The Pittsburgh Pirates are in baseball’s playoffs for the first time since 1992. Their spot in the postseason was assured after a barnburner of a game in Chicago that saw a home run in the top of the ninth and a play at the plate in the bottom of the inning to end the game (beautiful pics available at Deadspin). But as a great story by Travis Sawchik in the Pittsburgh Tribune-Review makes clear, the Pirates’ successful season was not built around long balls and thrilling plays. It was built around data, analysis, and organizational change.
The story relates how in 2013 the Pirates fully embraced data-driven baseball, especially on defense. They shifted infielders around like crazy from batter to batter, and also got their pitchers to throw more sinkers and other balls low in the strike zone in order to induce ground balls through the infield.
They did this mainly because they were tired of losing — after 20 consecutive below-.500 seasons — and didn’t have other options for turning the team around. As Sawchik writes,
With little money to spend on free agents, few impact prospects ready for 2013 and many outside the organization calling for a regime change, [general manager Neal] Huntington and [manager Clint] Hurdle discussed how to get more production from their returning roster.
The way to do this, they concluded, was to start really listening to the geeks the team had hired:
Before… Huntington…, the organization did not have an in-house analytics department. Entering this season, the Pirates had five full-time staffers dedicated to data architecture and analysis and their own in-house proprietary computer system…
The Tribune-Review article details what the analytics department came up with, how it worked with the players and coaches to put the new approaches into play, and how successful they’ve been. It’s a great read for baseball and data fans alike; please do check it out.
Here are the generalizable insights about data-driven management I took away from the 2013 Pirates turnaround:
Experience and Folklore are Good. Analysis is Better. After more than a century of playing baseball, there’s a lot of received wisdom about where to position the fielders to stop as many batted balls as possible. And MLB players themselves have built up a huge base of experience that gives them good intuition about where exactly to play in a given situation.
Even with all this, though, they’re still not positioned optimally:
[Director of Baseball Systems Development Dan] Fox researched where balls historically most often had been hit. He took evidence to Huntington that suggested the Pirates should change their defensive alignment. Fox suggested the Pirates not only increase their use of shifts but also alter where defenders, particularly infielders, are placed in base defenses.
“We’ve played infield positions conventionally for years, but if you look at hundreds of thousands of balls put in play, data shows that we’re actually off a little bit,” Huntington said.
The Geekery Must Pervade. Data-driven approaches work best when everyone follows them, all the time. Infield shifts don’t work if pitchers don’t keep the ball low in the strike zone, so position players, pitchers, and their coaches all have to sign up to the new approach. And to get maximum benefit, everyone has to turn off their own intuition and go (literally) where the data take them.
This takes time and is not natural for people:
“When I first came over to the Pirates, you could consider me as an old-school guy. But numbers don’t lie,” [third base coach Nick] Leyva said. “Halfway through last year for myself, I really bought it. … I was probably using maybe 50, 60 percent of what I was getting from stat guys last year. Now I’m close to 100 percent.”
Some still aren’t used to it:
There has been some resistance. Starting pitcher A.J. Burnett was visibly upset with shortstop Clint Barmes’ positioning in a game at Texas last week. “I don’t have a problem with Clint Barmes,” Burnett later told reporters. “I have a problem with (expletive) shifts.”
Burnett should get over his problems. He’s having his best season in years.
You Don’t Have to Change the Guard; The Old Guard Can Change. Burnett’s attitude notwithstanding, most of the Pirates personnel have bought into data-driven baseball, even though few of them have longstanding geek tendencies:
“It was definitely a transformation in understanding the game. It has been for me for the last 10 years, especially the last five years,” Hurdle said. “You have to get involved in the information. You’ve got to read. You’ve got to study. You can’t just stick your head in the sand and just say, ‘It doesn’t exist. It doesn’t count. It doesn’t make sense.’ ”
If crusty, old-school baseball managers can learn become data-driven, then the ones in your organization have absolutely no excuse.
Data Keeps Delivering. Michael Lewis’s Moneyball came out in 2003 and popularized the story of the geek takeover of baseball, which has transformed the game. When I watch the Sox at Fenway, for example, scoreboards tell me about OBP and OPS for the batters, stats that were never shown or discussed when I was a kid watching the Cubs.
Yet even a decade later, there’s still ample opportunity to use data to make significant improvements, not incremental ones, in how to play the game. Detailed data are now available on every pitch thrown, and every ball hit. This trove is being mined to change our understanding of, among other things, how to play defense. Legendary baseball man Branch Rickey once remarked in frustration that “There is nothing on earth anybody can do with fielding.” But that’s not the case any more; thanks to newly available data, the Pirates are doing a lot with fielding.
Your Competitors are (Probably) Missing Out On This. Amazingly enough, the Pirates are still in the minority of teams who have substantially changed (let alone ‘revolutionized’) their approach to fielding and pitching by being data-driven. Most teams still play defense the same way it’s been played for at least a century. I imagine they’ll eventually catch on and change, but the game of baseball shows us that demonstrably better ideas — at least those associated with data and analysis — do not spread quickly, even when competition is intense and performance gaps are visible and large.
What side of that performance gap do you want to be on?