Big Data Baseball Page 18
The Pirates are known today as a data-heavy, forward-thinking organization that deploys analytics heavily in decision-making. While Huntington did not hire Gayo, he kept him in a senior role because Huntington believes traditional scouting remains paramount in amateur-player acquisition, particularly in Latin America. The Pirates have flooded the Caribbean with scouts in attempt to collect more intel on prospects than other organizations. In the Dominican Republic, play is far less organized than in the United States, so no meaningful statistics are produced like those at the Division I college level. No algorithms or databases can help you find talent on ramshackle diamonds where grazing cattle sometimes share the outfield. In the Dominican Republic, a scout must find lanky, malnourished sixteen-year-olds and have the ability to predict what they could become. Perhaps part of this is innate, but more than anything else, it is the product of intensive study over years.
“It’s raw scouting in its purest form,” Moises Rodriguez, the Cardinals’ director of international operations, told Baseball America. “There are no stats. If you have scouts that know how to evaluate, and if you create a system to evaluate guys that are shown to you by your scouts, you’re going to succeed a lot more.”
This willingness to go where others would not and to see players as others had not had led Gayo to Starling Marte.
* * *
Santo Domingo, the Dominican capital, has two airports. The main airport, Las Américas International, to the east of the city, is where the vast majority of commercial traffic arrives and departs. To the north of the city is La Isabela, an airport with but one airstrip, which caters to small planes and charter flights. Adjacent to the airport exit, alongside an access road, is a ramshackle baseball field with a red-clay infield. Gayo travels to fields like this one on the outskirts of Santo Domingo, arriving with a lawn chair and two new, fresh baseballs he calls “pearls.” He gives each team a ball, and he’ll sit and watch under a searing sun.
Many scouts and evaluators in the Dominican Republic prefer to stay around their academy and have buscóns bring their talent and teams to the complex for workouts and games. Buscóns are quasi agents, sometimes coaches and sometimes even father figures, who represent young Dominican players. Gayo has always had a problem with people bringing talent to him. He’s a scout. He reasons that he is supposed to go in search of talent, not have it brought to him. Moreover, when the talent is brought to the complex, the buscón controls the environment and showcases the talent. Gayo wants control. More than anything else he wants to see games—not batting practice and speed-pitch sessions.
Gayo plays a strategic sales game that often goes something like this: A buscón will approach Gayo and attempt to sell him on his shortstop, not knowing Gayo has his eyes on the right fielder. Gayo tells the buscón that the Yankees like the shortstop and will surely outspend the Pirates for him. He’s too expensive, but what about the right fielder? You want $20,000? How about $25,000? That’s how Gayo signed Willy Taveras. Gayo found Marte, like Taveras, on a field on the outskirts of the capital. In the spring of 2006, few saw Marte as a future big leaguer.
The opportunity to find an undervalued Marte was largely tied to his being showcased as a shortstop. When Gayo first saw him, he saw him the way all the other scouts did: a player lacking the instincts and hands for the infield. He didn’t like Marte enough to sign him upon his first look. But Gayo was intrigued. He wondered how Marte would fare in the outfield.
“I remember back to [Ernest] Hemingway, he used to talk about bullfighters and the gracefulness of bullfighters. That’s something I’ve always liked. Clemente had that. He played with fire but there was grace to his game. Starling had that. There was a grace to his game,” Gayo said.
Gayo asked Marte to go to the outfield. Though a malnourished, 160-pound eighteen-year-old who was dealing with parasites like many Dominican teenage prospects, Marte began unleashing laserlike throws to the infield. Gayo’s eyes widened to saucers. He then asked Marte to run. After watching Marte run several sixty-yard dashes on an unkempt playing surface, wedged between cattle-grazing grounds and tin- and sheet-metal-roofed homes, Gayo’s jaw dropped. Marte clocked a 6.48-second sixty-yard dash. If you can run a 6.6, you can run very fast. If you can run a 6.5, you’re flying.
