clock menu more-arrow no yes

Filed under:

Basketball by (and beyond) the numbers

New, comments

While numbers never tell the whole story, there is a story they tell. And that story can help us understand the team -- and the game -- we love. For that reason, I'm happy to announce that Seth Partnow will be joining PtR as Statistics Editor. He will helping us in the way we use numbers in the stories published at Pounding the Rock. Here's his first post to introduce himself to the site.

Andy Lyons

First of all, cheers to J.R. Wilco for asking me to come aboard as PtR's first Statistics Editor. I wanted to take this opportunity to introduce myself and what that role means.

I'm a long-time NBA fan. I have academic background in economics and a law degree. I played a little bit (more accurately: I was on the team) for a small college, and I write about basketball, often within an analytics-based mindset numerous places around the internet, most notably at Nylon Calculus. I've come up with a few metrics of my own, largely based on the data made public from the SportVU player tracking system. One is an attempt to value the "rim protection" ability of NBA bigs. The other I've named "True Usage" and is a measure of offense as a collaboration between teammates rather than examining only the contribution of the shooter.

I should fess up right now: the Spurs are not my favorite team. That said, I had two Spurs in my pre-season Top 25. But if I had to declare allegiance to just one team, rather than to the game itself, it would be ... a team to be named later.

So, why am I here, and why should you care? Well, let's begin with a talk about basketball by the numbers.

Statistics can be a wonderful tool to increase the understanding and enjoyment of basketball. For someone like me who approaches the game from a problem-solving standpoint, statistics help frame and refine what my eyes are telling me about what is happening on the floor. It is vital that the two things, "the numbers" and "the eye test" go together. Without the eye test, we're all just staring at spreadsheets -- and we probably get enough of that in our everyday lives and jobs already.

One of the things that keeps the game on the floor from coexisting with the game in Excel, is the use of stats as a bludgeon. "This number says my guy is the best and your guy sucks, and if you disagree with me, I'm a better, smarter basketball fan than you."

That's an extreme version of something that anyone who engages with the NBA on the internet encounters at one time or another, and if there's one thing I hope to bring to PtR its an approach that helps us avoid that sort of argumentative pose.

Stats can tell us something about who's better, and ranking players is always fun. But context is vital. Almost every metric or measure of performance of NBA players illustrates what a player DID, not what he IS. Matt Bonner is a very effective player in short spurts because Pop only uses him in situations where he can succeed. Meanwhile Rudy Gay became a pariah in Toronto because of his poor performance, when really he was being asked to do too much. With his usage north of 30%, he became a shot-chucking liability. Returned to a more sensible role in Sacramento, his numbers came back to something like his Memphis norms.

Players get hurt, play out of position, are experimented with, have their minutes yo-yo'd, fight with the coach, have contract years. All of these things can affect players' performance and cause guys to perform at different levels from one year to the next. Well, aside from Tim Duncan, that is.

Stats are best used with a certain degree of humility. No matter how comprehensive any "one-number to rule them all metric" claims to be, it is laden with built-in assumptions, valuations and abstractions that render the result much more a best guess than a definitive answer. There are enough different systems of evaluation to allow one to "find a number" to support nearly any position you want to take.

Which brings me to my own philosophy on basketball analytics. To quote my friend Ian Levy, "The value of a statistic is always derived from how well it matches the question you are asking." The questions we are best equipped to answer are the bite-sized ones. The question, "Who is a better shooter, Tony Parker and Russell Westbrook?" is far easier to tackle than simply "Who's better?" You can certainly use many different stats to argue for one or the other, but it is vitally important to explain, if only to yourself, why the numbers you've chosen are important ones.

My goal as the Stats Editor here is to help ensure that the numbers are used in a way that encourages discussion and increases understanding of the game -- not to shut down disagreement and assert intellectual superiority. I'm always available on twitter @SethPartnow to answer stats questions you might have. One last thing, while I'll often have AN answer to a question, there are very few instances when I will claim to have THE answer.

I hope I can help enhance everyone's knowledge and enjoyment of the game during the coming season.