# A Billy Beane for Basketball?

According to this article in Wired, a man named Dean Oliver is trying to do for basketball what Bill James and Billy Beane did for baseball: create and exploit new metrics in order to better distinguish players who win from those who simply generate gaudy traditional stats. The Wired article is written by Hugo Lindgren, who is in some measure responsible for Freakonomics. Hugo used to be an editor at The New York Times Magazine, and it was Hugo who sent me to Chicago a couple years ago to interview Steve Levitt. Thanks, Hugo. And nice piece. (Thanks to Adam Scott of www.angryman.ca for the link.)

#### rarely right

I would expect there to be Billy Beane's for every professional sport - I'm sure that at some level there already are. One of the reasons why is that Fantasy Leagues lend themselves to endless statistical analysis of the players and games.

Here are two applied statistics questions I have about sports:

1. In tennis, is it more effective to hit the second serve at the same speed as the 1st serve or is it more effective to take speed off it and increase the odds that it lands in? Each player can do a simple analysis of their success rates on 1st serve vs. 2nd serve and come to a meaningful conclusion.

2. In golf, is it better to blast one's drive as far as possible or just bunt it down the middle of the fairway? Top pros have foresaken high levels of driving accuracy under the premise that it's easier to make birdie from the rough 100 yards out than it is to make birdie from the middle of the fairway 175 yards out.

#### drm7

There was a similar (and fairly balanced) article in Sports Illustrated about a month ago.

On the topic of golf, long hitters have profited from reducing the length of their drives - Phil Mickelson consistently fell short in the majors until he became more conservative. Davis Love III (a tremendously long hitter) didn't win regularly until he shortened his swing. Even Tiger Woods has taken a little bit off of his drives.

#### Dr. Funk

If you assume that speed of a serve and accuracy are inversely related, it's fairly easy to show that 2nd serves should be slower. The data backs this up.

#### Erik

As a former math major and a lifelong basketball player and fan, I'd recommend unconditionally one of Dean Oliver's peers. John Hollinger has written an annual review of new statistical methods, team analysis and player reviews.

Basketball teams are well behind baseball in using statistics to measure performance. Each team does have its own metrics beyond the basics you read about in the papers. Pat Riley was one of the early coaches to start tracking hustle stats (passes contested, loose balls run down, shots contested, etc.). The key difference is these stats though are primarily to motivate players and reward hustle plays rather than measure players' effectiveness.

#### deano

As Freakonomics was one of my favorite recent books, it's good to see my basketball work getting mentioned here.

I think the statistical sleuthing of Levitt is very much the type of thing we do in basketball. Though I don't highlight the high level math in Basketball on Paper (the book that got me my ticket to the NBA), it does take very high level analysis to handle basketball's complex interactions. It is not surprising at all that the type of very careful regression analysis done to support Freakonomics helps a lot in hoops. The key thing is building an underlying structure for understanding the game, which is what Basketball on Paper was about.

Though I can't speak precisely to the tennis and golf questions, I can say that what you're looking for are rules of thumb. And rules of thumb are called that for a reason -- they're useful but they're clumsy. They help if you don't know much, but can hurt you if you ignore the details. If you're lousy at getting out of the rough, you better stay on the fairway (my game). Basketball has a lot of rules of thumb. We use them, but we would rather look for better info.

#### Dan Rosenbaum

This was a nice piece on Dean Oliver, and he has been instrumental in making basketball statistics respectable. I also do some work on basketball statistics, and Dean has been a great influence on all that we do.

But I am an economics professor and I have to use this opportunity to tell my "Steve Levitt" story.

This was back in 1997 when I was about to be in the "market" for Ph.D. economists. I was in O'Hare waiting to fly to Boston for an NBER conference. And I look across the way and I see none other than James Heckman (who since has won a Nobel Prize). He is talking to a guy who looks to be my age who I assume is another graduate student like me.

Well, I listen on and off to their conversation but I don't introduce myself. I get on the plane, get off in Boston, and get in the cab line. At that point I know that Heckman, this student, and I are all going to the Royal Sonesta and I suggest that we get a cab together.

We get in the cab and I sit in the front and Heckman and this student sit in the back. The whole way I listen to their conversation thinking "My God, is this how smart all of the graduate students on the market are going to be? If so, there is no chance in Hell that I am going to get a job."

