Pretty much since basketball’s inception, players have been categorized into five traditional positions: point guard, shooting guard, small forward, power forward, and center. As the game has evolved and small ball lineups have become more prevalent, the line between these positions has become more opaque and new positional categories emerged, like primary ball-handlers, wings, and the all-inclusive big men.
Many different analyses have been conducted to develop new ways to categorize players. @SethPartnow utilized radar charts to visualize different point guard archetypes, see here. Most recently, serving as the inspiration for this article, @CrumpledJumper used play types to categorize different types of players, see here. I developed an idea of using players’ shooting location as a way to categorize players with similar shooting distributions. Specifically, the percentage of shots a player took from the paint, floater-range, mid-range, and 3-point land. This analysis looks at where a player shot from, rather than how effective he was at shooting to come up with these player groupings.
NBA.com stats were used to pull total shots taken from each range for each player. I then was able to calculate an attempt rate percentage based on these totals. In order to keep players with a reasonable shot sample size, I only included players who took at least 200 shots in the regular season.
To do this categorization I used a common machine learning technique known as a hierarchical clustering in Python. Hierarchical clustering allows us to group data into clusters (for this analysis we can think of these clusters as player types) based on two or more variables. In this case we have four variables: 1. attempt rate from paint, 2. attempt rate from floater range, 3. attempt rate from mid-range, and 4. attempt rate from 3. Hierarchical clustering proves useful when you don’t know how many groups you want to end up with. It allows us to go level by level and start with more general groups and then gradually get more specific.
Let’s now look at the first split. As we can see in the graph below the first split is split upon players who live in the paint and players that don’t.
If we now go down a level and look at the next split, the 3-point shooting group breaks into two more specific groups: 3-point specialists and a more well-balanced group that shoots from everywhere, see below.
So as of now there are three groups: a very paint heavy group, 3-point specialist group, and a more well-balanced group.
We could keep going and split the data into more and more clusters, but I think six clusters helps to paint a nice picture. This level of grouping is descriptive enough while not being too granular, and can really allow us to see some interesting player types and trends.
The final six groups are:
Rim Runners — These are guys who live in the paint (70 percent paint attempt rate), typically rolling to the rim as lob threats or catching dump of passes. Guys like Rudy Gobert and Clint Capella are notable examples.
Finesse Bigs — This group is a bit different than the Rim Runners, where they utilize more skill and deft to find their shots. Whether that’s operating with their back to the basket or facing up and driving. These guys aren’t always getting right into the basket, and are frequently operating in the floater range where 26 percent of their shots are coming from (the highest for any group). Ben Simmons, Jonas Valanciunas, and Giannis Antetokounmpo all fall under this group.
Inside-Out — A bit of a twist on the Moreyball Men where the largest amount of their shots are coming from in the paint (37%) and then 3-point line (31%). This is an interesting group because it contains bigs, wings, and point guards. Some examples are LeBron James and Russell Westbrook.
Mid-range Aficionados — A throw-back player type where the largest quantity of their shots come from the midrange at 32 percent. Some examples include, Kris Middleton and Myles Turner.
Moreyball Men — As the name suggests, players that fall under this group come from Darryl Morey’s school of thought chucking up a lot of three pointers (44 percent 3-point attempt rate) and when there not shooting from the three they’re in the paint (24 percent paint attempt rate). Some examples include the heir apparent James Harden and Josh Hart.
3-Point Specialists — These guys live behind the three-point line with a whopping 64 percent of their shots coming from beyond the arc. Notable guys include Steph Curry, Kyle Korver, and Eric Gordon.
Now that we have identified these six player types, are any of these groups more or less efficient than the others? We can utilize average points-per-shot by group to dig deeper into this, see below.
Rim Runners are by far the most efficient player type, once again showing that even though we are living in the 3-point age, paint attempts are still the most valuable shot attempt in the NBA. 3-Point Specialists are next in line behind the Rim Runners, while Mid-range Aficionados are unsurprisingly the least efficient player type.
I’ve also created a visual that shows the most and least efficient players in each player type as well as overall rank on PPS throughout all groups. You can search and filter on specific players to see where they fall.
So, we are now beginning to see this trend that the more extreme player types are the most efficient (guys who only shoot in the paint or only shoot 3s). At a common sense level this makes sense, if you specialize in one area you can get really good at it. Is it possible to be a number one option on your team in one of these shooting profiles? To see if there is a relationship between offensive workload and player type, we can bring in usage percentage to see the distribution of players on this spectrum. Usage percentage measures the amount of possessions a player uses, including shots, free throws and turnovers.
As exemplified by the two visuals above it’s very hard to be high usage when you are only shooting 3s or rolling to the basket. The number of players who are high usage is much larger in the Moreyball Men, Inside-Out, and Mid-range groups. By looking at the average usage percentage by cluster group visualization we can see the average usage rate for Rim Runners and 3-Point Specialists is a very small, both at approximately 17.5 percent. This demonstrates the nature of these player types. Star players just don’t have the luxury of only taking one specific type of shot, they are required to have a diverse skill set and typically shoot at all three levels: paint, midrange, and at the 3. They often have to take the tough shots that other players on the team can’t create for themselves, upping their usage and diversifying their shot portfolio.
I’ve put together a subjective list of some of the top offensive players in the league to show which different groups they fall in. You can add or remove other players to the visualization as you wish using the filter.
We can see many of them fall in the Moreyball Men category, a testament to where the league is going with threes and paint shots. No Rim Runners break the list and there is only one 3-Point Specialist. Low and behold it’s Steph Curry, but as we know Steph isn’t the rule he’s the exception to the rule.
I’ve also created a visual showing team breakdown of different player types, see below.
The percentage in each box shows the percentage of players on a specific team that belong to that player type. You can check out your favorite team and click into the box to see which players fall into which bucket. I had a fun time trying to guess before I looked.
Houston only has Rim Runners, 3-point specialists, and Moreyball Men. Even a team like Brooklyn, who wasn’t a particularly good offensive team last year has a nice concentration of 3-Point Specialists and Moreyball Men. They play a modern style and are taking the right types of shots.
One of this year’s surprise teams, Indiana, shockingly sported a good offense this year. A lot has been written about their love for mid-range shots. They have zero players who were 3-Point Specialists or Rim Runners, the two most efficient player types. And only one guy, Bojan, who fell into the Moreyball group. Can they maintain this hot mid-range shooting next season, or will they fall down back down to earth?
This does go to show that if you have players who are efficient at taking inefficient shots, that can be okay. It is quite possible to have a good offense when not shooting a lot of threes and paint shots, just look at the Timberwolves. People were worried last year (as was I) about their lack of shooting, but all in all, talent tends to win out.