player wanders from team to team until he’s bitten by a radioactive hitting coach and triples his home-run total or meets a sabermetrician on the road to retirement and suddenly sees the light. Enough of them have, though, that it’s swaying player performance on a league-wide level—changing the composition of coaching staffs, scouting departments, and front offices; altering the way general managers construct rosters; popularizing formerly frowned-upon training techniques; and determining who wins the World Series and individual awards. Even in its early stages, though, this movement is also raising privacy concerns, exacerbating baseball’s anti-spectator trends, and possibly leading to labor strife.

  On a more fundamental, broadly applicable level, it’s overturning old beliefs about the immutability of talent. In baseball’s old-school scouting parlance, “guy” is a versatile label, employed, one scout says, “like how Smurfs use the word ‘smurf.’” A non-prospect is not a guy, or (said dismissively) just a guy; a prospect is a guy; and a top prospect is a “GUY,” or a guy-guy. Players aspire to “guy” status. As former Red Sox prospect Michael Kopech said after Boston traded him to Chicago for ace starter Chris Sale (who recorded the first and last outs of the 2018 World Series): “All I wanted to do is show them I could be a guy for them.” Players also hope to hit their “ceilings,” a scouting term for an athlete’s alleged best-case outcome.

  Mike Fast, a special assistant to the GM of the Braves and a former Astros research and development director, says that whereas traditionally teams subscribed to labels like these, the franchises at the forefront of the latest, greatest revolution are realizing that “everything” is subject to change. We’ve entered an era in which the right type of practice produces more perfect players, and the earliest adopters of data-driven development are leaving the laggards behind. “I think the idea that analytics is leveling the field is completely backwards,” Fast says. “Analytics is tilting the field far beyond how it has ever been tilted before.” Fast’s colleague Ronit Shah, an Astros scout turned Braves R&D analyst, echoes that sentiment, saying, “The possibilities and the upside are pretty much limitless.”

  Talking in terms of “guys” and “ceilings” suggests that there are identifiable limits. Yet more and more players are figuring out how to go from non-guys to guys or from regular guys to guy-guys, which raises a radical possibility: Maybe there’s no such thing as an absolute ceiling, or the ceiling is high enough that no one knows where it is. And maybe more guy-guys are out there than we ever believed before.

  These new peaks in performance aren’t just the product of better technology. They’re a manifestation of a new philosophy of human potential. Increasingly, teams and players are adopting a growth mindset that rejects long-held beliefs about innate physical talent. One of the only innate qualities may be how hard players are willing to work. Scouts have historically graded players based on five physical tools, but in an era of optimization, a player’s approach to practice is a once-unsung sixth tool that affects the other five.

  “This decade of baseball,” Bannister says, “is all about an inefficiency on the player-development side.” To elaborate, Bannister borrows an analogy from Forrest Gump. “For a long time, baseball players were almost viewed as a box of chocolates,” he says. “They came in endless varieties, and you were just trying to find the best ones. As we started to be able to collect information on players and learn at a rapidly growing pace, we started to realize that the reason the best players are the best players is that they got closer to perfection with the way their bodies moved, as far as executing a certain pitch or taking a certain swing.” For information-friendly teams, Bannister continues, the pursuit of perfection has shifted from “finding bodies that are already doing things well or close to perfect” to asking, “How can we leverage the data and what we’ve learned from the data to get closer to that perfect pitch or perfect swing?” That, Bannister says, is “where the rabbit hole begins.”

  It’s also where the outlying lives of big leaguers begin to apply to our own. Only a small subset of people needs to get great at baseball. But if experienced players in a centuries-old sport can be better than they thought, it suggests something exciting. Maybe we all have hidden talent. And maybe everyone can be better at whatever work they do.

  The index of Michael Lewis’s Moneyball, the 2003 book about the Oakland Athletics that became a bestseller and the source of such severe front-office FOMO that copycat teams across the sport soon molded themselves in Oakland’s analytical image, contains nine subheadings for the listing “players, professional.” There’s “tools of” (the first page of the book), “scouting and recruitment of” (all of chapter two), “sight-based evaluation of” (three entries), and “trading of” (all of chapter nine). There’s “use of statistics in evaluating” (somewhat misleadingly, only one reference). There’s even an entry for what happens when “players, professional” fail to produce: “designating for assignment.”1

  But Moneyball’s index omitted an important potential tenth entry: “development of.” The oversight stemmed from a blind spot of the book—and until recently, of baseball at large. Perhaps in part because of his own history, then A’s general manager Billy Beane—a former first-round pick with raw talent to spare who never learned to translate his tools into on-field success—didn’t devote much time or attention to developing players, at least as Lewis told the tale.

