Basketball on Paper was published 15 years ago. ESPN’s Real Plus-Minus was introduced five years ago. The Sloan Sports Analytics Conference is over 10 years old. It is clear that basketball analytics has a firm foothold in the greater world of basketball and is here to stay. However, was it always going to be like this? Are there other possible histories where analytics did not become a part of the basketball ecosystem?
I intend to explore these questions here, as well as related branches of thought in subsequent pieces, in an effort to take a critical look at the state of the NBA today. The pieces will primarily concern the role of analytics and its influence on the game, but they will be varying degrees of technical and generally rely more on heuristics than hard numbers. Although I have a rough outline of ideas I want to cover in these pieces, part of the journey is seeing which new branches sprout up as we move through these exercises. Now on to the fun part - was analytics as it is known today always going to happen?
Let’s not bury the lede here - some iteration of analytics was always going to happen. It’s well documented that coaches such as Dean Smith used statistics to better track, measure, and evaluate team performance, which in turn influenced team strategy. There were other possible paths for analytics to have taken up to this point, but the current state strikes me as one of the most common outcomes. Other paths were possible to reach this point and the timeline was perhaps sped up a bit thanks to the overall progressive nature of the league, especially under Adam Silver, compared to other leagues, but something akin to this state was likely all along.
If we trace back the taken path, we can piece together the crucial junctions where technology, events, and individuals influenced the adoption of analytics. For example, SportVu’s tracking cameras and the data that came with it, including Second Spectrum and the products they build on top of that data. Prior to that, Synergy Sports was responsible for pushing forward both video analysis and analytics by classifying play types, aggregating player and team statistics for them, and presenting them on a per possession basis and with stats such as effective field goal percentage.
The Sloan Sports Analytics Conference has become a tentpole event and driving force of sports analytics as a whole over it’s history, with papers, talks, and presentations both advancing analytics technically and broadening it’s social capital. One of it’s founding members is none other than Daryl Morey, the famous general manager of the Houston Rockets, whose success has lent credibility to analytics both socially and within the tight-knit basketball world. The conference has become a must-attend for analysts across all sports and inspired similar conferences to sprout up across the country, providing space for ideas to spread across sports and disciplines.
Finally we have individual contributors such as Kevin Pelton and Dean Oliver. Pelton has increased public awareness of basketball analytics through his work with Basketball Prospectus and ESPN.com. Oliver is considered by many to be the first intentional user of more advanced statistical methods to analyze basketball after becoming a player-coach at Cal Tech. He helped broaden knowledge through his seminal book, Basketball on Paper, and will be transitioning back to the sideline this year for the Wizards.
These are some of the more important contributors to basketball analytics as we know it, but the path still goes back through coaches like Dean Smith. The level of analytics utilized by coaches has had a place in the game for a long time and would have continued to have a role at every level. Coaches are responsible for winning games, and winning games requires finding an optimal strategy for your roster. They have always used recorded stats - coaches come up with their own statistics all the time and often have assistants record non-traditional ones - to inspect the game. Sure, the technical revolution with computing power has allowed for innovation both within sports and other industries, but that simply sped it up. It didn’t create analytics and the associated roles with teams out of nothing any more than the advent of video technology lead to the proliferation of video scouts and coordinators.
But what if some of these things hadn’t happened? What if Yao Ming breaks down one season sooner and Morey’s Rockets falter, leading to his firing? Does Sloan still become the preeminent conference? What if Oliver didn’t get a chance to build on his basketball knowledge with Bill Bertka’s scouting company and decided to focus solely on engineering? Do we still have analytics in basketball?
In all likelihood, we do. Although the conditional probability of seeing people of Morey and Oliver’s ilk now that they’ve already existed increases, there have long been people interested in both math and sports. It was only a matter of time that such a person took a chance on seriously pursuing a career in sports over something like finance or engineering. The increases in recording and storing statistics fall in line with society-wide trends, and the increase in actual capital in the sports world to go with the always-high social capital associated with it suggests the likelihood has only increased over time. Thus, if we assume that the individual contributors would have come about anyway, what about the actual contributions?
The shape of the contributions may be different, with maybe more focus on statistical modeling methods other than regression analysis. Or perhaps on expected points, similar to the expected goals models in hockey and soccer that dominated the early iterations of analytics in those sports (to my knowledge). It would have depended on the backgrounds and interests of the contributors, but the general direction and narrative around would have been similar. In fact, the conclusions reached by analytics in the actual world likely intersect heavily with the worlds of our alternate paths because the underlying math remains the same. A three pointer is still more points than a two pointer and free throws continue to be the most efficient shot on average. The only ways these conclusions are not the same is if players’ skill level decreased significantly or the NBA changed the rules (which I’ll touch on in a future piece) in our alternate paths.
Let’s assume this argument is convincing and take the conclusion that analytics would exist in some form. It could be something akin to what it was five years ago, or maybe someone like Steve Ballmer buys a team in 2008 and invests heavily in analytics, pushing the discipline to a point similar to where it will be five years from now. Are the consequences still the same? I think yes, since analytics has and still would have illuminated pathways to innovation that were previously unseen, and discouraged the game from exploring others, regardless of the specific structure.
It may have taken a while to convince coaches of data’s utility when taken out of their hands and into those of full-time analysts, but now the two can mesh and provide a synergistic effect on each other. Chances are a Steph Curry-like shooter would have come along at some point anyway, and the right coach paired with that player would have given us similar results to the Curry/Kerr pairing (though Curry was certainly already elite prior to Kerr). But the growing acceptance of the three point shot thanks to analytics helped make Steph’s decisions acceptable by lowering the threshold on what constituted a “good” three point attempt.
Or maybe a different revolutionary player would come instead and the game would have taken a different turn. Maybe that happens five years later. Who knows. It’s all possible, but the point here is that it’s possible irrespective of analytics. What it really depends on are the players, the rules, and the coaches (I’ll explore the rules aspect in another piece). They are the driving factors, while the technological advancements in video and computation act as tools supplementing and enhancing them.
Analytics doesn’t “solve” basketball or provide some sort of silver bullet any more than the strategic machinations of coaches (aided by improvements in video). Good statistical analysis depends on context as much as, or perhaps even more than, naked eye scouting. And despite preseason statistical projections providing media fodder about not playing the game on paper, analysts want games to be played and see the season unfold as well. Games beget data and data begets analysis, which can then be used in future games to help a team win. The goal is not to limit teams or limit the experience of the game to what we think will happen on paper, but to provide a different perspective.
Analytics may have put guardrails on the way the game is played, but they are simply new versions of limits that already existed. The NBA game has always been fairly homogenous with respect to play style because it tends to be a copycat, conservative league in this regard. Coaches will use what they learned from other coaches and implement it with their own team. This mentality is what lead many coaches to play it safe and impose their own limits on how the game was played, influencing it’s evolution.
Though we are still seeing the impact that wide spread analytics unfold, especially how it relates to trends, there’s no world where it does not become a part of the game. The path to get here was winding and could have taken many different turns, but the history of analytics is not very surprising in the end. Teams and coaches want to win games and have used rudimentary analytics for these means since the beginning of the game. Now analytics are established as part of the basketball ecosystem and likely would have been under any conditions, the next pieces here will consider how analytics affect the overall ecosystem.
Comments