The defensive impact of individual NBA players is hard to measure, as it's so intertwined with their teammates' performance. Austin Clemens, undeterred, has taken an ambitious shot with his new Adjusted Impact by Court Location model, which attempts to parse out individual impacts on opponent shot selection (the square size, in the chart above) and field goal percentage (the square color).
You can look up the 2013-14 output of over 300 players on Clemen's site, but the chart above is for Roy Hibbert. You can see how Hibbert's particular defensive presence bears out on shot attempts: He appears to free up perimeter defenders to chase guys off the three-point line (the models shows eight percent fewer threes when Hibbert was on the court) because he's waiting behind them to threaten shots at the rim (which the model reports opponents also avoided, taking three percent fewer shots). That leaves them to shoot from midrange, where opponents were plus 11 percent on attempts. Hibbert was a standout but not an anomaly; Clemens found that, in general, interior defenders make a much larger impact on opponent shot location than wing defenders.
The methodology here is a bit tricky, but it's related to Adjusted Plus/Minus. For every shot taken, we know the location, the shooter's average field goal percentage from that location, and who's on the floor. (Someday soon we'll be able to analyze defense based on the positions of players on the floor—not just the shooter.) By binning shots into 1' x 1' locations on the floor, Clemens was able to create a regression model that measured each player's impact on opponent shot location (square size) and field goal percentage (square color), separated from lineup performance.
Some important caveats before we dive in further:
- Clemens's model is new and still in development. Like any new model, it should be approached with healthy skepticism. While the player graphs largely pass the smell test, there are exceptions; James Harden comes out looking pretty good, for example.
- There are only so many NBA shots each year, and slicing and dicing by lineups and 1' x 1' floor location means we're not dealing with huge samples. The grey squares are "statistically insignificant" in terms of the player's impact on field goal percentage, but the significance bar being applied here is very low (p < 0.5). I would have preferred a model that used larger bins and returned a bit more confidence, but that's essentially how the charts are being read anyways. (The eye is drawn to clusters of reds and blues, instead of individual spots.)
- These charts are only showing the plays that end in shots, and forcing bad shots is just one aspect of good defense. Players that excel at forcing turnovers and defensive rebounding, among other skills, will see their defensive abilities underrepresented in these charts. This could explain the middling graphs for players like Marc Gasol and Joakim Noah, for example.
We these caveats in mind, here are the defensive breakdowns, based on regular season numbers, each remaining team's most common playoff lineup.
There was some concern this year that LeBron James lost a defensive step this season, and this seems borne out by his chart, which show a particularly large increase in three-point percentage for his opponents. Shane Battier has a lot of orange on the perimeter as well, although he also contributed 27 drawn charges—third in the league despite playing just 20 minutes per game—that aren't captured here. Chris Bosh had a mediocre defense rating this year (104), but these graphs make him out to be a defensive beast.
The Pacers had NBA's best defense this season, allowing 96.7 points per 100 possessions. While Paul George and Lance Stephenson don't come out looking great in these charts, their weaknesses are more than covered by the defense strengths of David West, George Hill, and, especially, Roy Hibbert.
With these five players taken together, the Spurs looking like a scary team to attack at the rim. During the regular season, they scored 4.1 more points-per-game in the paint than their opponents, the fifth-best mark in the league, and the best of the remaining four teams.
The Thunder had a great defense in 2013-14—their 101.0 points per 100 possessions was fifth-best in the league—but you wouldn't guess it from this individual charts. All five players are associated with improved opponent shooting beyond the arc, although the presence of Perkins and Ibaka, like Hibbert, also appeared to cause fewer threes to be attempted. This field goal improvement hasn't extended to the playoffs; Thunder opponents are shooting .342 from beyond the arc, in line with the Spurs (.345) and Pacers (.339) figures and well below that of the Heat (.387).
I'd encourage you to go explore some other players in Clemens's dataset. I'm not fully convinced of the results, but it's a worthy attempt to measure defensive impact, and very fun to look at.