Big Data Baseball Read online

  Even beyond identifying the nearly invisible potential of Russell Martin as a must-have free agent, Fitzgerald considers his most important work to be gathering and analyzing day-to-day material for opponent-scouting meetings, which are a combination of reports and voices. The advance video scouting is done by Toregas, a former backup and minor league catcher for the Pirates, who by August and September was breaking down the last 45 games of each opponent in advance of each series. Reports also came from live scouts in the field, following upcoming opponents, who were responsible for providing up-to-date anecdotal snapshots of players and teams, findings that do not show up in a box score. And beginning in 2012, and playing an even bigger role in 2013, was the analytical data supplied by Fox and Fitzgerald. Fitzgerald says much of the work done is to try to confirm findings or observations by advance scouts through objective analysis. Ideally, they want evidence everyone can agree on and work from.

  “It’s math-y, but there’s still the whole arts-and-science debate,” Fitzgerald said of implementing big data. “I’d argue there is an art to that sort of stuff. I think that’s the biggest thing.”

  But where do the subjective and the objective meet? Fitzgerald cites an example. Say a batter excels against fastballs on the outer half of the plate from left-handed pitchers. But then what about the two lefties in the Pirates bull pen? Tony Watson and Justin Wilson both have rare, for left-handed pitchers, upper-90s fastball velocity. Maybe that batter has done damage against four-seam fastballs on the outer half, but how many 97 and 98 mph fastballs has he actually seen and hit? Here is where Fitzgerald and Fox dig deeper, seeking more subjective and objective information.

  “So it’s an art in the sense of what we get from the raw information doesn’t always tell the whole story. How were the pitches set up? Were there runners in scoring position that were tipping pitches [to the batter]? At the end of the day you can make the argument that [the art] is just refining the data, but in a way there are still situations that come up where there is gray area and you have to massage through it,” Fitzgerald said. “I like to think that is where we do damage, where we can get value out of it.”

  Fitzgerald says that five years from now every team will have incorporated data into their day-to-day operations. So the question then becomes how well will individual clubs do on the tough, fifty-fifty calls? What is the decision-making process? Will the field staff consistently stick with decisions made in advance meetings? That’s where the new advantages will reside. Through a growing respect and appreciation for the different skills sets of old-school and new-school camps, made possible by time and familiarity, the Pirates created a massive communication advantage. It not only allowed data-based findings to reach the field in 2013, but the communication allowed the data to be refined by coaches and players asking the right questions, questions sparked by observation.

  Part of the sport’s struggle in communicating big data ideas to on-field staff is undoubtedly tied to presentation flaws, and in the 2012–13 off-season, Fox and Fitzgerald focused on how they could overcome this and better present their data to best get it adopted on the field. This went beyond selling the team on shifting. How could all data findings most easily be accepted by players and coaches in advance meetings?

  “The biggest thing I’ve found, and I think Dan and the other guys would agree, is these guys understand visuals like that,” said Fitzgerald as he snaps his fingers, “versus if we give them some [statistical] data laid out on paper.”

  Recall, one change for presentation the Pirates made in 2013 was to purchase a video and information tool from TruMedia that allowed for the easy creation of data-visualization heat maps from the Pirates’ statistical databases linked to video. These are critical tools in the pregame scouting meetings.

  “Say, so-and-so has a thirty percent miss rate against changeups below the zone. We can pull up all the video on that,” Fitzgerald said. “Or, ‘Here’s the cluster he’s done damage against a certain pitch, and then here’s the video that ties to that.’ You can say, ‘I see a hole here; let me go look at this hole on video.’ I think it kind of speeds up the process of finding weakness.”

  Ray Searage brings an important piece of luggage with him on every road trip, a large plastic crate, roughly nine cubic feet in size. When he opens the crate, all that is visible at first are towels, as if he does not trust hotels to launder their bathroom essentials. But after the Pirates pitching coach removes several layers of fabric, his most important ally in game planning emerges: a laptop.

  “Two have already been smashed,” said Searage of his previous laptops when explaining his seemingly overly cautious packing.