But I keep that to myself and we get to the Royal Sonesta and check in. I check in to my room and Heckman checks in to his. Then this student tries to check in to his room, but they do not have a reservation for him. It was pretty late at night, so I figured Heckman would offer to share his room with this student that he had been talking with for hours. But I waited and waited and that offer did not seem forthcoming.

So I checked with the attendant and confirmed that I had two beds in my room (which I did) and then I introduced myself to this student and offered to share my room with him.

And then this student introduced himself to me. He was none other than Steve Levitt - who by that time I think was already tenured at Chicago. Which made me feel a lot better that not all graduate students were as smart as this guy.

Well, Steve was an incredibly nice and humble guy despite the fact that by that time he already was a pretty well known name. He was very gracious in giving me advice about the job market and I still remember him tiptoeing around the room in the morning so as not to wake me up.

I apologize for taking up so much space on this blog, but I thought this might be a story this crowd might get a chuckle about.

Best wishes,
Dan

#### daveslutzkin

Dr Funk: "If you assume that speed of a serve and accuracy are inversely related, it's fairly easy to show that 2nd serves should be slower. The data backs this up."

Yeah, but this assumes that speed of a serve has no effect on returnability other than accuracy. So logically you could serve very slowly but always put the ball in the corners and win every point.

The _perception_ is that it's better to slow down the second serve and not give away a free point, but if slowing it down makes it much easier for the opponent to gain control of the point, maybe it's better to hit it hard. You'll serve more double faults but when it goes it the opponent will more likely be on the defensive. The only real measure is second serve win percentage, not second serve non-fault percentage.

#### hart2412

Last week the Wall St. Journal Weekend Edition had a similar story about the growing use of statistics in basketball to exploit relative inefficiencies in player management.

I confess to being in the legion that believe in this approach, and have recently joined with 82games.com to help with a compilation of various statistics charted throughout an NBA game, with the hope of showing the power that numbers can have on team performance maximization.

Great job with this blog and keep up the good work.

#### DENVER LENDER » Blog Archive » Basketball by the Numbers

[...] I came across an article on the Freakonomics blog about Dean Oliver - a statistician for the Seattle Supersonics. He’s a statistics guy who’s trying to formulate a philosophy similar to Billy Bean regarding the stats are truly important for basketball players. The article itself appeared in wired.com [...]

#### rarely right

I would expect there to be Billy Beane's for every professional sport - I'm sure that at some level there already are. One of the reasons why is that Fantasy Leagues lend themselves to endless statistical analysis of the players and games.

Here are two applied statistics questions I have about sports:

1. In tennis, is it more effective to hit the second serve at the same speed as the 1st serve or is it more effective to take speed off it and increase the odds that it lands in? Each player can do a simple analysis of their success rates on 1st serve vs. 2nd serve and come to a meaningful conclusion.

2. In golf, is it better to blast one's drive as far as possible or just bunt it down the middle of the fairway? Top pros have foresaken high levels of driving accuracy under the premise that it's easier to make birdie from the rough 100 yards out than it is to make birdie from the middle of the fairway 175 yards out.

#### drm7

There was a similar (and fairly balanced) article in Sports Illustrated about a month ago.

On the topic of golf, long hitters have profited from reducing the length of their drives - Phil Mickelson consistently fell short in the majors until he became more conservative. Davis Love III (a tremendously long hitter) didn't win regularly until he shortened his swing. Even Tiger Woods has taken a little bit off of his drives.

#### Dr. Funk

If you assume that speed of a serve and accuracy are inversely related, it's fairly easy to show that 2nd serves should be slower. The data backs this up.

#### Erik

As a former math major and a lifelong basketball player and fan, I'd recommend unconditionally one of Dean Oliver's peers. John Hollinger has written an annual review of new statistical methods, team analysis and player reviews.

Basketball teams are well behind baseball in using statistics to measure performance. Each team does have its own metrics beyond the basics you read about in the papers. Pat Riley was one of the early coaches to start tracking hustle stats (passes contested, loose balls run down, shots contested, etc.). The key difference is these stats though are primarily to motivate players and reward hustle plays rather than measure players' effectiveness.

