I acquired all of the Division I team data since 2002, and from that we can observe some fascinating trends and data relationships in the data. This is a multipart series exploring some of that data.
Definition
In Dean Smith's book,
Basketball: Multiple Offense and Defense, Smith discusses evaluating a team's offense by its points per possession (PPP in other parts of this season, but PPS for this discussion). This is calculated by dividing the number of points by the number of possessions in a given period. Smith defines a possession as any loss of ball control. So, turnovers, trips to the free throw line (1+1 and 2), and
any field goal attempt are considered a possession's endpoint.
The number of possessions using this method is easily tallied from a box score, so long as the number of field goal trips is known. If it is unknown, I've determined (after about 900 UNC games) that these miscellaneous free throw trips are most accurately estimated with this equation:
The result is a combined possession tally for a game that generally averages somewhere in the 150s.
In the last 3 decades we have seen the growth of Analytics, the practice of evaluating performance by multiple mathematical factors. This has given better understanding of the game, and in many ways, made the game more interesting from a fan's perspective. In this period we've seen a new definition of a possession ending: any time the other team gains control of the ball. That definition shares with Smith's definition the turnover and free throw components, but differs with the field goal attempt component. The Modern Definition is calculated by determining the Smith Method total, and subtracting offensive rebounds. This difference has resulted in possession totals that are generally in the lower 130s.
Because possessions are in the denominator of the Points Per Possession quotient, the lower number of possessions in the Modern Method results in a higher points per possession than the Smith Method. The NCAA didn't start publishing Offensive Rebounds until the 2015 season, so I can only demonstrate the differences over the last 10 seasons:
In the graph we see that the annual average Points per Possession for the Modern Method (PPM) is higher than that for the Smith Method. The key, though, is that the shapes of these lines are relatively the same. So, the two methods, by themselves, don't really tell different stories.
In the 3,484 team seasons since 2015, PPS averaged
0.903 (0.012 std dev) and PPM averaged
1.036 (0.0134). PPM is 14.7% higher than PPS, so we can convert Smith numbers by multiplying them by
115% (1.147) for comparison.
Target Values
Dean Smith's goal for his offenses was 0.95 points per possession, and his goal for the defense was to allow no more than 0.85. What does this mean with the Modern Method. If we look at Smith's goals in terms of standard deviations from the national average, and convey that to the Modern Method's average and deviations, we can compute the equivalent range to be 0.98 - 1.09. Therefore, when you hear sports radio personalities talk about about teams and their points per possession, remember that Dean Smith would have targeted
1.09 for offense and 0.98 for defense.
Next up: Points Per Possession against Winning Percentage (Smith vs. Modern)