  There in the middle of any road clubhouse, surrounded by players’ lockers and near the center-of-the-room sofas and televisions of a common clubhouse layout, Searage can often be seen before games with his laptop opened, examining various heat charts of opposing hitters and Pirates pitchers. Searage looks at the screen and then writes down his thoughts on a legal pad, formulating a recommended game plan. Fox and Fitzgerald supply him with their analytical findings, having combed through the video advance report and having tried to identify areas with likely question marks. For instance, a switch-hitter might have been strong from the left side lately but historically is a much better right-handed hitter. What should the recommended plan of attack be? That is where art meets science and objective meets the subjective.

  Searage prefers to receive the advance statistical and video reports on opponents two to three days before a particular pitcher’s start. Then he begins poring over the data, consulting with Russell Martin and the pitchers to outline a plan. During the last series of the regular season in Cincinnati, never was there a more important strategy session during the last two decades in Pirates history.

  The Pirates arrived in Cincinnati on the final weekend of September already knowing they would play the Reds in a one-game play-off to determine which team advanced to the National League Division Series (NLDS). The Pirates and the Reds had clinched the NL’s two wild-card berths, and the Cardinals had secured the division title, but something critical remained to play for: home-field advantage. The Pirates had a 50-31 record at PNC Park, a .617 winning percentage, compared to a .525 mark on the road. Home field was even more valuable to the Reds, who owned a .636 winning percentage at the Great American Ball Park and were just a .500 team on the road. The winner of the weekend’s series would gain roughly a 10 percent edge in advancing to the NLDS.

  What most influences home-field advantage is not what you might think. It is not primarily tied to park dimensions, travel fatigue, or familiarity. It’s umpire bias. University of Chicago behavioral economist Tobias Moskowitz and Sports Illustrated writer L. Jon Wertheim concluded in their book Scorecasting that home-field advantage is predominately tied to the home-field team’s enjoying the benefit of more borderline strike-ball calls, with umpires either consciously or subconsciously influenced by the environment. Moskowitz and Wertheim based this on examination of millions of pitches tracked by PITCHf/x and QuesTec—computerized systems that track pitch location.

  “In baseball it turns out that the most significant difference between home and away teams is that the home teams strike out less and walk more—a lot more—per plate appearance than road teams,” Moskowitz wrote.

  Moskowitz also found that the larger and louder the crowd, the more umpire bias—whether consciously or subconsciously—could be created. A one-game play-off, such as the wild-card game, could make for a raucous environment.

  But home-field advantage was also important because Cincinnati’s Great American Ball Park was one of the most hitter-friendly venues in the National League. The Reds were built for run production with sluggers such as Joey Votto, Shin-Soo Choo, and Jay Bruce. While the Pirates were built for run prevention at pitcher-friendly PNC Park.

  While the Pirates arrived in Cincinnati knowing whom they would play in the wild card play-in game, they also entered knowing who would pitch for
them: left-hander Francisco Liriano.

  The analytics, the video scouting, and the advance reports all agreed: the top three Reds hitters—Votto, Choo, and Bruce—were all left-handed, and all struggled with the low-and-away slider, Liriano’s best pitch. Searage could see these weaknesses clearly on the heat maps generated on his laptop. A package of printed statistical reports and heat maps were placed in each player’s locker prior to a series. In those reports they could learn of the .321 on-base plus slugging percentage (OPS) left-handed batters produced against Liriano in 2013. The mark was the all-time single-season best by a left-handed pitcher who had faced at least 100 left-handed hitters, according to Baseball-Reference.com’s “play index” tool. Liriano was even better than Randy Johnson, who might have been the most terrifying left-handed pitcher for a lefty hitter to face in the game’s history, and who in 1999 limited lefties to a .331 OPS. The six-foot-ten Johnson threw in the mid-90s with a sidearm motion and also had a wipeout slider. Liriano’s improved ability to get ahead with a low-to-mid-90s two-seam fastball—in part tied to Martin’s pitch framing—allowed for his slider and changeup to become even more effective. Hitters were facing less favorable counts, more susceptible to either missing or pounding Liriano’s off-speed pitches into the ground.