#### deano

As Freakonomics was one of my favorite recent books, it's good to see my basketball work getting mentioned here.

I think the statistical sleuthing of Levitt is very much the type of thing we do in basketball. Though I don't highlight the high level math in Basketball on Paper (the book that got me my ticket to the NBA), it does take very high level analysis to handle basketball's complex interactions. It is not surprising at all that the type of very careful regression analysis done to support Freakonomics helps a lot in hoops. The key thing is building an underlying structure for understanding the game, which is what Basketball on Paper was about.

Though I can't speak precisely to the tennis and golf questions, I can say that what you're looking for are rules of thumb. And rules of thumb are called that for a reason -- they're useful but they're clumsy. They help if you don't know much, but can hurt you if you ignore the details. If you're lousy at getting out of the rough, you better stay on the fairway (my game). Basketball has a lot of rules of thumb. We use them, but we would rather look for better info.

#### Dan Rosenbaum

This was a nice piece on Dean Oliver, and he has been instrumental in making basketball statistics respectable. I also do some work on basketball statistics, and Dean has been a great influence on all that we do.

But I am an economics professor and I have to use this opportunity to tell my "Steve Levitt" story.

This was back in 1997 when I was about to be in the "market" for Ph.D. economists. I was in O'Hare waiting to fly to Boston for an NBER conference. And I look across the way and I see none other than James Heckman (who since has won a Nobel Prize). He is talking to a guy who looks to be my age who I assume is another graduate student like me.

Well, I listen on and off to their conversation but I don't introduce myself. I get on the plane, get off in Boston, and get in the cab line. At that point I know that Heckman, this student, and I are all going to the Royal Sonesta and I suggest that we get a cab together.

We get in the cab and I sit in the front and Heckman and this student sit in the back. The whole way I listen to their conversation thinking "My God, is this how smart all of the graduate students on the market are going to be? If so, there is no chance in Hell that I am going to get a job."

But I keep that to myself and we get to the Royal Sonesta and check in. I check in to my room and Heckman checks in to his. Then this student tries to check in to his room, but they do not have a reservation for him. It was pretty late at night, so I figured Heckman would offer to share his room with this student that he had been talking with for hours. But I waited and waited and that offer did not seem forthcoming.

So I checked with the attendant and confirmed that I had two beds in my room (which I did) and then I introduced myself to this student and offered to share my room with him.

And then this student introduced himself to me. He was none other than Steve Levitt - who by that time I think was already tenured at Chicago. Which made me feel a lot better that not all graduate students were as smart as this guy.

Well, Steve was an incredibly nice and humble guy despite the fact that by that time he already was a pretty well known name. He was very gracious in giving me advice about the job market and I still remember him tiptoeing around the room in the morning so as not to wake me up.

I apologize for taking up so much space on this blog, but I thought this might be a story this crowd might get a chuckle about.

Best wishes,
Dan

#### daveslutzkin

Dr Funk: "If you assume that speed of a serve and accuracy are inversely related, it's fairly easy to show that 2nd serves should be slower. The data backs this up."

Yeah, but this assumes that speed of a serve has no effect on returnability other than accuracy. So logically you could serve very slowly but always put the ball in the corners and win every point.

The _perception_ is that it's better to slow down the second serve and not give away a free point, but if slowing it down makes it much easier for the opponent to gain control of the point, maybe it's better to hit it hard. You'll serve more double faults but when it goes it the opponent will more likely be on the defensive. The only real measure is second serve win percentage, not second serve non-fault percentage.

#### hart2412

Last week the Wall St. Journal Weekend Edition had a similar story about the growing use of statistics in basketball to exploit relative inefficiencies in player management.

I confess to being in the legion that believe in this approach, and have recently joined with 82games.com to help with a compilation of various statistics charted throughout an NBA game, with the hope of showing the power that numbers can have on team performance maximization.

Great job with this blog and keep up the good work.

#### DENVER LENDER » Blog Archive » Basketball by the Numbers

[...] I came across an article on the Freakonomics blog about Dean Oliver - a statistician for the Seattle Supersonics. He’s a statistics guy who’s trying to formulate a philosophy similar to Billy Bean regarding the stats are truly important for basketball players. The article itself appeared in wired.com [...]