Dean Oliver is the Director of Production Analytics at ESPN, leading the development of sports analytics for the network. He worked for seven years in the NBA with the front offices of the Denver Nuggets and Seattle Supersonics, using analytics to improve trades, free agency, the draft, and coaching tactics. The contents of his book, Basketball on Paper, are now used broadly across the NBA. Oliver has a Ph.D. in engineering and has years of experience as an advance scout, collegiate scout, and in coaching.
GSABR: How did you get your start in sports? How has your education and
basketball experience helped you get to where you are today?
DO: I played basketball as an undergrad at a D3 school. But in high school, I read Bill James books on baseball and looked to apply them in basketball. A big thing was taking the approach that James had, but setting aside the details to understand basketball as a very different sport. I just sat down and charted basketball games in some detail and that opened up a world of structure that I hadn’t seen before.
GSABR: What were some of your day-to-day responsibilities with the Sonics and
Nuggets? How were they similar/different from what you do with ESPN?
DO: With a team, the responsibilities are either on the personnel side or coaching side. It varies over the course of a season. Leading up to the draft and free agency, it is a lot of personnel evaluation. This also can come before a trade deadline to evaluate potential trades. Prior to a season, once a roster is pretty well set, there are a lot of coaching side responsibilities. During the season, there are ongoing coaching side responsibilities, some of which being automated, others that are not.
At ESPN, it is entirely different. We are really developing frameworks for evaluating teams and players across sports. The application to making decisions has as yet not happened, though it probably will.
GSABR: What aspects of analytics are teams and companies most willing to buy
into? What aspects are they most reluctant to buy into and why?
DO: Personnel evaluation seems to be the easiest for teams to buy into because the methods for getting it right are a little more straightforward. Coaching tools are tougher because they really involve managing large groups of people and how they fit, a more complex math problem.
GSABR: With ESPN’s wide appeal and entertainment-based segments, how do you
reconcile arcane advanced stats and analytics to make them appeal to a mass
DO: The story is the thing. Understand the story. If stats can help tell that story, they aren’t arcane. A lot of times, conventional stats are considered misleading. That’s a situation where a more analytical stat fills in a gap and, while it may need some explanation, that explanation should be part of the story.
GSABR: What skills should students looking to break into sports analytics have
and how can they leverage them to do good and actionable work?
DO: Data – understand data and databases, how to get data in and how to get it out.
Programming – understand how to automate projects, how to logically manage data.
Statistics – know the tools for getting structure out of data. Regression in its many forms is very useful. Theoretical stats also helpful, as can be engineering concepts.
Communication – Being able to easily communicate analytical concepts and numbers is not necessarily easy. It’s not just supporting an opinion with one number, it is understanding the analysis that goes into a number that usually has the power.
Special thanks to Dean Oliver for his time and insight
Interview by Nik Oza, Georgetown Class of 2016