I’ve done a lot of work recently on the concept of shot selection, including a breakdown of Boston University’s threes-and-layups offense against UMass-Lowell and Northeastern’s paint-focused upset of Georgetown. In this analysis, I’ve usually started with the assumption that certain types of shots are more efficient than others — in particular, that dunks, layups and three-pointers are generally good shots, and that two-point jumpers are generally bad.
But to find the magnitude of shot selection’s impact on offensive success — and to convince anyone who might still be skeptical — I’ve taken a deeper look into that assumption, using data from all Division I teams over the last two full seasons (2011-12 and 2012-13). Below is a series of charts showing the relationship between each of the three main shot types — shots at the rim (layups and dunks), two-point jumpers, and three-pointers — and offensive efficiency.
Coaches constantly say they want players to take more shots at the rim, and with good reason — teams that take a greater fraction of layups and dunks score more points. The relationship is statistically significant, as are all relationships shown here, but practical terms, it’s not enormous: A 10-percentage-point increase in shots at the rim is expected to raise offensive efficiency by about .015 points per possession (roughly a point per game). That’s nothing worth ignoring — plenty of games are decided by one point — but it might be smaller than you’d expect.
In contrast to the first category, coaches rarely say they want players jacking more threes (Joe Scott presumably excepted). But maybe they should — teams that shoot more threes also score more points, and believe it or not, the relationship is actually stronger than it is for shots at the rim. A 10-percentage-point increase in three-point attempts is expected to raise offensive efficiency by about .02 points per possession (roughly 1.5 points per game).
If you’re curious, the outlier on the far right is 2012-13 Lamar, which took only 11 percent of its shots from beyond the arc — and went 3-28 with one of the nation’s worst offenses. (Of course, Lamar shot poorly on both twos and threes, and turned the ball over very frequently, so shot selection wasn’t really its biggest problem.) This year, Pat Knight loosened the reins somewhat — the Cardinals have taken a whopping 20 percent of their shots from three-point range, more than three other D-I teams.
Since shots at the rim and three-pointers both are associated with higher efficiency, and since these three categories are collectively exhaustive, two-point jumpers must go in the opposite direction. Indeed, the wasteland between the rim and the three-point line lives up to its reputation, as teams that take lots of long twos perform worse offensively. A 10-percentage-point increase in two-point jumpers is expected to reduce offensive efficiency by about .025 points per possession (nearly two points per game).
Now, there are lots of exceptions to these trends. The team from the last two years that took the highest share of two-point jumpers, 2011-12 North Carolina (55%), actually had a good offense (1.09 PPP). The team that took the fewest, 2011-12 Hartford (16%), had a crappy offense (.89 PPP). Even in the third and strongest relationship, the r-squared was just .05 (different rates of two-point jumpers explained only 5% of the variation in offensive efficiency), leaving a lot of success that is explained by other factors. From a coach’s standpoint, improving shot selection is probably less valuable than finding the best lineup combinations — and certainly less important than recruiting good players in the first place.
But on balance, minimizing two-point jumpers is a good idea: Of all teams that took less than a quarter of their shots from midrange, 67 percent scored more than a point per possession, compared to just 38 percent of all other teams. (So far this season, Boston College, BU and Northeastern all have two-point jumper rates below 25 percent, while Harvard is close at 26 percent.)
The main takeaway, which was well-established already, is that long two-point jumpers are inefficient shots. But the other interesting finding is that, not only are three-pointers not bad shots, they may even be more valuable than shots at the rim. For all the complicated metrics that have been introduced, and all of the resulting old school-vs-new school fights, perhaps the most important insight of basketball’s statistical revolution is that 3 is greater than 2.