An Analytical QnA with Seth Partnow
If you would like to read this article in Greek, you can click here
A few weeks ago, a book entitled "The Midrange Theory" was published and made available around the world. The book discusses the practice and impact of basketball analytics, not in terms of pure numbers, but a creative narrative which can help us understand the game of basketball better, but also in a new, different way of thinking.
This book is authored by analytics expert Seth Partnow (known in the Twitter world as Anchorage Man). Mr. Partnow currently works as an Αnalyst at The Athletic, providing NBA and basketball Analytics in his articles. He previously served as Director of Basketball Research for the Milwaukee Bucks, from the 2016 off-season through the end of the 2019 season. Prior to that, he was the lead writer and editor of The Nylon Calculus, a leading basketball analysis website.
It is my great honor and pleasure to have Seth as a guest in my Substack, and I have the opportunity to give you a written Q'n'A on his book and his thoughts on Analytics in general. (The photos with the diagrams are from his book, and Mr. Partnow has given me permission to use them).
Let us plunge into the conversation:
-Seth, you called your book "The Midrange Theory." Usually a title says a lot about the content. So what is the Midrange Theory and what should we expect to find in its pages?
First of all, the title, “The Midrange Theory” is an homage to one of the favourite hip hop albums of my youth, A Tribe Called Quest’s “The Low End Theory.” As might be expected, the title chapter discusses the rise of the three-point shot and commensurate decline in midrange attempts by both describing in more detail what has actually occurred – I don’t want to spoil the book, but there are specific types of midrangers which have vanished while others have barely changed in frequency – while also adding in the changing competitive context in terms of rule changes, strategic advances and improving/diversifying player skillsets which have contributed to the differences between today’s game and that of yesteryear as much or more so than the application of data.
But that’s just one chapter. I cover a number of other topics, including the draft, how and why the playoffs are different from the regular season, challenges in interpreting statistical production as well as detailing the actual practice of analytics within NBA franchises. I’ve also included a “basketball analytics 101” primer as an appendix to the book.
-A month Ago Charles Barkley said: “First of all, they (analytics) are just stats. They just gave it a name. It's kind of like yoga. Yoga's nothing but stretching. They just call it yoga so they can charge more. I tell people, yoga's just stretching. They gave it a different name to charge you for it. The statistics, they change the name — 'We're gonna charge you for analytics now' — and they just raised the price, but it's nothing but stats.”. What do you think of this statement, and of people distrusting Analytics in general?
One of the driving forces behind the book is to push back against “analytics” being defined by the people who know the least about it. As far as people distrusting the discipline in general, my other aim with the book is to have that better definition be in the language of basketball rather than in the language of statistics or math or computer science. I think an understanding of just how much statistical analysis agrees with and reinforces core pieces of basketball conventional wisdom can move us from a point of antipathy and antagonism to a healthier discussion coming out of those points of agreement.
For example, the Four Factors are simply a restatement and explicit definition of truly fundamental basketball wisdom: take and make good shots; take care of the ball; avoid fouling on defense; finish defensive possessions with a rebound. That’s genuinely elemental stuff and explained in that way makes the adoption of eFG%/TOV%/FTAr/DREB% a much smaller leap than is often portrayed between “traditionalists” and “analysts.”
-In your book you have a lot of different diagrams with statistics. Do you think visualisation is a way to help people understand analytics and the game in general, and also an easy way to convince people about analytics? And if this is the case, how can we convince them?
Communication is as much about knowing how the recipient best takes on information. For many, well-visualized data better illustrates the context and important takeaways from statistical analysis far more convincingly and succinctly than would a data table or a lengthy verbal description. It doesn’t always work, and might not be the best way to communicate with any given individual, but it’s an essential tool to have in the chest if we’re trying to reach as many people as possible.
-How do you divide your time between reading Stats and watching basketball games? (NBA or NCAA as an example of scouting)?
I probably watch 3-4 hours of basketball a day, while spending perhaps a similar amount of time “doing stats.” Though often I’m doing both simultaneously, as a lot of the analysis I end up doing is spurred by something which has caught my interest in a game I’m watching.