“Not only 6.48,” Gayo said, “but you could put a crystal glass on his head and it won’t fall down. This guy is running and you’re like, ‘It’s beautiful.’ This is like the most beautiful thing you’ve ever seen.”
But could he hit?
In a batting practice, Marte sprayed line drives all around the field. He demonstrated balance in his swing. Gayo still had questions; it was hard to tell if a young player would have an understanding of the strike zone or hit for power, but he saw that Marte could get his bat to the ball. In his swing, Gayo saw Marte had balance in his shoulders, waist, and knees, and power in his wrists. With a regular healthy diet and time in the weight room he could gain strength and really be something, Gayo thought.
Something else that Gayo liked in Marte was that he could compete. In game situations, when Marte struck out or made a poor play or decision, it did not carry over to his next opportunity, and he did not lose confidence. Gayo was sold. He offered Marte $85,000; he officially signed with the Pirates on January 4, 2007. Every other team had missed on Marte because of what he couldn’t do: he couldn’t play an adequate shortstop. Gayo focused on Marte’s strength: he could cover ground like a deer.
Marte worked his way quickly through the Pirates system. With face-of-the-franchise Andrew McCutchen solidly entrenched in center field, Marte, a natural center fielder, was shifted to left field when he was first called up to the majors, near the close of the 2012 season. While a move off center field is usually viewed as a negative for a prospect, playing left field in PNC Park was more difficult and therefore more valuable. Marte proved that he was up for the job in his first full season in the majors in 2013.
On August 18 in the top of the twelfth inning in a tie game with the Pirates against the Diamondbacks, Arizona second baseman Aaron Hill lifted a fly ball into shallow left field that to onlookers appeared would land and easily score the runner on second base to give Arizona a 1-run lead. Instead, Starling Marte came out of nowhere, at least from out of the edge of a television camera’s periphery, to make a sliding blur of a catch to end the inning.
Had Marte been raised in the United States, he might have been a wide receiver or defensive back somewhere in the NFL. He’s a sleek, six-foot, 190-pound athlete with muscle definition, the type of body rarely found in a major league clubhouse. On this August evening, Marte exploded out of his stance and took a perfect route to an interception point. Marte had been covering ground like that all season for the Pirates. His defensive range was a key reason why the Pirates remained a game ahead of the Cardinals in the middle of August.
According to the statistic defensive runs saved, Starling Marte ranked first among left fielders in 2013, saving 20 more runs than the average left fielder. This statistic used video scouting and analytics to attempt to measure the range of an outfielder. Teammate Andrew McCutchen ranked sixth among center fielders with 7 runs saved. According to ultimate zone rating, a statistic that uses a similar methodology to defensive runs saved, the Pirates had two of the top twenty outfielders in the game in 2013. Only Oakland, Pittsburgh, Arizona, and Boston had two outfielders in the top twenty. Boston won the World Series, and Oakland led the American League in wins. It spoke to the importance of defensive play, the importance of having an athletic outfield.
But August 18 was the last game Marte played in for the Pirates for nearly a month. Eight days earlier in Colorado, a fastball from Rockies relief pitcher Josh Outman struck Marte on his left hand. Marte fell to his knees and grabbed the hand. The Pirates training staff came out to the batting area to examine Marte, who grimaced in pain. He continued to play through the injury for a week, but while it had not affected his fielding, he struggled to grip the bat and swing with full forc
e and became hesitant to slide headfirst on the bases.
Marte complained of pain in his hand again following the August 18 game, and on August 24 the Pirates placed him on the disabled list. In his absence the Pirates played around .500 baseball. They lost 14 of 29 games and went from 1 game ahead of the St. Louis Cardinals in the National League Central to 2 games behind them, and just ½ a game ahead of the Cincinnati Reds, on September 18. The Pirates’ lead in the wild-card race had shrunk to 5 games over the Washington Nationals and 7 games over the Arizona Diamondbacks. While their grasp on at least one of the two wild-card spots seemed firm, it was no longer automatic. The fans on their public forums of talk radio, Twitter, and Internet message boards collectively groaned, “Here we go again.” “Ownership is cheap.” “Epic Collapse III!”
