Saturday, July 09, 2005

Some ideas about the use of capologists and stats experts in NBA front offices

By Dan T. Rosenbaum

I would like to start out with a big disclaimer. I have never worked for an NBA team in any way whatsoever, so you should take what I say with a big grain of salt. But I have talked to a lot of people in a lot of NBA front offices, and I think I might have some useful ideas (and surely some bad ideas) about how to best utilize capologists and stats experts. With teams paying skyrocketing costs for coaches, I figured that some brainstorming about other front office personnel might be interesting.

How teams use capologists

Most teams have a capologist and a stats expert (usually separate people). The capologist often is a lawyer or accountant outside of the basketball operations department - sometimes even outside the organization. Their capologist duties often are only a small part of their overall responsibilities, so they concentrate on learning the complex rules governing trades, free agency, and the luxury tax. This leaves precious little time to think strategically about how to use these complex rules to benefit their team through trades, structuring free agent deals, etc. Also, putting together forecasts of future salary caps, luxury taxes, and league-wide free agent activity is not generally done in any systematic manner. And perhaps most importantly, by not being a "basketball person," a capologist often is seen as a resource for asking questions and not as an integral decision-maker.

But there is a tremendous amount of heterogeneity across teams, so it is a big mistake to overgeneralize. The lawyer/accountant model of a capologist described above is not the only model. Other teams invest salary cap expertise into their basketball operations people. The Chicago Bulls and San Antonio Spurs are two of the most successful models for these two different models of capologists.

Take, for example, a trade made in October 2002 that was way under the radar screen. San Antonio traded Erick Barkley to Chicago for cash considerations and Chicago immediately waived Barkley. What was the purpose of this trade?

San Antonio probably forecast that their team salary would be above the luxury tax threshold (and below the cliff threshold). So Barkley's salary would have cost the Spurs $765,600 in salary, $765,600 in luxury taxes, and about $2.4 million in lost luxury and escrow tax distributions. Even if they paid Barkley's full salary and included a small bribe in their deal with the Bulls, they could have saved more than $3 million in a move that did not affect their basketball team one iota. Three million dollars likely would pay the salary and overhead for a capologist for more than a decade.

And it is no coincidence that San Antonio and Chicago pulled off this deal. Chicago has lawyer Irwin Mandel who is reported to be one of the best salary cap experts in the league. Mandel is not a basketball person but he has been with the Bulls forever, and almost plays the role a President would play on other teams. When general manager John Paxson meets with agents, it is not uncommon for Mandel to be with him during negotiations.

San Antonio has Sam Presti, whom general manager R.C. Buford has called "one of the talented young minds in the game." Presti is a rarity in the business being a capologist and stats expert. Also, general manager R.C. Buford reportedly is one of the best salary cap experts in his own right.

While Chicago represents perhaps the best application of the non-basketball person model of capologist, San Antonio employs what I believe is a better model. They have invested salary cap expertise into their key basketball people, in essence creating a culture that is probing for arbitrage opportunities based upon the salary cap. And that, I believe, is one of the reasons why San Antonio is one of the best-run franchises in the NBA. Because their key basketball people have a solid understanding of the nuances of the salary cap, they see opportunities that are missed by teams that departmentalize basketball and salary cap expertise.

Another example is the trade that brough them Hedo Turkoglu and Ron Mercer two years ago. They served as broker for a trade between Indiana and Sacramento, and in return got a couple serviceable players that did not cost them any salary cap space in the next summer when they were looking to sign Brent Barry and re-sign Manu Ginobili. Such a move may not seem like much, but little moves like than can be what separates the championship teams from everyone else.

How teams use stats experts (and my philosophy on basketball stats experts)

The stats expert often is a young person who may have worked his way up in the organization, often starting out working in the video department. Typically he has very little formal statistical training, but learns about basketball stats through reading and talking with other experts.

Another common approach is to hire a statistician (or two) with some advanced training, either in-house or as an outside consultant. Often these statisticians may technically be quite proficient; they may produce advanced statistics with dizzying levels of math. But do they have the basketball knowledge needed to ground their analyses? Sometimes, but not always.

And therein lies the problem. We all use statistics. Even the oldest of the "old school" basketball people look at box scores and per game statistics. Stats experts have value not because they are the only ones looking at stats. Their value comes from having good judgment about stats - good judgment that comes from a solid understanding of both basketball and statistical analysis more generally.

