Editor's note: David Roher is co-president of the Harvard Sports Analysis Collective, a student organization. See his other work at the group's web site. This is the first installment of a new series of blog posts about sports statistics and strategy.
To refresh your memory: it was a see-saw game, and the Celtics' six-point lead with around three minutes to play had dwindled to one point with less than a minute remaining. As the Mavericks looked for the go-ahead shot, Kevin Garnett intercepted Jason Kidd's pass. He safely dumped the ball off, and the five green jerseys meandered to their side of the court. More than 20 seconds after his steal, Garnett attempted a jumper. I'll leave out what happened next, not to spare you the pain, but to show that it's the process that matters here, not the result. It would have been nice if Garnett had made the basket. Duh. My issue is how long it took him to take the shot.
Readers might think that it's better to slow the pace when ahead late, and they're almost always right: longer possessions mean fewer possessions and fewer opportunities for the losing team to claw back. Factor in the Celtics' normally slow pace (91 possessions per game, 20th in the NBA) and it's an even stronger case. However, there’s another important factor here: the end of a quarter is a special scenario in which a team can gain an extra possession. When teams get the ball with 30-40 seconds to go in the first three quarters and get a shot off quickly, they deny their opponents a chance to run out the clock because they are almost guaranteeing themselves another possession.
The strategy is called the 2-for-1, and the statistically minded Celtics use it a lot in the first three quarters of a game (they'd used it at least once in their previous six outings). One of the reasons it works is that possessions ending in the first half of the shot clock usually result in more points than those that end in the second half. In a study my club did two years ago, we found the expected return from a possession was highest after an attempt just two seconds into it. This figure plateaued until about seven seconds into the possession, after which it gradually declined until the shot clock expired. The total decline was worth about .25 points, or about 22 points over an entire game. Think about it: as the clock winds down, players more often shoot because they need to get a shot off, not because they have a good chance of making the basket. As a result, a team isn't sacrificing much by shortening its first possession.
The problem with evaluating the 2-for-1 in the fourth quarter is that there are often more important strategic motives. When my club did a study of the tactic two years ago, my colleague Dan Yamins found that he had to exclude fourth-quarter data. After Garnett stole the ball on Friday, he and his teammates were probably thinking about those other factors that screwed up our study: if they could just make one more basket, they would be up 3 and have a great chance of winning.
They probably weren’t thinking about what would happen if they missed and the Mavericks followed with a basket, or even if the Celtics made a two-pointer and the Mavericks followed with a three. They'd want an extra possession, one with enough time to get a decent look. But by taking around 21 seconds in their possession, they were effectively dismissing these situations.
As it happened, Garnett missed and Jason Kidd hit a three with 2.5 seconds left. The Celtics were forced to attempt an inbounds-pass alley-oop that sailed straight out of bounds. The game was effectively over. Again, it's not the result that matters, but the result happens to provide a convenient example of the benefit of taking a faster shot: even in the worst possible case, the Celtics would have had enough time to send the game into overtime or win it with a three.
We do have to consider the impact on the best-case scenario: that the Celtics hit the initial shot. In this case, taking a quick shot admittedly increases the chance that the Mavericks would repeatedly foul the Celtics, extending the game as long as possible. However, the additional chance of fouling is slight and rarely works out for the fouling team. Recall all the times you've seen a team try to foul at the end of the game, and then count how many times it actually worked.
I'd guess that the lapse in judgment occurred on the floor and not the bench: the Celtics are one of the most statistically inclined teams in the NBA. It's pretty difficult to weigh your strategic goals while playing in a demanding game, especially when the right call, the 2-for-1, is usually only applicable in the first three quarters. Regardless, Celtics fans should hope that their team will be a bit hastier in future similar situations.