Steven Shi
Like the protagonist of Bennett Miller’s widely acclaimed, 2011 baseball movie titled Moneyball, Dan Cervone, a Senior Analyst in Research and Development with the Los Angeles Dodgers, is an expert in baseball data analytics. He spoke at UCSB’s Data Science Club meeting on Jan. 23 to discuss the implications of data science in Major League Baseball.
Since the inception of data analytics in baseball, many franchises have raced against each other to develop cutting-edge analytic techniques and models in order to optimize player and team performance. According to Cervone, data science is “kind of like a Moneyball thing.”
“Data is one of the ways teams can differentiate themselves from each other,” said Cervone.
Indeed, for many baseball franchises, data science plays a defining role in determining team rosters and strategies. Using different data sources such as tracking data, play-by-play data, and hand charted data, Cervone’s analytics team helps the Dodgers break down each player’s performance into categories of “skill, luck, and circumstance,” allowing them to form in-game strategies..
“The most successful thing I had a big role in is our defensive positioning,” Cervone said, recalling his proudest accomplishment in baseball analytics. “It was a problem that we really had almost nothing when I came in, and there were many levels to the problem.” Cervone helped create algorithms and models to map out and optimize defensive strategies.
“It was a really fun accomplishment because the impact on the field was very direct and immediate,” continued Cervone. “It is a really good example of modeling that is done to stitch together different ingredients to optimize the problem.”
Yet, Cervone also notes that his team faces a unique set of challenges when applying or discussing their work. They communicate with a diverse set of people — many of whom do not have data science backgrounds — and must clearly explain the meaning and applications of the data
Furthermore, unlike other organizations that utilize and share new data science technologies, baseball franchises often isolate themselves to protect proprietary strategies. Indeed, Cervone notes that because “30 teams are competing against each other, [they] are more siloed than other industries.”
Cervone, a graduate of Harvard University’s Doctoral program in statistics, joined the L.A. Dodgers Analytics team in December of 2016 after the Dodgers came under new ownership. Cervone’s graduate work in sports analytics, along with his experience in modeling basketball and soccer, attracted the attention of the team’s front office.
Addressing students who are interested in sports analytics, Cervone advised, “I think the best way to get into the sport analytics field is to do some sort of research project in sports data, whether its a class project or creating a poster at a talk. We are constantly looking at Twitter, Reddit, and blogs for this kind of stuff. We definitely find it and it definitely helps people stand out.”