  Securing home-field advantage for the winner-take-all game was not just critical for the Pirates as a whole, it was critical for Liriano. In 2013 his ERA was 1.47 in pitcher-friendly PNC Park versus 4.32 on the road. After a series-opening win on Friday, September 27, against the Reds, the Pirates smashed six home runs against them on Saturday to secure home-field advantage. Come Tuesday, Liriano would be on his home mound, and Pittsburgh would be hosting a postseason baseball game for the first time in two decades. Finally, meaningful baseball would be played in October in Pittsburgh.

  * * *

  Russell Martin’s father never wanted a nine-to-five job. He refused to live a traditional lifestyle. Jazz was his passion. He worked in construction and was good with his hands. He built decks and took on other odd jobs. He did that work so he could frequent the Montreal subway as a street performer during morning and afternoon rush-hour crowds. His pleasure came from having a commuter pause, stop, and listen. He gave Russell the middle name Coltrane in homage to the great jazz performer John Coltrane. Russell senior told The New York Times the middle name was not simply meant as a tribute to Coltrane’s music but also to his free and independent spirit, which he hoped his son would emulate.

  With the Pirates, Martin presided over the stereo in the clubhouse, hooking up his iPhone and scrolling through the saved playlists. Attached to the exterior of Martin’s wood-grained locker is a list of music designations for each day of the week. There’s a day for rap and hip-hop, a day for alternative rock, and a day for Latin music. To create a more subdued Sunday-morning atmosphere following Saturday-night games, there’s reggae as players sporadically file into the clubhouse, often toting Starbucks coffee-to-go cups. Martin controls the playlist much the way he sets the tone in the clubhouse, players say. He comes to work every day setting a standard. He works as hard as any other player in recent memory with the Pirates, running stadium steps at 2:30 p.m. in September five hours before a night game, part of what he calls a pregame “activation” period. The songs Martin chooses to play all have a repetitive beat, which is a defining criterion for rock and roll and much of the modern pop/rock and hip-hop genres. Jazz lacks a repetitive beat. Jazz, the love of Martin’s father, is improvisational.

  “Jazz is still a huge part of his life, but I never really understood that music. I like a beat, a pattern,” Martin said. “In jazz they are just like one guy is going one way, another guy is going off that, the other guy is going off that. It’s like you are all playing together but there is no pattern. It’s strange.”

  However, Martin tries to avoid a repetitive beat as a catcher, endeavoring to be without patterns in one of his most important tasks. Despite all the data at the hands of coaches in the dugout, professional catchers still have an important in-game task: they call the pitches. In the NFL and the NBA, coaches call the plays. Even in college baseball catchers turn to the dugout for pitch signals from coaches. But in professional baseball, catchers are responsible for calling pitches.

  Martin takes the responsibility seriously. Being an analytical player, he plays a significant role in designing game plans, examining the strengths of opposing hitters versus that day’s starting pitcher, and reviews scouting reports created by the analytics team. In particular Martin likes to study first-pitch swings, percentage of swings, percentage of out-of-zone swings, and two-strike approaches. He wants to get a sense of the hitter’s tendencies, but also of his psychological disposition: Is he aggressive or passive? Martin studies heat charts trying to locate the areas of a strike zone a hitter is effective in. For instance, in a start at San Diego, the Pirates television broadcast displayed a heat zone, a data visualization, of how Padres left-handed hitter Seth Smith performed against pitches in certain parts of the strike zone. Smith hit well against pitches in the middle to outer half of the plate. So Martin had Pirates starting pitcher Gerrit Cole pitch in, under, and over the hot areas to record a three-pitch strikeout.

  Martin understands the importance of big data; for him it provides an edge at the margins. But the game is still largely instinctual for him, and he relies heavily on observing and adjusting off what he sees. If his middle name and his father’s love of jazz have impacted him in one area, it is on the baseball field in pitch sequencing. Martin strives to avoid patterns.

  “My style of catching and calling a game is a little like jazz,” he said. “There’s no beat to it, there’s no sequences. There’s just feel. I almost feel that’s how I can call a game. I’ll call eight changeups in a row or eight sliders in a row if I feel like that’s the right thing to do.”

