Tobacco Road Report

What are “tempo-free” statistics? Tempo-free statistics are a means by which to adjust basketball statistics for number of possessions during any given game. Since basketball statistics are greatly affected by the pace of play and the number of possessions each team has, it makes sense to adjust the traditional basketball box score with possessions serving as a reference point.

The traditional statistics used to measure player or team performance in basketball – total points, assists, rebounds, average team points per game, etc. – are all affected by the pace of play. This is opposed to baseball where the number of innings and outs determine a minimum number of opportunities for each team. As we often say of statistics, we want to compare apples-to-apples, not apples-to-oranges, and adjusting for tempo allows us to do this.

We at TRR are not the first to use tempo-free statistics or promote their efficacy. Our colleagues Ken Pomeroy (www.kenpom.com) and John Gasaway (bigtenwonk.blogspot.com) have long promoted the effectiveness of tempo-free statistics, and creating possession-based statistics goes at least as far back as Frank McGuire and Dean Smith. While we are not the first, count us in as a supporter of more meaningful statistics for the game of basketball.

Over the next few installments of our tempo-free statistics series, we will bring you a selection of possession-based statistics that we feel are particularly useful for evaluating team performance. This installment includes discussion of estimating the number of possessions per game, which is used in a variety of ways to detail the level of offensive and defensive efficiency of any team, a subject we leave for later posts.

Counting the Number of Possessions
The logical first step in computing possession-based statistics is to calculate the number of possessions a team had during any given game. To do this, one needs to think of the different ways a possession can end in traditional box score statistics (as opposed to literally charting every possession of a game). An offensive possession can end with a field goal attempt, a turnover, or a foul that results in free throw attempts.

We view offensive rebounds as extending any given possession and, consequently, adjust the number of possessions downward by this amount. Our methodology, therefore, generally follows that of Pomeroy, Gasaway, and Basketball on Paper by Dean Oliver (www.basketballonpaper.com).

The tricky part comes into play when thinking about possessions that end in free throw attempts. Some attempts are one-and-ones, some are two shot situations, and others are three point plays the old fashioned way. As a result, we cannot simply subtract off the number of free throw attempts like we can with offensive rebounds. The correct measure is the fraction of free throws taken which result in a change of possession?

Fortunately, Ken Pomeroy has put some effort into watching a large enough sample of games to arrive at an estimate, and we take him at his word when he says 47.5% of every free throw taken results in a change of possession. As of our last reading, Gasaway uses this number and Oliver uses something in the neighborhood of 40%. Any faithful reader willing to examine a large enough sample and report back to TRR should feel free. Your efforts will be duly noted and appreciated within our hallowed halls.

Therefore, applying the above methodology results in a number of possessions equal to
Possessions = (FGA+TO – OREB + 0.475*FTA).

As an example, consider the following statistics for Boston College in their game with Florida State on January 14, 2006. During this game, the Eagles attempted 58 field goals (FGA), committed 11 turnovers (TO), grabbed 13 offensive rebounds (OREB), and attempted 27 free throws (FTA). This results in 68.8 possessions for the Eagles during the game.

We caution our readers that this number is merely an estimate, but one that has proven fairly effective over time at estimating the true number of possessions in a game. A quick reading of the other users of tempo-free stats will show that some use additional fine tuning from here to arrive at what they feel is a more accurate estimate. However, we prefer to keep it simple and use the above measure.

We’re now ready to go. Computing the number of possessions during a game is the first step towards unlocking the secrets behind tempo-free statistics. Return to TRR for subsequent posts on how we use the number of possessions to evaluate how each team in the ACC stacks up against the competition on a tempo-adjusted basis.