Fueled largely by his defensive value, Marte finished the 2013 season ranked as the twenty-eighth most valuable position player in baseball with 4.6 wins above replacement, despite missing 30 games to a hand injury. Still, if he couldn’t grip a bat and swing it effectively, he couldn’t be in the starting lineup—a considerable loss. But it was tough to say exactly how effective Marte was since there was no perfect way to evaluate defensive performance. But that was all about to change.
Accurately measuring defense is one of the last great frontiers of on-the-field analytics, along with injury prevention and softer sciences such as a player’s desire to compete and excel at his craft or clubhouse chemistry.
Throughout the twentieth century the public and teams largely made subjective judgments about defenders. Few analytical tools existed to measure defensive play. One of the few was the highly flawed statistic of fielding percentage, the number of successful assists and putouts divided by the total chances to either assist on or produce an out. But fielding percentage does not take into account the range of a player. For example, an error for one player might be a ball another player cannot reach. Moreover, an error is a subjective judgment of an official scorer.
John Dewan wanted to change baseball’s thinking not just about overall team defense—by using defensive alignment and shifting—but also about individual player defensive value, by creating metrics such as defensive runs saved. Like Dewan’s system, the Pirates’ in-house defensive-valuation systems debit or credit defenders on their ability to turn batted balls in certain zones into outs. But Fox believes, like Dewan, that no system created to date is by itself enough to give a completely accurate portrait of an individual defensive player. Fox notes that the Pirates try to account for the ballpark and the type of pitcher the defender is paired with when evaluating a fielder’s performance, but still, even the best of the current metrics are fraught with biases and unknowns. Since player movement is not being measured yet, there cannot be an accurate assessment of ability. An accurate portrait would require the new player-tracking system, Statcast.
* * *
On March 1, 2014, at the MIT Sloan Analytics Conference, Joe Inzerillo of Major League Baseball Advanced Media (MLBAM) took the stage at Hynes Convention Center in Boston for a much-anticipated presentation, akin to baseball’s version of the late Steve Jobs unveiling an Apple product. Behind him on the stage was a large drop-down screen for the projection camera. At that time much of MLBAM’s focus was on improving the digital reach and experience for its customers, which made MLBAM one of the most important and profitable baseball subsidiaries. Each major league team has an equal stake in MLBAM, and Bloomberg.com estimated each stake to be worth $110 million in 2013. The organization, baseball’s digital arm, has turned away offers of more than a billion dollars from equity firms seeking to buy a stake in the company, according to media reports. But with the success of its partnership with Sportvision’s PITCHf/x system, MLBAM became curious about creating its own data, and at the 2014 Sloan conference, Inzerillo announced its most exciting venture to date.
Before an eager audience of analytically inclined people in the baseball community, Inzerillo made the astonishing claim that he and his team had solved baseball’s greatest mystery. He announced that every movement, every step, every throw on the field, would soon be measured and quantified.
“The Holy Grail has always been finding this defensive side, the runners, on an empirical basis that we could see,” Inzerillo told the audience. “Now we are actually going to be able to see this phenomenon directly produced and then analyze the data and see what it means.… Baseball is a game of inches; now we are going to be able to tell you how many inches.”
Then from his handheld tablet, Inzerillo began a remarkable video presentation of a single play from a 2013 Braves-Mets game at Citi Field. In the bottom of the ninth, the Braves had a tenuous grasp on a 2–1 lead against the Mets. The Braves’ excellent relief pitcher Craig Kimbrel had struck out two but also hit a batter and walked another. The tying and go-ahead runners were on base when Justin Turner drove the ball into the left-center gap. Off the bat, the drive looked as if it would at least tie and possibly win the game. But Braves center fielder Jason Heyward, who typically plays right field, made a diving catch to end the game and secure a win for the Braves. It was a remarkable play, but the graphics that accompanied Heyward and Turner on the screen were even more incredible.
