Using the Shot Clock Wisely

An area that many visitors to 82games never seem to find are the pages featuring the shot selection and shot clock usage numbers for teams and players. For shot type, we break it out by Tips, Dunks, "Close" shots and "Jumpers" (essentially anything outside the close range of the other three). Then we also divide shot clock usage into four segments:

Internally we even go so far as to cross reference the two views...so for instance looking at a player's jump shots in the 21+ second range. To see some samples of these, check out the Timberwolves in-depth stats or how about comparing Eddy Curry to Shaquille O'Neal.

For the teams we have just posted sortable tables that look at all the teams together on offense and defense in the different shot clock sections. With this data at hand we though it would be worthwhile to run a quick correlation/regression test to try and deduce if there is any strategic value to emphasizing performance in specific time sets.

1) Points Per 100 Possessions by Shot Clock Segment
By looking at a number based on points per one hundred possessions, you wrap up shooting efficiency, foul drawing and free throw shooting, turnovers, and offensive rebounding to get a very good overall team efficiency rating. Taking this and then running the correlations (for just this current 2003-04 season) to the team Wins on the season reveals the following:

With only the 29 observations it's pretty slim pickings in terms of coming to strong conclusions. However, the first read suggests that playing good defense in the early parts of the shot clock is a positive sign (all those coaches who harp on getting back on defense must be feeling pleased right about now), whereas offensively being able to score effectively in a half court offense is more important than high efficiency in quick baskets.

Overall when you take the offensive points per 100 possessions minus the defensive points per 100 possessions, the net arising from this shows that the correlation to wins is best in the quick shots domain, tapering off gradually as the shot clock usage extends.

This season, the top five teams in net for the 0-10 second category are the Pacers, Spurs, Pistons, Mavericks, and Heat.

Of course, the 95% Confidence Intervals for these correlations on the net stat says we can't accept the conclusion that this is indeed the appropriate order at the 95% level.

2) Correlation of Possession Usage Percentage by Shot Clock Segment
This time we'll look at the percentage of possessions used in a certain shot clock segment. So for example, on offense the Golden State Warriors "resolved" 40% of their possessions within 0-10 seconds, 24% within 11-15 seconds, 22% within 16-20 seconds, and 14% in the 21+ second range.

First off the correlations here are very, very weak indeed. A .01 correlation for example suggests next to no connection to the team wins. Part of the problem may be the way we are handling offensive rebounds, namely that the rebound counts in the shot clock segment of the missed shot, whereas the follow-up shot counts in the 0-10 class if taken quickly. A better look might be to only consider "first shot" possessions and exclude subsequent events after a missed shot.

You might think that a team that resolved fewer possessions than its opponents in the 21+ "rushed" stage of the shot clock would be at an advantage, but surprisingly the top teams in that regard were Boston, Sacramento, New Jersey, Dallas, and Phoenix. Two of those teams had losing records, whereas the Pacers had the 6th worst mark in that category, yet the best overall won-lost number.

3) Regression on Pts per 100 Poss. by Shot Clock Segment
Not that there's much to be gained from a regression run on 29 observations with eight stats per team in play, but here are how the numbers come back...with an R^2 of .97

Throwing the numbers all into the pot at once, projects the most important shot clock segment stat to be offensive efficiency in the 0-10 second range, followed by defensive efficiency in the 11-15 second range. The standard errors are once again vast, meaning the 95% CI for the Offensive 0-10 seconds is actually .81 to 1.36 in range (go 2x SE higher and lower).

Look for the easy/quick basket but don't get reckless...play tough defense especially in the early parts of the shot clock...and you may be slightly more successful than you are now!