-In Chapter 4 of your book, you write about "Goodhart's Law": Goodhart's Law is simply expressed as, "When a measure becomes a goal, it ceases to be a good measure." You cite Russell Westbrook as an example of this law in your book. In my opinion, Westbrook is a stat-padder player and overrated, but to a large segment of basketball fans, he is a bona fide superstar. How do you think we can get people to use stats outside of the boxscore and also point out some measurements like efficiency?
Very delicately? But seriously, pretty much everyone accepts that there are players who are ‘better’ than their box score stats. For that to be the case, it literally must also be true that there are some players who are worse than their own statistical record. Of course good look convincing a partisan of a particular team or player that their favorites are on the wrong side of that dichotomy!
To the larger question, I try to refocus the discussion on what truly matters, at least in terms of basketball as a competition. A player is only as good as the degree to which they help the scoreboard move in a positive direction for their team. To the extent statistical accumulation serves that purpose it’s great, but sometimes players can put up gaudy numbers of their own without necessarily increasing the team’s total production, and part of the purpose of analytics is to identify which players are doing this and which are actually helping.
To circle back to Westbrook, the argument that his triple doubles overstated his impact isn’t the same thing as saying he was not an impactful player. Rather it’s a discussion of degree of impact. While there are certain player types that analysts tend to think are just not good at all (think volume-shooting scorers with mediocre efficiency who don’t contribute much in other areas) the more disagreements usually concern players who are ‘overrated’ but still very good. If I opine that the player you think is an MVP candidate is only about the 20th best player in the league, that means I still think he’s an All-Star!
-As a Greek, I have to ask you a question about Giannis, since you worked for the Milwaukee Bucks for 3 years (2016-2019). Do you think the development of his game is due to a player seeing the stats and saying I need to get better, or is Giannis an example of a person with great work ethic and hard work?
Giannis has a truly maniacal (perhaps even at times detrimental) work ethic, so that part of it was always a given. Him being put in the right situation to both highlight his capabilities but also expand the decision-making portions of his game was what took him from being an All-Star to being an MVP.
-In chapter 9 of your book, you analyze how difficult it is to measure the defensive impact of a player and a team in general, since defense is the interaction of 5 players at once. How do you think people should look at defensive stats, and is defense a type of game where the eye test is better than analytics, as opposed to offense where we have a lot of better metrics?
Defense is an area where there is still a lot of foundational work to be done simply to describe what players are doing before we can evaluate how well they are doing those things and/or how impactful they have been. We haven’t quite cracked the code of coming up with something like “usage vs. efficiency” for defense, but I think this is an area where continued work with player tracking data can start to shed some real light not just upon who is good or bad defensively, but also why they are and identify some of the things which are happening on the floor which are particularly helpful or harmful to a team’s defense.
-In Chapter 13 of your book, talking about the NBA draft, you said this: “Playing basketball well is not only about having the right tools in the chest, but about selecting the right one, quickly, and using it properly.” I think that's a great quote about basketball IQ. I think basketball IQ is more important than a trained body, like Nicola Jokic's. We do not have specific stat about basketball IQ, but how do data analysts advise teams in draft selections to take players who understand the game better?
The simplest way I can express this is that it’s always important to ask “but can he play?” For a player with evident physical abilities and attributes but mediocre statistical accomplishments, the dearth of production is at least a strong suggestion that no, he cannot play if we define “playing ability” to be that intersection of being able to identify what the correct action is and then to execute that action successfully at game speed.
-Aside from being a data analyst ,Seth, you are a basketball fan. Is it easy to overcome your bias when the stats on a particular subject say something different than your opinion?
Fighting against one’s own biases is always a challenge, even when you know you have biases to overcome. So, not it’s not easy. But I try!
-It's a fact that three point shots and shots at the rim predominate these days. Do you think the league will return to the era of the 90s, with lots of mid-range shots, or will we see a greater increase in three-point shots in the future?
I think we’re approaching an upper bound to the number of threes taken, but I don’t think we’ll see much of a move back the other way. As I discuss in the title chapter, there are lots of good reasons why the bulk of threes attempts are threes and not long twos, but you’ll have to read to get the full explanation.
If this Q'n'A captures your imagination about the contents of the book, you can purchase Mr. Seth's book from both Amazon and the official book publisher. Here are the links
Triumph Books
You can also follow him on Twitter to easily find the interesting content he uploads !!!!
https://twitter.com/SethPartnow
If you would like to read this article in Greek, you can click here
Really enjoyable! Looking forward to reading the book!