## Review: *Basketball on Paper*

By **Kevin Pelton**

For Hoopsworld.com

Dec. 11, 2004
*Dean Oliver's* Basketball on Paper *was published earlier this month, the first published book from the author long recognized as one of, if not the premier statistical analyst of the game of basketball. "Page 23" reviews this year's second book on NBA statistical analysis.*

At the risk of succumbing to hyperbole, *Basketball on Paper* is a revolutionary strike for statistical analysis of the game of basketball.

With all due respect for what John Hollinger has done with *Pro Basketball Prospectus*, the annual is not particularly revolutionary. There have been NBA statistical annuals before, and there will be again.

To my knowledge, there has never been anything particularly like *Basketball on Paper*. Describing it is difficult, one of the most challenging things about writing this review. I find it easier to say what the book isn't than what it is.

It is not an annual. As I discussed during **my review of ***Pro Basketball Prospectus: 2003-04 Edition*, the annual format is, in many ways, restrictive. While the modern NBA obviously gets more attention because of its relevance, it is not Oliver's main focus.

It is not a ranking of the greatest players in NBA history, or even current players. Check out Elliott Kalb's *Who's Better, Who's Best? in Basketball* if that's what you're looking for. Oliver does devote a chapter to looking at the statistics of the NBA's all-time greats and evaluating their legacies, but this too is not his focus.

So what is *Basketball on Paper*? I would describe it as Oliver explaining how to think about the game of basketball, using statistics as a tool and an aid. Though Oliver would rightly be described as the leader in NBA statistical analysis, he is not dependent on the numbers. Instead, what he does would more accurately be described as using statistics to help us confirm what we see and catch what we miss.

I happened upon a nice quote to sum up my feelings on this strategy in Bill James' *New Historical Baseball Abstract*:

"We all know many things and many different types of things which are not reflected in the statistical record. Acknowledging this, a good statistical analyst is sometimes able to reach out and draw areas of the game which were previously undocumented inside the tent, inside the focus of the statistical record. Sabermetrics is sometimes able to invent a way to correct for one or another distortion of the statistical picture."

Oliver's most notable effort in this vein is the three chapters he devotes to the topic of "teamwork", which include his effort to quantify chemistry, at least in an on-court sense. After exploring how interactions between players fail to show up in the statistics that are publicly available, Oliver explains how he works around these limitations in evaluating players in a team context. Oliver also uses information like a team's performance with a given player out of the lineup or before or after his arrival to get a broader perspective on his impact.

Also insightful, in terms of articulating a thought, if not creating it, is Oliver's point that the "holy grail" of rating systems that most statistical analysts are searching for (myself, alas, included) is not possible because at best we can only measure players' outputs -- which are affected by hundreds of other variables, particularly team context.

That Oliver does not rigidly adhere to a statistical dogma should not really be a surprise. While most statistical analysts' experience on the basketball side of things is probably limited at best -- I myself have not played organized ball since the sixth grade, victimized by a game that lacks athleticism, height, and shooting ability -- Oliver did play in college and was an assistant coach and did some scouting thereafter. He has also consulted for NBA teams, notably the Seattle SuperSonics.

As a result, Oliver's work is applicable to all sorts of people, as he outlines in his introduction. In particular, *Basketball on Paper* is written with an eye to coaches. Discussions of risky and conservative strategies and when to use timeouts should particularly appeal to these readers. Legendary former North Carolina coach Dean Smith, whose endorsement appears on the back of the book, describes it as "a unique and surprisingly practical addition to a coach's library". Who am I to argue with Dean Smith?

Odds are that you, the reader, are most interested in how *Basketball on Paper* is relevant to fans. If you're reading this column, I can only assume that you are interested in evaluating players and coaches in an analytical manner, and that is what Oliver is doing -- albeit not so much with the goal of determining who is better, as I explained earlier, but digging further to figure out how teams win. In particular, I think Oliver's chapters on how to evaluate a game by looking at a box score and how to evaluate a team over the course of the season are interesting to fans, while the discussion of the players ("Great ones, freaks, and specialists") he has chosen to evaluate makes for fascinating reading.

