Basketball Sabermetrics

Shamrock · March 25, 2004

by Harlan Schreiber (1/10/04)

Dean Oliver is one of the foremost writers in the field of statistical analysis of basketball. Basketball has been his passion for over 15 years. First, he established The Journal of Basketball Studies and later he founded APBR_Analysis the premier chat session for discussion of dynamic and complex basketball related issues. He has recently created a stir in the basketball analytical community with his book "Basketball on Paper," a thorough re-examination of all facets of the game of basketball through the lenses of statistical analysis and a host of other innovative analytical tools.

Question: What inspired you to write "Basketball on Paper."

Dean Oliver: The long story is that I wanted to write something like this for a long time. Reading Bill James [analytical baseball]books in the 1980s inspired me to try to create similar tools for basketball. I tried a lot of different methods that didn't lend much insight. Finally in 1987, I started to have some breakthroughs by recording stats manually which eventually led to a lot of the tools I now use. I toyed with the idea of writing a book for a while in the late 1990?s and I wrote a few chapters. Finally in 2001, I e-mailed Bill James some of my work. Three months later, James e-mailed me and said that he couldn't help me because he was writing his own book on basketball. I thought uh-oh. I better write something before he does and that spurred me to sit down and actually create the book.

Q: When did you actually first start writing this book?

DO: Back in 1994-95, I started the Journal of Basketball Studies on the web. It forced me to write down a lot of the methods I had been thinking about in my head so that people could understand that. I experimented a lot and getting my ideas out in the public really helped me refine my theories. I actually sat down to write the book in February of 2002.

Q: Do you have any new ideas on the horizon that we have not seen in the book?

DO: I think 50-60% of what I do is in the book, so there are many things that you haven?t seen. These are particularly regarding game strategy that come from my understanding of a number of different disciplines, including political science and game theory.

Q: What kind of statistical training and basketball training is in your background?

DO: I went to Caltech, a small science school of only about 200 per class, where I received an engineering degree and played, coached, and took statistics for the basketball team. I then went to get my PhD in statistical applications to environmental engineering at the University of North Carolina. I went to UNC in some part because of its great basketball tradition. There, I did collegiate scouting with Bill Bertka [then the Laker assistant coach and currently a consultant to them] and his scouting service, Bertka Views.

Q: Have you heard from Bill James since your book was published?

DO: I did send him a copy of the book out of appreciation for his inspiration. I heard back from him in early-December and he said it was ?excellent writing? and that he ?learned a lot.? I was elated because he is not the type to heap praise lightly.

Q: Do you consider yourself the "Bill James of basketball?"

DO: What I do, in terms of background, merging basketball and science is rare. I think my focus is different than James'. I focus on different issues and strategies. I try to look into quantifying chemistry, which is not a big issue in baseball. There a lot of other basketball analysts who primarily evaluate individual numbers and I look at some other stuff.

Q: What other basketball writings have influenced your work?

DO: More than anything, I've read a lot of coaching books. Dean Smith and Frank McGuire talked about the value of possessions long before any statistical analysis of basketball came around and they actually did some basic statistics. I've obviously also read Dave Heeren, the creator of Tendex, Martin Manley, Bob Bellotti, David Claerbaut, and John Hollinger. A lot of their work focuses on value of players by placing linear weights and attempting to find the true worth of certain statistical markers.

Q: What do you think of the linear weights model of basketball analysis?

DO: It's a good screening tool. It adds the good statistics and subtracts the bad stuff. This works in certain contexts. I personally don't think that finding a value or weight for an assist or a rebound illuminates our understanding of how to build a good basketball team or how to most efficiently use players.

Q: What is the untapped frontier in basketball analysis?

DO: Evaluating college and high school stats and translating them into meaningful predictors for an NBA career. There have been a number of small scale attempts to date but nothing too substantial. This is because the data, especially for high school players, is so limited at this point. But if someone could come up with a reliable system it would be very helpful.

Q: Do you think that we have meaningful defensive statistics for individual players?

