NEW YORK — Even when New York Mets outfielder Curtis Granderson’s batting average had plummeted to a nadir of .122 with just one home run in early May, manager Terry Collins still etched his veteran slugger’s name into the No. 5 spot in the middle of his lineup card the very next day.
That night, Granderson doubled twice to snap a 3-for-49 slump. In the following game, he parked his second home run. In the last two months, he’s hit 12 home runs with a .299 average and 1.037 on-base plus slugging percentage (OPS) that ranks in the National League’s top-three.
The Mets’ patience in waiting for Granderson’s breakout was a result of more than just Collins’ faith that his perennially slow-starting power hitter would again heat up. Though Granderson’s classically outward-facing stats such as average, OPS and home runs were all abysmal after the season’s first month, his underlying advanced metrics showed good process.
Granderson wasn’t chasing bad pitches out of the strike zone nor swinging and missing at a higher rate than usual while his exit velocity — the speed at which the ball leaves the bat — suggested regularly solid contact. His batting average on balls he put in play indicated that he was “tremendously unlucky,” Mets assistant hitting coach Pat Roessler said.
“He was really pretty good,” Roessler added, while speaking at a panel during the annual convention for the Society of American Baseball Research, SABR, which spawned the sabermetrics name for advanced use of data in the sport. “You try to use that to keep their confidence up and keep them in the right frame of mind.”
“We know in here that it’s more than just your average to determine if you’re playing well or not,” Granderson said in the Mets’ clubhouse last week.
This recent example follows more transformational uses of advanced data metrics by the Mets: three-time All-Star Jay Bruce revamped his hitting approach; young pitchers Seth Lugo and Paul Sewald were encouraged to emphasize certain pitches based on their spin rates; the front office settled on Lucas Duda over Ike Davis as its starting first baseman in part because of his consistently higher exit velocity; the coaches position fielders based on intricate algorithms; an analytics staffer accompanies the club on all road trips and sits accessibly in the clubhouse during all games for quick consultations; and those are just a few of the publicly reported examples that have leaked and trickled out in the past few years.
At the SABR panel, a fan inquired whether the Mets implement any neural networks or other machine learning methods in their work. Appearing to be caught off guard by the questions, senior coordinator of baseball systems Joe Lefkowitz replied, “Ummm . . . yes?” He acknowledged that it’s a “technique we use” without divulging anything about its specific purpose.
“It’s obviously a big part of our department’s purpose for existing: We’re trying to use the most advanced statistical techniques we can that hopefully not all the other teams are using,” Lefkowitz said, although with more than 250 analysts across the 30 major league clubs, it can be assumed that the Mets are not alone.
Running the Mets’ baseball operations for the past seven years, which included a 2015 trip to the World Series, has been general manager Sandy Alderson, who in a previous post mentored Oakland A’s executive Billy Beane, who of course has become synonymous with baseball analytics thanks to Moneyball. Two of Alderson’s top lieutenants were J.P. Ricciardi and, until two years ago, Paul DePodesta, each of whom previously advised Beane. It’s little surprise, therefore, what interest the Mets have in incorporating objective data into their decision-making.
“We’re simply trying to enhance the probabilities of success,” Alderson said. “We can’t guarantee success. We’re not going to be unnecessarily or excessively impacted by the results. We have to focus on the process and enhance our chance of success.”
(Not every decision is analytically-minded, however. Alderson said recently that the decision to sign Tim Tebow — the Heisman Trophy winner and former NFL quarterback whose celebrity transcends football — to a minor league deal was, at least in part, an entertainment decision. He revealed that the scout listed in the media guide as having signed Tebow was actually a merchandise executive: James Benesh, the manager of venue services.)
Get The Latest MLB Tech News In Your Inbox!
In the case of Bruce, he has taken the advisement of Roessler and hitting coach Kevin Long to focus on what he does best, which, the numbers show, is hitting the ball in the air to the pull side (rightfield, for a lefty like him). The results: his 20 home runs are the most he’s hit in the first 80 games of a season, and his OPS is only four points shy of his career best from 2010. Under the hitting coaches’ tutelage, Bruce can rattle off his hard-hit rate, his flyball rate and the batting average on ground balls into a defensive shift.
“There are so many numbers and percentages and rates and ratios that you can chew on, but how do you melt that down to make it applicable?” Bruce said. “A lot of people, I think, tend to maybe run from information a little bit. Well, I like to have information. I’m not saying I’m going to use it all. I wanted to figure out how to apply that information I got in the most simple form. For me, it was get balls you can hit in the air and you can drive. And it’s really helped.”
Trackman’s military-grade Doppler radars record 27 data points on every play in the majors and have been available in all 30 ballparks powering MLB’s Statcast tracking system for the last three seasons. Some clubs, such as the Mets, have been using Trackman on their own for a half-dozen years or more. T.J. Barra, the club’s director of baseball research and development, said that duration has been long enough to give the front office an understanding of the reliability, predictability and usability of those numbers.
For instance, they’ve advised pitchers to promote a certain pitch in his arsenal based on its spin rate and model how launch angle — the vertical angle at which the baseball leaves the bat — can predict production, helping the Mets turn Statcast data into “something that’s useful, not just cool,” Barra said.
While the rate of defensive overshifts (putting a third infielder on one half of the field) have risen significantly in the past few years and sometimes have become a proxy to measure an organization’s reliance in sabermetrics. The Mets rank in the bottom five in the majors for deployment of shifts this season but not for lack of data — quite the opposite. The analysts help the coaches position fielders via a wide array of variables including not only the opposing hitter’s tendencies but also the Mets’ pitching game plan for the evening. It just so happens that the Mets haven’t faced too many extreme pull hitters expected to continue that pattern when facing New York’s pitching staff.
In fact, Mets first base coach Tom Goodwin joked that the analysts are always trying to push the envelope with new defensive formations.
“It might be coming where we have four outfielders,” Goodwin said. “T.J. has been dying to do something like that. These guys, when they get together and they get in that room, you never know what’s going to come out.”
That’s why, for pregame meetings, he joked, “We kick the techies out.” Instead, Barra meets with each coach for 15 minutes before each game to run through a plan, which is evidence of some symbiotic harmony between the coaches and the analysts. “As long as you have a healthy relationship when you’re doing it, you can put your guys in the best possible situation where they’re able to make a play,” Goodwin said. “That’s really what both sides want, when you talk about the gut and you talk about the stats.”
This dichotomy of old-school coaches and scouts relying on their gut to make decisions while the analysts and younger executives have pushed for more data-minded determinations has been perpetuated and exaggerated ever since the release of Moneyball. The most successful clubs have married the two into a joint process. Barra gave an example of how the analytics staff helps prepare scouting reports: the Mets had never faced that night’s starting pitcher, the Phillies’ Ben Lively. Through Trackman, however, they had full data on his entire repertoire of pitches, so the analytics crew was able to compare him to similar pitchers New York had faced.
The Mets understandably have been less forthcoming about a number of their in-house initiatives, including providing assistance to the strength and conditioning staffs, researching the best management of pitch counts and inning limits to protect pitchers and more.
Not every area of the game can be quantified or projected, of course, but as Alderson said, “The key thing is making sure you don’t underestimate what you can control, and you do your damnedest to make sure that what is in your control, you do as well as possible.”