  Much of the drama in Moneyball’s narrative arises from transactions: picking players in the amateur draft, trading for undervalued relievers, and signing the unsung Scott Hatteberg, whose patience at the plate went underappreciated at a time when runs batted in and batting average still reigned as the game’s most prized offensive indicators. As Moneyball portrayed it, Oakland’s ability to compete despite noncompetitive payrolls was about being better at acquiring players. “You can identify value or you can create value,” says former San Diego Padres senior quantitative
analyst Chris Long, one of a wave of stathead hires who flocked to front offices in Moneyball’s immediate aftermath. Ideally, you’d do both, but Oakland’s cutting-edge efforts, initiated by Beane’s nonplayer predecessor Sandy Alderson, were focused on the former. Moneyball’s subtitle promised to reveal “The Art of Winning an Unfair Game.” Apparently, developing players wasn’t part of that art.

  That’s not to say the A’s weren’t promoting players from within; though one would hardly know it from Moneyball, they did have homegrown heroes. Some of them, though, had been top draft picks, always slated for stardom. In Lewis’s book, Beane adopted a deterministic view of player performance, downplaying the idea that players could be capable of changing their ways. Oakland’s draft strategy was akin to clever actuarial work: the A’s determined that picking certain types of players had panned out in the past, so they made more of those picks (college pitchers) and fewer of the riskier kind (high-school pitchers). They also noticed that walks were worth more than the market realized, so they targeted hitters who took them. As a consequence, the homegrown half of Oakland’s early-aughts lineup was less patient than the half acquired through trades, leading Lewis to note that “the guys who aren’t behaving properly at the plate are precisely those who have had the [proper] approach drilled into them by A’s hitting coaches from the moment they became pro players.”

  Because his own prospects had proved unable or unwilling to master traits that the players he imported already possessed, Beane concluded that if plate discipline could be taught, “we’d have to take guys in diapers to do it.” In 1984, another A’s firebrand, Oakland manager Billy Martin, had expressed the same sentiment in even more absolute terms: “You got your mules and you got your racehorses, and you can kick a mule in the ass all you want, and he’s still not gonna be a racehorse.”2

  In fairness to Beane, no one else in the early 2000s was thinking too much about making mules into racehorses. The year Moneyball debuted, Mark Armour and Daniel Levitt, coauthors of Paths to Glory: How Great Baseball Teams Got That Way, wrote: “Other than some analysis of the influence of pitch counts on young pitchers, there has been little research outside of the professional baseball community on such things as methods for developing a young hitter’s power or how to teach a young pitcher to gain better command of his breaking ball.”3 Matters weren’t much more advanced inside that community. Current A’s general manager David Forst, Beane’s longtime top lieutenant, remembers the team dictating that its minor-league pitchers throw a certain percentage of changeups per game and talking about bumping every minor-league hitter up to a 10 percent walk rate, partly by forcing them to take pitches until the opposing pitcher threw a strike. Those methods, Forst says, “seem pretty rudimentary now compared to what we’re capable of doing,” but more advanced development was difficult because “we didn’t have the tools to implement it or measure it.”

  Granted, there wasn’t much need for a forward thinker like Beane to focus on remaking mules when there were so many discount horses around. The A’s could construct a winning team on the cheap by pairing the players their draft approach produced with other clubs’ low-hanging Hattebergs. Hatteberg himself signed with the A’s in 2002 and went on to be their third-best hitter for a single-season salary of $900,000, only three times more than the MLB minimum. “Evaluating was way ahead of developing,” Forst says.

  But Beane’s edge at adding players gradually dwindled, partly because Moneyball’s success inspired imitators and partly because sabermetrics—a movement formative figure Bill James described as “the search for objective knowledge about baseball”—was starting to sweep the sport even before the book became a flashpoint. As Beane said just two months after Moneyball made it to stores: “The old days of getting something for nothing are over. There are too many good [GMs] out there now.”4

  Suddenly, other teams were holding on to their Hattebergs, and in Oakland, economic realities reasserted themselves. The A’s missed the playoffs in 2004 and ’05, failed to finish with a winning record from 2007 to 2011 and, after a brief renaissance, finished in last place from 2015 to 2017. Ironically, the young prospect whom Beane had traded in 2002 to clear room for Hatteberg, Carlos Peña, later blossomed into a far better hitter—and a more prolific walker—than Hatteberg had been. It took time, but the first baseman broke out, even though there was little in his recent performance profile to suggest he would.

