Interview with David Berri, author of _Stumbling on Wins_

Today features something a little different. Sports economist David Berri has a new book out, Stumbling on Wins. I was a big fan of Berri's first book, Wages of Wins, and regularly read his blog. I sent him a few short questions.

PVW: Whenever I read you write about your work, you seem to put it in terms of behavioral economics, but I've never actually heard you use the term. Do you identify yourself within that camp as an economist?

To answer your questions I am going to refer to the summary of behavioral economics offered by Alan Krueger at Princeton.

Krueger offers the following description of behavioral economics:

As Krueger notes, this is one of the fastest growing fields in economics. Stumbling on Wins is but one book drawing upon research from this field. For example, in Predictably Irrational, Dan Ariely presents evidence – drawn from laboratory experiments -- that human beings frequently fall short of the rational ideal. Nudge – by Richard Thaler and Cass Sunstein -- show how once we understand that people are not perfectly rational better policies can be designed. And How Markets Fail (by John Cassidy), The Myth of the Rational Market (by Justin Fox), and Animal Spirits (by George Akerlof and Robert Shiller), all demonstrate how behavioral economics can explain the causes of our recent economic downturn. So this is not just a field that is growing, it's also an important area of research.

PVW: It's been a few years since I read your excellent first book, Wages of Wins. As I remember it, the book was disproportionately weighted towards basketball, though it certainly had some chapters in other sports as well that would have been worth the price on their own. Does this book contain a similar percentage about basketball?

Stumbling on Wins covers research from baseball, hockey, basketball, and American football. The NBA is the primary focus of Chapter Two (Defending Isiah), Chapter Six (The Pareto Principle and Drafting Mistakes) and Chapter Eight (Is it the Teacher or the Students?). Hockey is the primary focus of Chapter Three (The Search for Useful Stats). American Football is the focus of Chapter Four (Football in Black and White), Chapter Five (Finding the Face of the Franchise), and parts of Chapter Seven (Inefficient on the Field). And baseball comes up in Chapter One (Maybe the Fans are Right) and parts of Chapter Six and Seven.

Essentially the book is trying to show that the Moneyball phenomenon (decision-makers in baseball were not valuing players correctly for much of baseball history) is something we see in all the major North American sports.

PVW: Did any parts of your new book surprise you when you first did the research?

Let me refer to what has become a classic cliché in answering your question. Specifically, it was Albert Einstein who said, "If we knew what it was we were doing, it would not be called research, would it?"

I think that quote captures the unpredictability of research. For each study we are asking some basic questions: What determines draft position of an NFL quarterback? What factors determine the pay of an NBA free agent? What factors determine wins in the NBA? How consistent are NHL goalies across time?

The answer to each question isn't known before we start investigating the issue. And the answers – given what we understand of the conventional wisdom – are all somewhat surprising.

PVW: What parts of the new book do you think will cause the most controversy?

This is hard to answer. The story that was picked up on first was the story on the impact a Final Four appearance has on an NBA's players draft position. I would not have guessed that story was going to be noted first and that demonstrates how hard it is to predict how people react to the stories you tell.

PVW: In 5 years, what overvaluations and undervaluations in the NBA do you think will persist?
This is one of the key questions addressed in behavioral economics. How quickly does new information get adopted by decision-makers? In traditional economics it is imagined that information is adopted instantaneously. In reality, though, the process seems much slower. We note in the book, baseball had the data to calculate on-base percentage since the 1860s. But it doesn't appear that it was correctly incorporated in the valuation of players until the 21st century. That record suggests that the overvaluation of scoring in the NBA might persist for some time to come.

PVW: If charges drawn was kept as an NBA stat, do you think it would improve the explanatory power of Wins Produced?

No, the explanatory power is based on the connection offensive and defensive efficiency has to wins. But it would allow us to connect part of what is currently counted for the team to the individual players. So our evaluation of players would be improved (although I doubt it would be much of an improvement).

Let me offer a few more details on Wins Produced:

The model used to measure player performance begins with the simple notion that wins in the NBA are determined by a team's offensive and defensive efficiency. From this statistical model we can derive the value – in terms of wins – of much that is in the NBA box score. These values tell us that wins in the NBA are primarily determined by shooting efficiency and the ability of a team to capture and maintain possession of the ball. This means that an NBA player helps his team win when he is an efficient scorer, grabs rebounds, captures steals, and avoids turnovers (i.e. the possession factors). Blocked shots, assists, personal fouls do matter. But shooting efficiency and the possession factors matter more.

The model explains 95% of team wins and is much more consistent across time than the plus-minus models. But it does tell us that players who score inefficiently, and/or who are below average with respect to the possession factors, do not help a team win very much. And this can be true even if the player scores more than 20 points per game. As a consequence, Wins Produced can produce results that contradict what people generally believe about NBA players. Again, popular perception is driven by scoring. So as we argue in the Stumbling on Wins, perhaps what we see from Wins Produced shouldn’t be that surprising.

PVW: Now, for a change of pace: which do you think will be easier to model: american football or football (soccer)?

I am currently doing more work on both sports. My sense is that soccer will be easier. But then again, research is often surprising. So I am not sure that will be my answer when my co-authors and I finish our research.

Thank you, Professor Berri.

Note: for the sports fan clicking through, this is a blog about Texas politics, so you'll probably be disappointed. The only post even remotely related is Rick Perry sponsoring Bobby Labonte. Although I'm not sure that the sabermetric statheads are NASCAR fans, perhaps they are.

Posted by Evan @ 04/08/10 12:09 AM


Previous Entry | Home | Next Entry


No comments yet

Add Comments

No flames or impolite behavior. HTML will be stripped. URLs will be transformed into hyperlinks.