DO: Officially recorded defensive statistics ? no. But I'm relatively content with the kind of system I?ve proposed. In researching this topic for the book, I learned a lot. Most notably, having a defensive system ? as in a defensive scheme on the floor ? is very important. Instinctively, I think I knew coaching made a difference but sitting down measuring this issue confirmed it. I've heard through the grapevine that some teams have developed their own defensive evaluation systems and do some more detailed analysis of individual defensive effort. I haven't seen any of this yet but I'd love the chance to see what's out there.

Q: Have you gotten any feedback about the book from anyone inside the NBA?

DO: I showed the book to my people with the Sonics, where I do some consulting work. [sonic President] Wally Walker, assistant GM Rich Cho, and scout Yvan Kelly gave me very positive feedback which felt great. They were obviously already familiar with my work because they tracked me down through my JOBS website back in 2000. A West Coast scout for the Celtics has also been feeding my ego for a few weeks now.

Q: Do you plan on writing any follow ups to Basketball on Paper?

DO: When I finished this book, I had another 15 chapters I'd love to have written but couldn't fit in. Now that this book is out I've been thinking about how I'm going to top it. I'm still thinking about that. There are a lot of add ons that I use that would be good for another book but I can't say I'll have anything out right away.

Q: Since you don't really use linear weights, how do you evaluate a player statistically?

DO: Evaluation varies from player to player. Take Allen Iverson. Whether he's shooting 12 shots or 30 shots, his shooting percentage is modest. He is always taking bad shots but he'll make a decent amount of them regardless of whether he's being draped by a defender or he's wide open. Conversely, he misses a bunch of shots when he's not being guarded.

For some reason, Brent Barry is a lot more efficient at 12 shots per game than he is at 15-20 shots. We understand these scenarios observationally but my challenge is to identify the efficiency curves of each player. This is a tool that I had to develop. It tells a lot about how players fit together, which to me is much more important than the linear weight method which doesn't address context.

Another good example is in Los Angeles with Shaq and Kobe. Last year, they were surrounded by non-scorers so they looked like the best players in the league. This year, with Gary Payton and Karl Malone, linear weights show Shaq and Kobe too be way down in production. This doesn't reflect their true value. The efficiency curve tells a different story of their value. This addresses the question of how to optimize efficiency and how to best distribute shots. The curve told us that in 2002-03, when Shaq or Kobe couldn?t use possessions (through injury, for example), Rick Fox could best handle the excess, even though Derek Fisher and Robert Horry were more effective with Shaq and Kobe around. So really, the efficiency curve addresses the most interesting questions to me about optimal distribution of offensive possessions.

Q: Have you spoke with any coaches about the book?

DO: Well, I'll speak with [sonics coach] Nate McMillan next week. Some of his assistants have gotten copies of the book as well. I also know that both the Van Gundies have copies. I have talked directly with high school and collegiate coaches about the book and they have a ton of interesting questions, which is great. They even address issues that I hadn't thought of like developing tables that tell a coach when to go for a three pointer or foul late in a game.

Q: What do you think about the paradigm that Hubie Brown is trying to create in Memphis, with ten players getting equal opportunity minutes?

DO: I don't necessarily like egalitarian teams in the abstract. But the Memphis idea is good because they have the talent that lends itself to playing a lot of guys without having much drop off in quality. It also avoids the issue of fatigue from riding the same players too long. There are some possible chemistry issues from this type of system but a strong willed coach can lessen this issue.

This is where some of the psychological and people management theories I have read about become important. Some players will respond to this system and consider it fair and there are certain guys who will always be ticked off. If the coach emphasizes hard work, most players will respond to him, at least in the short term.

Q: Do you think the discourse in the basketball analysis field is at the same level as it is in the baseball?

DO: In terms of quantity, no. But in terms of quality, we are doing very well. There has been some revolutionary stuff produced in baseball but it has slowed down in recent years. Innovations in basketball analysis are coming at a more rapid rate now and more people are coming out with intelligent basketball analysis and the discourse continues to get better. There is an opportunity for someone who is willing to work at it to make a good contribution in this field in the coming years.

Dean Oliver is the author of Basketball on Paper. He can be reached at http://www.rawbw.com/~deano/.

HOOPSANALYST

Contact us: [email protected]

The Van Gundies are reading this man's book...hmmm.