Moneyball-Brad Pitt-Jonah H
Brad Pitt and Jonah Hill star in the film “Moneyball,” which hits theaters today, and Hill’s character is based on Mets executive Paul DePodesta, who spoke at the Strata Summit this week.
Long before “big data” analytics was trendy, Paul DePodesta was taking it to the majors, and today his story will be told on the big screen.
A period of DePodesta’s career is depicted in the film “Moneyball,” which premieres today on more than 3,800 screens across the U.S. The film is based on the best-selling 2003 book by Michael Lewis and chronicles the data-driven revival of the Oakland Athletics. It follows Athletics general manager Billy Beane and DePodesta, who used computer analysis to identify undervalued players. The character based on DePodesta has been renamed Peter Brand and will be played by Jonah Hill.
During a presentation at the Strata Summit in New York on Tuesday, DePodesta, now vice president of player development for the New York Mets, reflected on the role of performance analytics in baseball and the lessons that can be applied to data-driven organizations. When he arrived in Oakland, small-market teams with limited budgets like the Athletics were at a disadvantage in bidding wars with wealthier teams from markets like New York and Boston, DePodesta recalled.
“We had to come up with a different way,” DePodesta said, “It was like shopping at 7-Eleven while preparing a gourmet meal at the same time.”
Data vs. the subjectivity of scouting
The solutions Beane and DePodesta adopted were influenced by a school of statistical analysis in baseball called sabermetrics (named after the Society for American Baseball Research), and were often at odds with traditional methods of scouting players.
Depodesta 1
“Evaluating players was very subjective,” he said. “We had a whole new set of evaluation criteria that we’d never seen before. We didn’t solve baseball’s problems, but we did reduce the inefficiencies in decision-making.”
DePodesta spoke to an audience of executives and data scientists about the process of data-driven decision-making and how to avoid analytical errors that lead to erroneous conclusions. Often, the challenge is looking at the data dispassionately, which often involves filtering out emotional reactions to the data or to player performance.
“We’re always looking for causality, and that can be deceiving,” DePodesta said. “We often get stuck on things and don’t always know why.”
Common biases in data analysis
“Confirmation bias” is easy to set in, DePodesta said. “Once we make a decision, we resist any information that doesn’t fit that conclusion,” he said.
A particular problem in baseball is “appearance bias,” or the idea that some players resemble better baseball players than others. DePodesta said this is also an issue in business, citing Malcolm Gadwell’s data point on height and business success. Gadwell said that while only 3.9 percent of American men are over 6’2″, about 30 percent of Fortune 500 CEOs are over 6’2″.
To make good decisions, we need to get rid of those biases.
“We relied on data as a flashlight in the cave, a guiding light,” DePodesta said. “We said, ‘If I can’t prove it, I don’t believe it.’ We had to keep asking the simplest questions. The only thing we stuck with was an open-minded mindset.”
