Is There a Relationship Between Usage and Efficiency in the NBA?
No.
There were 276 players to play at least 1,000 minutes in the 2019 NBA season per stats.nba.com. Here is the plot of their USG% (the percentage of a team’s field goal attempts, free throw attempts, and turnovers used by the given player) and TS% (points per combined field goal attempts and free throw attempts) from the same source:
If efficiency was inversely proportional to usage we would expect a line roughly descending from Mitchell “Mitch” Robinson with his #1 efficiency and #265 usage to Russell “Russ” Westbrook with his #251 efficiency and #10 usage.
If efficiency was directly proportional to usage we would expect a line roughly ascending from Bruce Brown at #271 and #274 to James Harden at #1 and #28. (Or arguably Giannis Antetokounmpo at #4 and #9 just beyond the J in James, but someone in the upper right anyway.)
We instead get a square, and while the regression suggests there is almost certainly no relation between usage and efficiency, if anything efficiency slightly increases with usage.
.
Okay, but what if we look at individual players over time? There were 99 players to play at least 1,000 minutes in each of the last five seasons, let’s see how the USG% to TS% graph looks over those five seasons for the three players who saw the largest swings in USG%:
Giannis Antetokounmpo got more efficient as he took more touches, Harrison Barnes got less, Brook Lopez was efficient with his least and most touches and nowhere in between. We could not ask for less helpful data in determining the usage-efficiency curve… but what if there are many usage-efficiency curves?
Obviously Giannis Antetokounmpo’s team is better with the ball in his hands than without, but what if Giannis himself is too? What if he’s not great at setting screens and therefore ineffective as a roll man in addition to his struggles as a spot-up shooter? Last year he took only 75 shots as the roll man, on par with Daniel Theis and Harry Giles III even though he took twice as many shots as they did combined. We see Giannis settle for a long jumper out of isolation and imagine him throwing down an alley-oop as an alternative, but what if it’s the other way around?
Obviously Harrison Barnes is the classic case illustrating how a player only has so many high quality shots and has to settle for inferior ones to increase their volume, as his raw number of three pointers plateaued in Dallas while his overall field goal attempts skyrocketed, but a deeper look shows the issues ran even deeper: after taking 208 corner threes in the two years in the sample he was in Golden State, he took a lower raw number in his two (full) Dallas years with only 148. Not only was he adding bad looks, he was subtracting good ones. Perhaps he was overcompensating? Perhaps in addition to taking on a larger responsibility his team was just worse?
Obviously Brook Lopez started taking threes which is why his low usage year is so good since as a complementary player he only takes those now, but he was taking threes last year in his worst efficiency season too, and the year before when he was still a primary option in Brooklyn. He’s posting up way less in Milwaukee than LA but he was posting up way more in Brooklyn than LA. The key is that he was just posting up better in Brooklyn by drawing dramatically more fouls, which with his career 80% mark from the line is a decisive factor.
.
So we can talk ourselves into explanations for why any given player’s efficiency changed when their usage did, past tense, and some of them might even be right! But it’s always good to have a control too, so let’s look at the three players who saw the smallest swings in USG%:
Marco Belinelli at least has been on a(t least one) different team every year, so we can pretend there’s a reason his efficiency has twice the variation of Harrison Barnes’ with only a tenth(!!!) of the usage variation. How did Andre Iguodala go from 62 TS% on 11 USG% to 54 TS% on 12 USG% in back to back years? How did he get back to 59 TS% on 12 USG% the next? He took about three quarters of his shots at the rim or from three each year, he was on the same very very good team, what explanation could there possibly be?
.
If projecting how efficient a player will be with the same usage year over year is shaky, projecting a usage-efficiency curve for any given player when there is no league-wide or generalizable individual trend is surely a fool’s errand.
