Putting the Team on Their Back? The Usage and Efficiency of NBA Superstars in Critical Situations

By Matthew Goldman, economics Ph.D. student at the University of California-San Diego and Justin M. Rao, Ph.D., economic researcher at Microsoft Research

Do star players “put the team on their back” in critical situations? While it’s relatively easy to remember a game winning shot, it can be hard to tell just how the offensive load is shifting over the course of the game. In this post we use 5 years of play-by-play data to answer this question quantitatively.

We examine how two key metrics of offensive performance vary across the game state, which we define with the current score margin and time remaining at the time of a shot. The first is “usage rate”: the fraction of possessions in which the player takes the team’s first shot. The second metric is “efficiency”: the number of points on a given possession in which the player took the first shot (note we include free-throws made if the player was fouled and points scored off an offensive rebound).

In work that we have submitted to the 2013 MIT Sloan Sports Analytics Conference, we identified an interesting pattern in team efficiency: the team that is currently losing, especially late in the game, has significantly higher efficiency than the leading team. We were careful to control for offensive and defensive line-ups and threw out “garbage time”; what we found is that a given line-up gets better when the team is down late and worse when the team is up late in the game. Put simply: currently losing motivates (or currently winning makes you complacent or both).  And this is true, even if the game remains well within reach for both teams.

Here we delve in to this effect for the game’s top players. In Figure 1 we present 3-D plots of the usage rates of Kobe Bryant and LeBron James as a function of game state.

Figure 1: Usage rates for LeBron James and Kobe Bryant

Kobe and LeBron both shoot over 30% of their team’s shots on average when they are on the court. Two patterns are clear for both players: they shoot more in the second half and they shoot more when their team is trailing, especially towards the end of the game. This effect is particularly striking for Kobe, who shoots over 40% of his team’s shots when trailing by 10 late in the game. When down by 5 points late, he shoots 37% of his team’s shots, far greater than the 27% he takes in a typical first half situation. Strikingly, when the Lakers are up by 5 points late, a game that is still within reach for the opponent, Bryant’s usage drops to near 1st half levels. LeBron shows a muted version of the same pattern: he shoots 33% of the team time when trailing late vs. 30% of the time when holding a small lead late.

Both these players, especially Bryant, exhibit we what we call “loss averse motivation”–currently losing motivates them to action. “Loss aversion” is a concept in economics and psychology that says, roughly speaking, people are more motivated to avoid losses than to seek similarly sized gains. When trailing late in the game, losing becomes more and more likely; the thought of a relatively likely loss appears to light a fire under these guys, especially compared to the thought of a relatively likely win.

What happens to efficiency? Figure 2 gives similar plots for points per shot attempt (efficiency):

Figure 2: Efficiency for LeBron James and Kobe Bryant.

Kobe has the highest efficiency when trailing early in the game. Late in the game he does slightly better when trailing as opposed to leading, but the effect is not dramatic. Recall he is shooting much more when trailing, typical usage curve logic would say he should do worse per shot when shooting more, but if anything he does better. This is particularly striking evidence of motivational loss aversion. LeBron is also more efficient trailing, as seen by the downward slope from left to right. He is 5-8% better down late in the game vs. being up by a similar margin.

Kobe and LeBron are not unique in responding to losing. In Figure 3 we plot the efficiency rates for Chris Paul and Kevin Durant, both of whom show similar patterns.

Figure 3: Efficiency for Kevin Durant and Chris Paul.

However, the loss-averse pattern is not universal either. Our analysis has identified 3 key patterns. The first is the loss-driven group. The second is what we call “zero margin” types, these guys are most motivated when the game is close. The third we call “first halfers”.

Two good examples of zero margin players are Dirk Nowitzki and Russell Westbrook. Figure 4 gives their usage patterns:

Figure 4: Usage for Dirk Nowitzki and Russell Westbrook.

Notice both tend to shoot more along the spine of the graph, which is at margin=0. They want the ball in their hands when the game is close, especially late. Interestingly, although Westbrook likes to shoot when the game is close in the 4th quarter, he does worse in these clutch moments. In contrast, Nowitzki ups his game. These patterns are shown in Figure 5:

Figure 5: Efficiency for Dirk Nowitzki and Russell Westbrook

Two examples of our “first halfer” category are elite big men. Figure 6 gives usage rates for Dwight Howard and Tyson Chandler:

Figure 6: Usage for Tyson Chandler and Dwight Howard.

First halfers do just as the name implies, they shoot the ball much more in the first half. Both of these players’ usage rate drops by more than half over the course of the game. Late in the game, they rarely get looks, but the effect is not driven by the final 2 minutes, it’s is consistent over the course of the game. So while elite ball handlers like Kobe, LeBron and Chris Paul shoot less in the first half and more in the second half, elite big men seem to show the opposite pattern.

Overall we find that indeed some players put the team on their back in key situations. A loss-averse type does this when their team is losing late—they seem to hate thinking they’ll lose. Interestingly they don’t shoot more with a narrow lead, even though that lead could slip away with poor performance. Some players are more motivated by a very close game rather than losing per se, however the effect is not as dramatic as that we see in the loss averse types. And finally we can see that the players who definitely do not put the team on their back are first half big men.

About the authors: Matthew Goldman is an economics Ph.D. student at the University of California-San Diego. Justin M. Rao, Ph.D., is an economic researcher at Microsoft Research in New York City. The pair has had two research papers in the Sloan Sports Analytics Conference over the past two years, with last year’s entry, “Pressure vs. Concentration: The Asymmetric Effect of Pressure on NBA Performance,” earning the ESPN Fan’s Choice award.

Editor’s note: The views expressed in each post are those of the author(s) only and not those of the conference organizing team or blog sponsor.