By Erik Johnsson:
Since its birth in 1979, we’ve watched the 3-point shot evolve from a signal of desperation to a cornerstone of offenses around the league. Any fan of the sport knows that teams have gradually embraced the long ball in place of the mid-range, and they show no sign of slowing down. Every year since the 3-point arc was extended to 23.75 feet in 1997, the league has steadily and substantially increased its 3-point attempt rate. To fans, players, and coaches alike, it is clear that basketball will only continue to drift beyond the arc.
Luckily, we don’t need a crystal ball to see what a future NBA might look like. Over the past decade, teams have consistently upped their appetite for threes when it matters most: clutch time. With less than 5 minutes remaining and within 5 points of their opponents, teams have always been more willing to step behind the arc – and not just because of last-second heaves. If we remove all shots taken in the last second of the game, and all shots beyond 40 feet, the league still shoots far more threes in the clutch than it does otherwise (for reference, Damian Lillard’s shot to finish off the Thunder was 38 feet from the basket).
If what happens in clutch time is any indicator of the future of basketball, then we’ve only scratched the surface of 3-point shooting in the NBA. When games get tight down the stretch, teams become increasingly reliant on the 3-ball – even if they shoot it less efficiently. In fact, in the 2018-2019 NBA season, only four teams took fewer 3-point shots in the clutch than their starters did in the first quarter:
It appears that teams hoist up more threes in the clutch, attempting to raise their efficiency when it matters most. But what shots are these new 3-pointers replacing, and who shoots them? Are coaches just putting more three-point specialists on the floor in clutch time, or are players choosing to take more of their shots from deep?
To answer the first question, we can look at the league’s shot selection in the clutch compared to the shot selection in the opening minutes of each game. If we divide the court into six different zones, we can compare the frequency of shots from each zone during the two time periods. On the left is a shot chart of all shots taken in the 2018-19 NBA season, colored by zone. On the right is the whole league’s shot distribution, in and out of clutch time. As you’ll notice, those new threes aren’t just repurposed long twos; teams are taking more threes, even in favor of shots from the restricted area.
Of course, there’s a second way to interpret this information. Instead of offenses deliberately changing their shot selection down the stretch, perhaps defenses tighten up and force their opponents to take deeper shots. This might explain why, on average, players seem to raise their three-point attempt rate about .035 percentage points in the clutch. This difference is statistically significant (p < .05), and is shown by the blue line on the plot below. Only players with at least 40 shots attempted in the clutch are shown:
Note: If we lower that threshold, the trend actually gets stronger; we start to capture players like Wesley Matthews who appear to function almost exclusively as sharpshooters in the fourth quarter, but don’t get nearly as many looks as other, more ball-dominant stars. For the sake of having a more accurate sample, these players were left out.
Players above the zero-line shot more threes in the clutch, and players below the line shot fewer. As expected, most players fell above the zero-line. Even Anthony Davis, who normally took just 13% of his shots from 3, jacked his 3-point attempt rate up to 30% when games were tight down the stretch. Though, there’s good reason to think that some of these additional threes were forced by the defense; in the clutch across the league, teams tended to shoot more threes in the form of pullups and step-backs, rather than straight up jump-shots. Knowing that these shots are generally less efficient than standard J’s are, it’s likely that clutch defenses are forcing their opponents into tough one-on-one scenarios (of course, some stars are more comfortable handling this than others are).
In reality, all this is probably the result of both offenses changing their playstyles to maximize efficiency, and defenses getting more stingy in the fourth. When defenses lock up at the end of games, they force opponents to lean on their biggest stars to generate points, many of whom rely on stepback and pullup threes to score most efficiently in one-on-one scenarios. But of course, stepbacks and pullups are less efficient than the types of threes that teams might get when they’re not in the clutch. Not to mention, a few of these shots are coming when offenses have no choice but to shoot threes, just to have a chance at winning the game. Though, with more granular data containing the position and orientation of each defender on the floor, we could ask whether or not clutch threes are more-heavily contested than others are. In turn, this might tell us which side of the ball is really dictating the hyper long-range brand of basketball we see in clutch time.
If you have any questions about this post, you can reach Erik on Twitter (@ejohnsson50) or email (ejohnsson@college.harvard.edu).
Bonus: In case you’re interested, check out the players who saw the biggest changes in 3-point frequency when it mattered most. As you’ll notice, AD isn’t the only player who might have taken some ill-advised threes in the clutch. Many players at the top of the scatter plot might have benefited from dishing the ball to a teammate more often (looking at you, Russ). Remember that some of the players (especially near the bottom) on this list might have misleading 3-point percentages in clutch time. Julius Randle only took five threes in clutch time, and he happened to make four of them. Giannis shot seven, and made three. No matter how hard they try, neither of these players will shoot that well in the long run, at any point in the game (sorry Giannis).