  In observing Martin and seeing the way he interacts with teammates, Fox was struck by something, another hidden value. While it was perhaps impossible to value intangible elements such as leadership and baseball makeup, Fox was blown away by Martin’s work behind the scenes and thought that these skills must have value. Fox was impressed by Martin’s ability to articulate his methods from pitch framing to pitch sequencing to hitting approach. It wasn’t just Martin’s pitch framing that Pirates pitchers raved about; they also praised him for his ability to game-plan. Jeff Locke lauds Martin’s ability to keep hitters off-balance. Although quantifying such skill through data is not known to have been accomplished, Locke believes it does have value. For example, earlier in the season against one of the National League’s best hitters, Joey Votto, Martin demonstrated his master pitch-sequencing psychology.

  “We went with back-to-back fastballs in the first at bat to Votto on [June 1]. We went right after him. It’s oh and two and he didn’t take a swing, and I’m sure that was kind of weird for him,” Martin said. “He knows how he is going to be pitched. When you throw something that doesn’t make sense, he’s like, ‘Whoa, he just threw me a changeup?’ He has to reprogram the possible sequences that are going to happen in his mind. It’s knowing who you are playing against.… If he’s beat in a certain way, does he come up in his next at bat looking for that type of pitch? There is no piece of paper that tells you that.”

  To Martin, pitch-sequencing is an art form that could benefit from some data analysis. While he’s able to employ some of the data at his disposal to create a rough plan, once the game starts, he has to adjust and evolve based upon his reads from the batters. Sometimes it’s observing how comfortable a batter is in taking a pitch. Was he jumpy or was he “soft” on the landing, tracking the ball with ease and letting it drop out of the zone without so much as a triggering of his hands? While a pitcher’s slider might be his best pitch according to data from PITCHf/x, if Martin doesn’t like its movement in a bull-pen session or feels it lacks its typical life during the game, he’ll stay away from it.

  “It’s always good to review [data],” Martin said. “
But instinctively you have to do what you feel is right. You have all that information to fall back on, but you don’t want it to cloud your judgment during the game.”

  Most important, Martin knows his pitch-sequencing ability won’t matter if he can’t get pitchers to trust him and his process. He knows his pitch calls are merely “suggestions” and the pitchers have the ultimate veto power with a simple negative, horizontal shake of their head as they look in for signs from the mound.

  Entering spring training in 2013, Martin knew little about the pitchers Locke and Liriano. Locke was young with little of a major league track record, and Martin had never played with Liriano. Prior to spring training, Martin asked Pirates video coordinator Kevin Roach for video on the two pitchers so he could study them. In spring training Martin felt he connected with Locke as they rotated from hot to cold tubs during a treatment session. They spoke for forty minutes and it was not all about baseball. Martin was trying to learn Locke’s temperament and level of aggressiveness. Martin also shagged fly balls with Liriano and others, picking their brains, searching for nuggets of their pitching philosophies. More than anything else he wanted his pitchers to feel confident on the mound. He doesn’t want them overthinking. He wants them to trust the call.

  “That’s the thing about baseball, it’s not a perfect science,” Martin said. “There’s not going to be one moment where that was the perfect pitch.… Typically, if a pitcher has conviction and he attacks, we are going to get a good result.”

  Pirates reliever Jared Hughes told a story to a reporter of how he learned to trust Martin that spring. In an early exhibition game, Martin called for a slider, and Hughes, a natural sinker-ball pitcher, in need of a ground ball to get out of a jam, shook off Martin’s call. Martin then called time-out and walked to the center of the McKechnie Field diamond to chat with Hughes. He wanted to know why Hughes had rejected Martin’s pitch selection. Hughes explained he needed a ground ball, he needed his sinker. Martin told Hughes he believed his slider could be a ground-ball pitch, too. Hughes thought about this. It was a spring training game after all, the stats didn’t matter much, and he was sure to make the team. He agreed to throw the slider, and the pitch did generate a ground-ball out. Through small moments like that, Martin built trust.