Friday, September 15, 2006

NBA time of possession doesn't matter much

Here’s a nice basketball study from … as it turns out, it doesn’t tell us all that much, but it’s interesting nonetheless.

In their “Random Stat” column, the (anonymous) author charts how time of possession relates to winning. It turns out there’s a small positive correlation of .13, meaning that teams who hold the ball longer have a slight tendency to be more successful. (But there’s a huge confidence interval around that .13, from -.24 to .+.47, “so it could be anything basically.”) And the bottom three teams made the playoffs, which reinforces the idea that the stat doesn’t matter that much.

The most extreme team was Phoenix, with a possession time of 46.94%. That means that instead of having the ball for 24 minutes, they had possession for only 22:28 – about one and a half minutes less than average. Assuming about 105 possessions per game, it means that every time they had the ball, they held it 0.84 seconds less than an average team.

Perhaps the reason time of possession doesn’t help much is that there are offsetting reasons why a team could be holding the ball less. On the positive side, they may be able to score faster. On the negative side, they may turn the ball over more. And the opposite applies to the other team’s time of possession; a good defense might keep the other team in possession for the full 24 seconds as they try (and perhaps fail) to get open for a decent scoring chance.

Another “Random Stat”
study on steals shows how this might apply. Phoenix’s Steve Nash led the NBA last year by having 137 of his passes stolen. Assuming that the Suns would have kept the ball ten seconds longer if the pass wasn’t stolen, that’s 1370 seconds of possession lost, or about 17 seconds a game – all due to just the one cause, Steve Nash getting intercepted.

But Nash’s teammate, Shawn Marion, led the league the other way, by stealing 121 opponents' passes -- so it’s a bit of a wash. As a team, the Suns were fourth in the league in stealing passes, which increases their time of possession. Offsetting that, they were dead last in steals off the dribble or loose balls – they stole only 122 balls that way, while the league leader, Charlotte, stole 292.

Basically, what the correlation of .13 for time of posession tells us (assuming the .13 is close to the real value, and not just random luck) is that, in general, the good reasons for holding the ball longer tend to occur a bit more frequently than the bad reasons. Which doesn’t tell us a whole lot about strategy, or even about the tendencies of any given team. But it’s kind of fun anyway.


Post a Comment

<< Home