But the bar for this good judgment is set very high. Basketball is not like baseball, where the repeated one-on-one contests between batters and pitchers and relative lack of interaction between teammates generates loads of data that can be usefully analyzed with the statistical tools taught in a good undergraduate statistics class.

In basketball, the centrality of teammwork and interactions between players requires that we need to account for a player's roles, as well as the players he plays with and against. Doing so requires a lot more creativity and more sophisticated tools than what are used by baseball analysts. An Ivy League graduate who is handy with Excel is not going to be as useful in basketball as they have been in the Moneyball world of baseball. It is not enough to be smart with numbers.

That is why I think the folks who end up doing stats in a serious way for basketball teams are going to have more formal training (or at least more extensive experience working with basketball stats) than the folks in baseball. The problem is harder in basketball, so the analysts are going to need a bigger toolbox they can work with.

(As I see with the graduate students I teach, the demand for skills working with complex data are growing exponentially in the job market.)

But a good practical understanding of statistical analysis is not enough. The centrality of teamwork and role players significantly increases the basketball knowledge a stats expert must have in order to make good judgments. This again makes basketball very different from baseball, where a lot can be learned sitting in an office and churning out numbers. Not so in basketball, where context matters.

So the teams that get this right are going to need two things - a person with solid training in (or extensive practical experience with) statistics/econometrics and a solid understanding of basketball.

The gold standard of stats experts - Dean Oliver - is no accident. Oliver who worked for the Sonics this past season is a rarity among stats experts. He is Ph.D. (in environmental engineering, I think) who has worked as a scout. He combines formal training, a solid understanding of basketball, and lots of experience (he has been at this for more than a decade).

Oliver, I believe, played a subtle but significant role in the Sonics' remarkable turnaround this season. Conventional wisdom gives McMillan almost all of the credit, but Mike Kahn of in a piece that is highly supportive of McMillan writes that McMillan "contemplated resigning," after management said "they expected to be in the 2005 playoffs." Kahn goes through a litany of reasons why McMillan supposedly was unhappy with this team, so one has wonder how in this atmosphere McMillan transformed from a mediocre coach to a great one.

Oliver certainly was not responsible for the distinct style the Sonics played with this season - a style that did a masterful job of exploiting the talents of its very good role players. But I think his numbers helped the whole staff have more confidence in this unorthodox approach. Oliver, like a lot of the players on the Sonics, was by no means a superstar with all of the answers; this was no Moneyball repeat. But he played a role and played it well. And if a stats expert can do that, he is earning his way.

That ideally is how a team uses a stats expert. Not to run a team as in the Moneyball world of baseball. But as a role player who works hard to become a good basketball person and makes all of the basketball people around him better at what they do. As a role player who can help make sense of advanced methodologies and concepts in basketball statistics. More importantly, as a role player who can help put data in the right perspective and understand when to pay attention to the data and when to ignore it.

This final point is greatly underappreciated. If there is anything my experience teaching and working with stats has taught me is that non-stats experts spend ungodly amounts of time looking at stats that are more misleading than useful. (Although highly trained "experts" can be worse if they don't understand basketball very well. Or if they just do not have good intuition for data.) I just cringe when I hear of teams using plus/minus data in ways that are almost surely more distracting than helpful.

At the end of the day, using stats wisely is about making good judgments. And it seems quite a risk when teams rely so heavily on stats "experts" with very little statistical training (or experience) and/or limited basketball knowledge.

[I will add more to this later in Part II.]

Last updated: 11:00 PM, July 14, 2005


Anonymous GrandAdmiralDan said...

Wait a minute, people get PAID for doing what I do for free :)


I too have been an advocate of NBA teams bringing in a capologist or two who meet the exact criteria you describe.

I believe that the Milwaukee Bucks GM, Larry Harris would share your views on this subject, since he is both the type of capologist you describe and the type of stats expert you describe. Perhaps not quite as gifted as the examples you give in either field individually, but he was able to use these skills to work his way up the organization all the way up to GM. If I am not mistaken, he started with the Bucks in the video department while his father was here (Del Harris), and decided to keep that job and stay with the Bucks when his father moved on.

Good post.

7/10/2005 10:06 AM  

Post a Comment

<< Home