How the Red Sox revived their analytics department

FORT MYERS, Fla. — Reputations tend to last beyond their accuracy. In the public consciousness, an ace may remain one after he's lost his stuff, a manager may be old-school after he's adapted and a baseball franchise may be at the forefront of the game's analytical community even after it's actually fallen behind.

When Dave Dombrowski was hired as the president of baseball operations in August 2015, one of the first things he learned was that the Red Sox, a paragon of baseball analytics to so many, did not live up to their reputation.

 "I was actually surprised when I came in, thinking that the Red Sox were No. 1 [in analytics staffing]," Dombrowski said. "They were not No. 1."

When the Red Sox won the 2013 World Series, they had three full-time employees in the analytics department, as well as Bill James as a regular consultant. Long the big-market darlings of the analytical community, Boston had been passed by several clubs who had devoted greater numbers to that part of their front office.

"Certainly all other teams are doing more in this area than they were 15 years ago, and a lot of teams are doing much more," assistant general manager Brian O'Halloran said. "In terms of the resources deployed toward that area, we had just fallen behind."

Overhauling the analytics department had long been an object of organizational conversation. Carmine, the analytical database ballyhooed and derided in equal measure, had initially been designed around the year 2000, at a time when mobile phones were just phones and HiFi meant more than WiFi.

The Red Sox couldn't rework Carmine, though, with its current resources. Those IT staffers who pitched in here and there were being dragged to more business meetings those days, and adding personnel to analytics fell behind adding to the scouting and medical departments on the organization's list of priorities.

"Time passes, things change," said Zack Scott, Boston's Vice President of Baseball Research and Development. Scott joined the Red Sox in 2003 along with Tom Tippett, who designed the prototype for Carmine with Diamond Mind, Inc. "It's something we wanted to do, but you have to make choices where you grow. You can't get everything all at once."

Losing seasons have a way of burnishing deeper introspection, though, and when John Henry read Big Data Baseball, Travis Sawchik's 2016 book about the use of analytics by the Pittsburgh Pirates, Scott found a more willing ear.

"It just sparked some questions [from him]. 'Hey, are we doing this?'" said Scott, who presented Henry with two project lists — the one the Red Sox could currently attack, and the theoretical one they could complete with a larger staff. "We were just hoping to be closer to the top. That was always what John wanted when he bought the team."

The Red Sox added one analyst in 2014, two more in 2015 and two developers in 2016. They've added two more full-time analysts for this season, as well as making a former part-time assistant into a full-time coordinator. A staff of three or four is now closer to a dozen.

As a result, the long-awaited transition from Carmine to a newer, more user-friendly database is underway. Beacon will gradually replace Carmine over the next two seasons.

"Carmine was really designed for a power user, and quite honestly a lot of those power users left," Scott said. "We weren't optimizing it, and a lot of the reason for that was people weren't as comfortable because they were newer people. They didn't know where to go."

"We believe our system needed some upgrading to get more contemporary, and we added personnel to make some changes and continue to grow," said Dombrowski.

Scott conceded that, with the analytical expansion, "you hoped it would happen sooner." But at least it's happened aggressively over the last three years, at a time when more and more data pours into the game. In the second half of the last decade, PITCHf/x data changed the way we analyzed pitching. Now, Statcast and other player tracking data are dramatically altering the way we analyze all aspects of the sport.

To steal Scott's phrasing, player tracking data is changing analytics from a question of what to a question of how. Let's take an outfielder's arm as an example.

An outfielder's arm used to be judged solely by the number of assists he accumulated. Deeper research allowed you to look at how often runners tested his arm; the best arms might not record too many assists, because nobody ran on them.

With player tracking data, you can analyze why a runner doesn't challenge an outfielder's arm. Is it because he gets to the ball more efficiently? Is it because of his arm strength or his arm accuracy? How should those skills be valued?

"Now we can look at it more granularly," Scott said. "We get to that next level down, which is more aligned with traditional scouting. You get to the tools and now measure the tools."

More so than a decade ago, every team in baseball understands the existential importance of a strong analytics department in the game. But as it gets harder to gain an edge in analytics, the ones you do discover become ever more significant.

"I wouldn't say analytics is more or less important than it was 15 years ago, but I do think it is more important now that we put more resources toward it just because of the amount of data," O'Halloran said. "The landscape of gaining competitive advantages is more and more difficult unless you deploy resources — and the right resources."

"There's always stuff on the horizon we probably don't even know are coming," said Scott. "All of a sudden you're going to have this new dataset, and you want to always be in a position to start analyzing that as soon as possible. There's a huge benefit to being first on something."

At its heart, though, being a Red Sox software developer or analyst isn't that different from being Boston's shortstop or center fielder.

"Everyone needs to do their job better than the 29 other people that are in a similar position," Scott said. "If we all do that, we'll be better."