  By 2015, when Peña retired and the A’s finished in the cellar for the first time since Beane’s rookie year as a GM, almost every front office was heavily invested in identifying value via stats and analysis, and the most sophisticated clubs were way ahead of where the A’s had been at the turn of the century. That spring, MLB introduced Statcast, a network of cameras and radar that records the speed of every pitch, the velocity and trajectory of every batted ball, and the paths of every player in the field and on the bases in every big-league ballpark. That system supplanted the PITCHf/x and HITf/x systems, which had recorded the speed, movement, and inferred spin of every pitch and the speed and angle of every batted ball for several seasons prior. Below the big leagues, TrackMan (a component of Statcast) soon monitored all thirty teams’ minor leaguers from the highest level to the lowest.

  Although teams differed in how deeply they delved into tracking data, the info was widely available and far more revealing than the best low-tech alternatives from a decade before. Not until 1988—the same year that the influential James published the twelfth and last of his annual Baseball Abstracts—had baseball’s data collectors even noted the outcome of every MLB pitch. Less than thirty years later, a system that once would have seemed like a sci-fi figment was capturing the process that produced every outcome on the field at forty thousand frames per second.

  Bigger data required bigger databases and bigger departments devoted to analyzing their contents. In April 2016, a study Ben coauthored for FiveThirtyEight, a website that specializes in statistical analysis, charted the rapid increase in analysts employed by teams over time. By then, more than five full-time front-office members per franchise, on average, were working in research and technological development (a figure that’s still swelling, topping 7.5 per team by spring 2018). Every team in the majors employed at least one analyst, and every team but the parsimonious Miami Marlins employed more than one. Although the study found that the early adopting data-centric teams had reaped rewards worth as many as several wins (and tens of millions of dollars) per season from mining baseball’s big data before their competitors could, those benefits have shrunk as the front-office brain race has intensified. As baseball analyst Phil Birnbaum once observed, “You gain more by not being stupid than you do by being smart.”5 Teams have long since stopped being stupid about recognizing the good players right in front of their faces.

  Although the term “Moneyball” has come to be associated with specific strategies the A’s deemed most advantageous, it was never actually tied to any one method of team building or in-game management. It was more of a philosophy, one aimed at finding inefficiencies wherever they lay. “When people think of sabermetrics and Moneyball, a lot of it is what they see on the field, the way the game is played,” says Long, who has consulted for multiple teams since departing the Padres. “And most of [the value] is really off the field.” On-the-field changes are easy to see: in recent years, counterproductive tactics that statheads have decried for decades (and that the Moneyball A’s eschewed), like sacrifice bunting and inefficient base stealing, have fallen out of favor. But eradicating bad bunt and steal attempts offers only modest edges. Championships and playoff appearances depend on procuring—or creating—quality players. In the 1920s, teams called the experts who combed the country searching for fresh talent “ivory hunters.” In the 2010s, they call them “quants,” short for quantitative analysts. The goal is the same, but the methods are always evolving.

  In the summer of 2014, hundreds of stat-obsessed seamheads, including quants from fourteen teams, gathered in
Boston for an annual analytics conference known as Saber Seminar. Most of the speakers at Saber Seminar present research about running regressions or writing complex queries to expose some unsuspected sliver of value. But that year’s keynote speaker, then Red Sox GM Ben Cherington—whose team was fresh off a 2013 title—announced that the days of detecting hidden value that players were already providing were quickly coming to an end. “It sure felt like in ’02, ’03, ’04, we could more easily create a talent gap between the best teams and the worst teams, and you could more easily count on a bunch of wins before the season ever started,” he said. “That feels harder to do now.… Finding ways to optimize player performance and get guys into the higher range of possibilities is more and more important.”

  The higher range of possibilities: it may not sound sexy, but that’s where the wins are in a world where teams aren’t being bullheaded about on-base percentage and other once-overlooked contributions. Cherington was speaking a language in which his audience was well versed. A few months before Moneyball made it to the shelves, a young analyst named Nate Silver, who would go on to project political races, unveiled a new framework at the sabermetric-minded website Baseball Prospectus for projecting player performance. His creation, PECOTA—which ostensibly stood for player empirical comparison and optimization test algorithm but was really a “backronym” tribute to former infielder Bill Pecota, who represented a roughly typical player—became the standard for public projection systems. PECOTA presented its estimate of each player’s stat line as a single likeliest projection, accompanied by a range of less likely (but still conceivable) outcomes: a 10th-percentile projection sketched out a season where nearly everything went wrong, whereas a 90th-percentile projection presented one where the player far exceeded what the system expected.