One concern for potential readers may be their ability to understand the math in *Basketball on Paper*. Even I was a little intimidated when I first flipped through the book and saw the number of tables and formulas, but in the vast majority of cases Oliver explains things in such a way that someone with no formal math background beyond high school can understand. There are exceptions, like the chapter that relies on significance testing (to compare a team's performance with and without a player), and many of Oliver's formulas are far more complicated than anything you'll find in this column, but understanding these concepts isn't necessary to get the bigger points Oliver is trying to make.

The aforementioned James has often made the point that he is a writer who uses statistics, not the other way around. While Oliver probably wouldn't go that far, his writing style is more than capable of keeping the reader engaged throughout. Like Hollinger and the *Baseball Prospectus* crew, Oliver frequently takes a light-hearted approach, coming up with a theme song for Dennis Rodman to the tune of the Gilligan's Island theme and titling his chapter on coaching "Should I Firebomb Billy Donovan's House?" after a question he was actually asked by a Florida Gators fan.

For the most part, *Basketball on Paper* focuses on the NBA, where Oliver has done most of his recent work. He does, however, discuss at length women's basketball and the WNBA. Besides for a couple of columns I wrote last season, Oliver's own work at his website, **Journal of Basketball Statistics**, several years ago, and some occasional columns at the women's basketball website **Full Court Press**, the WNBA analysis is essentially unique and was particularly interesting to me.

Besides for the things I've already mentioned, a highlight for me was Oliver's chapter discussing the relationship between possession usage and offensive efficiency. Using charts, Oliver explains why the line of thinking that argues that because Allen Iverson is less efficient than some of his Philadelphia teammates, he should take less possessions, is, if not wrong, then overstated. (This is true because the initial evaluation is made based on average efficiency, not marginal efficiency -- how efficient the player would be with each additional shot.) We all know intuitively that there should be an inverse relationship between possessions and efficiency (in economic terms, this is the theory of diminishing marginal returns), but Oliver does a good job of making this relationship evident and easy to understand. Further research on this topic will help us evaluate high-possession, low-efficiency scorers like Iverson more accurate.

Oliver's chapter on evaluating defense should also be a draw for many readers, introducing new defensive stats he had a team of volunteers track in the WNBA in 2002. Adding these stats - forced misses, forced turnovers, defensed field goals made, defensed free throws made, and missed free throws defensed - to the ones already tracked - blocks and steals, basically - provides a much better picture of defense, as Oliver illustrates.

For me, however, this chapter wasn't particularly illuminating because, well, it's my defensive scoresheet from a Seattle Storm-Los Angeles Sparks game that Oliver uses in the book. I should take this opportunity to point out that I am an extremely biased reviewer of *Basketball on Paper*. In addition to the work on Project Defensive Scoresheet, a study of mine is briefly referenced in the book, and Oliver and I have worked together on some of my statistical projects. As a result, my review should be taken with a grain of salt.

I still think I can be objective about Oliver's work, though I'm certainly not objective about statistical analysis in general. As far as that topic goes, I don't think it gets any better than *Basketball on Paper*. Pete Palmer, along with James another leader of the statistical revolution in baseball, declares in his endorsement of the book that "it can be used as a textbook". I would take the same position. *Basketball on Paper* is not a book to be read once and put away forever. Instead, it can be placed on the shelf and pulled out whenever statistical insight is needed.

There has never been a basketball book quite like that I'm aware of, and that is the highest praise I can offer *Basketball on Paper*.

*Kevin Pelton is an intern for the Seattle SuperSonics and is responsible for original content on ***Supersonics.com**. He writes "Page 23" for Hoopsworld.com on a semi-regular basis. He can be reached via e-mail at **kpelton@sonicscentral.com**.