A Pandora’s Box of Score Keeping
Issue   |   Thu, 11/08/2012 - 20:13

By the time the NBA season rolls around every October, I’ve had way too much time to tout all the potential it represents. Fortunately, winning basketball games is a moving target, so there are surprises when rubber finally does meet road for NBA teams. For example, high-profile teams that sputter out of the gate (hello, Lakers!) are derided as being talented only on paper — which is to say that the expectations fueled by previous performance fail to materialize in the new season. So how do we make sense of that gap between performance and expectations? The records of performance, which feed expectations, are found in box scores.

Of particular interest is the disparity between exceptional performance and successful results — as is the case with super-teams that can’t figure things out. How it is that a team that stuffs the stat sheet could ever lose? It could be that performance in basketball is valued incorrectly. That is, it’s possible that in boxing in the stats we care about, we disregard those that actually matter for winning. So, do consistent discrepancies between performance and expectations belie the illusory nature of statistical categories? They don’t record everything that happens in a basketball game, but box scores have enough information for us to piece together the parts of the story of the game we care about. Our question, then, is whether box scores allow us to reconstruct the most important elements of a basketball game.

In the on-paper world of box scores, it’s pretty easy to win a basketball game: just score more points than the other team. But past the points column lie answers to the questions about the difference between garnering great performances and garnering successful ones that amount to wins. Field goals are a huge piece of that puzzle of course, because putting the ball in the hoop is the only way to get ahead. Similarly, the free-throws column is a good place to keep tabs on how well each team did in picking up the easiest points in the game. It’s tempting to look at those categories and link shooting percentages to winning, but let’s hold off on that just for a moment.

Instead, let’s dig a little deeper into the field goal and free-throw numbers. The number of shots made in each of those categories is clearly important, but what about that second number — the number of shots attempted? Basketball is an unpredictable game, but we can simplify things in our on-paper world. Forget everything you know about basketball for a moment, and focus on the scoring objective of the game. Every time a team has the ball, they should be looking to score, right?

In this way, basketball boils down to a simple game where either one team will end up shooting from somewhere on the court, or the other team will gain control of the ball before that happens and reverse the roles. Every single basketball play ends with either a shot or a turnover.

Go ahead and think about it — the shots may vary in nature and difficulty, but in the NBA, you’re guaranteed either a shot, or a change of possession every 24 seconds or fewer.

So because basketball teams win games by scoring points, and scoring is constrained first and foremost by the number of possessions each team has, successful teams should seek to win by increasing possessions. This maximization of scoring chances can be achieved in two ways: limiting your team’s turnovers while increasing the other team’s; and otherwise pushing up the number of possessions you have.

Fans of the Phoenix Suns will remember how well this two-pronged strategy works: the run-n-gun Suns of the early 2000s were famous for taking shots in seven seconds or fewer every trip down the floor. At that quickened pace, the Suns developed a possessions cushion over the rest of the NBA — all the extra possessions added up to more chances to win, and more room for error.

And that’s precisely what advanced metrics like points-per-possession (PPP) measure. For every possession — one of a limited number of opportunities each team has to do something with the ball — used by a player, how many points should a team expect? In broad terms, how much closer does each possession get the team to winning?

With that in mind, we can examine shooting percentages. Shooting efficiency is key, because it harks back to our key insight about possessions: scoring more with fewer possessions is a valuable skill indeed. Nevertheless, it is less important than it first appears. One of my favorite NBA quotes comes from Antoine Walker, who when asked why he jacked up so many three-pointers, responded, “Because there are no fours.”

To illustrate the point, consider this scenario: two teams have the same number of possessions in a game. One team shoots only two-point field goals, and the other shoots three-pointers exclusively. Even if the former team had a magical 50 percent shooting rate, the latter would only need a shooting percentage above 33 percent to win the game. Scoring is the most valuable skill in basketball, even at the cost of voluminous shooting. Other stats like rebounds, assists, and steals can also be assigned values in this system for both creating new possessions and generating quality shots with high PPP values.

Within this framework, our first conclusion is a blaringly obvious one: scoring is a function of scoring chances. Our second is more nuanced: that in any given game, a team doesn’t need to be absolutely efficient to win; a team simply needs to be relatively efficient with its scoring opportunities compared to its opponent. Assists, fouls, turnovers are all categories that reference the way in which a team ended a possession. In piecing together these seemingly disparate bits of information, it is possible to understand how a team spent its possessions, and what that would have meant for their success in the game. From there, we break into a new dimension of analysis: rather than analyzing performance as it is, we can derive optimal performance strategies for success.

For example, given the number of possessions a team uses per game, and the team’s shooting percentages, it is possible to provide an ideal distribution of two-pointers, three-pointers and free-throws. It remains to be seen how well these kinds of strategies could be implemented, but the potential is enormous.

Box scores provide a multitude of performance data, some of it more convoluted than others. Highlight-reel dunks are lumped together with off-hand prayers and uncontested mid-range jumpers in the field-goal category. First-quarter points have the same value as those earned in the pressure-packed fourth quarter on a score sheet. And yet there is solace in that dearth of information — basketball players, after all, don’t have quite that much on their minds.