The graphics showed Heyward’s real-time data. He had reached a top speed of 18.5 miles per hour, accelerated at a rate of 15.1 feet per second, begun 80.9 feet from the interception point of where the ball would fall for a hit, and covered that ground in 83.2 feet, nearly a perfect line to intercept the ball, a route efficiency of 97 percent. Heyward also displayed first-step quickness, his first reaction occurring just 0.2 seconds after the ball was hit. And it wasn’t only Heyward that was tracked—everything was tracked. The ball left Turner’s bat at 88.3 mph and had a 24.1 degree launch angle. It traveled 314 feet and had a hang time of four seconds. This was PITCHf/x on steroids.
For the first time in baseball history MLBAM demonstrated that just about every aspect of a remarkable play such as this could be measured. This was the next giant step for big data in baseball. It promised to push the total data points produced in the sport each season from the millions to the billions.
“We are just scratching the surface,” Inzerillo said.
How does it work? Statcast combined the information from two different systems to pull its data together. Using the radar-based TrackMan to track the ball and its movement, Statcast also tracks the players on the field using a twin set of ChyronHego binocular cameras, which have 3-D capabilities due to their stereoscopic arrangement. The cameras record every player’s movements on the field and are synced with the radar readings from TrackMan Doppler radar. The players’ and the ball’s movements are translated and turned into meaningful data by the system’s software. The Statcast system was already set up and gathering real-time data at the Mets-Braves game at Citi Field in 2013 and was planned to launch in three more stadiums in 2014, with the goal being to have the system in place in every park by 2015. What PITCHf/x did for recording pitch movement, location, and speed, MLBAM’s Statcast would do for every movement on the field.
But challenges are still ahead. While the data can be tracked and shared in real time, the accompanying real-time graphics have not been adapted. So much data—several terabytes—is produced per game that it limits who can process and have access to it. Unlike with PITCHf/x, teams do not want the Statcast data to become public, so that they can create their own proprietary metrics. Assuming the system does go into place, analysts will then have to create meaningful metrics from billions of new data points flowing into the game.
How is this going to change the game? What was once left to the subjective eye on the field will now be quantified. Just as John Dewan got the baseball industry to change the way it thought about collective defensive performance, Statcast has the potential to change the way we value and think about the performance of individual defenders, and also to help us better understand defense as a whole.
St. Louis Cardinals GM John Mozeliak feels Sta
tcast will have a significant impact. “It eliminates the intuitive or subjective opinion on who’s really good at defense,” Mozeliak said. “This will allow you to really have a clear definition on someone that can cover ground and how they do it. Now it is still very subjective analysis. As much as we try to push the envelope, this type of tracking system will really change how people think about defense.”
MLBAM’s system could place an even greater premium on defensive value as qualities such as range, accuracy, route efficiency, and arm strength will for the first time be precisely quantified. The Pirates were already investing in the bet that individual defensive value was more important than many other teams thought in 2013.
But in the middle of August, with Marte’s injury, that competitive advantage vanished. Now the Pirates would be challenged to give the city something it hadn’t had in twenty-one years—a home play-off game—without him. Without Marte, the Pirates needed someone else to step up.
11
ARMS RACE
In early September of 2013, at Yinzers, a well-known sporting-goods store wedged between warehouses and the outdoor market near downtown Pittsburgh, a large placard was placed outside the storefront marking down the final six victories needed for the Pirates to reach win No. 82—thereby guaranteeing them their first winning season in twenty-one years. With each win, another number—77, 78, 79, and so on—was crossed off. Thousands walked past the sign every week as the strip’s boulevard hosts the city’s popular Saturday-morning market. The sign became the city’s unofficial countdown watch. However, on September 9, there was a problem: Yinzers hadn’t been able to mark off a win for five days because the Pirates hadn’t won. The countdown was stuck on win No. 81. Though reaching 82 wins at this point in the season seemed mathematically inevitable, until it happened, Pittsburghers had a degree of doubt.