It is pretty clear to anyone that watches basketball, I think, that there are different offensive styles adopted by players who nominally play the same position. Among guards and wings the two most common archetypes are generally called something like Slashers, who initiate the offense driving to the basket and Stretch or Spacing guards who shoot from beyond the arc and stretch the defense.
In fact, the very first analytic post I did on the web last year was looking at shot selection among back court players using play by play data via HoopData, which has since stopped publishing. I decided to revisit the analysis this time with data from NBA.com with data from 2010 to 2014, again using a technique called k-means clustering to group players based on where they took their shots.
Based on an initial analysis, I found the variation in shooting locations between players was, indeed, primarily between shots from three and shots at the rim, with a secondary source of variation from mid-range shots. Shots in the paint outside of the restricted area were associated with trying to get to the rim, and corner three were a minor specialty location, but not enough to warrant its own group.
Any attempt to categorize data has issues with borderline cases, whether its the five cardinal positions with tweeners or shooting location styles. But, this configuration gave me good measurements on both cluster cohesion and separation between clusters, and, more importantly, conceptually coherent groups to examine.
So that left me with three groups, that I dubbed Slashers, Spacers, and Mid Rangers. Over the four year study period 41.6% were grouped as Spacers, 30.8% as Slashers and 27.4% as Mid-Rangers. Below you can see the averages in shot attempt percentages by the relevant locations:
So, Spacers shot from three, on average for 50% of their attempts. While Slashers were an almost inverse image, taking over fifty percent of their shot in the restricted area or the paint. Mid-Rangers, on the other hand, look almost indiscriminate in their shot selection shooting slightly more from mid-range than anywhere else.
This visualization of Wings and Guards in 2014 gives a decent picture of how the different groups operate with a few names attached (click on the image to see a live version of the visualization).
We can take the analysis a step further by getting the stats averages for each cluster. The different playing styles become pretty apparent, with higher average shooting efficiency by the Spacers, only partially offset by the increased free throw rate from Slashers (and Mid-Rangers coming in the bottom). On the other hand, Slashers tend to assist more and grab more offensive rebounds. The numbers for 2014, which are typical of the entire study period, are shown below.
Player offensive rating isn’t necessarily the best individual advanced metric for players, but consistently higher rating by Spacers is telling us something. For example, it appears that many Slashers play that way partly out of necessity given their lack of long range shooting ability:
That said, there are great and not so great players in each shooting group, shot location doesn’t in and of itself determine a player’s offensive value. You can see that just looking at the list of some of the best player’s in arguably the least advantageous group, the Mid-Rangers, including Chris Paul and Kevin Durant. Below are the players with over 1,100 shots last year:
One thing I intend to follow up on, is any different effects of offensive styles between positions. My preliminary numbers indicate that shot selection groups are most important for shooting guards and least important for point guards, with small forwards in between. In other words, point guards in the different groups may be able to contribute to offensive production in other ways, namely distribution, while shooting guards may be less able to offset their lower shooting efficiency.