Stat Dive (part 23): Points Per Possession (Smith vs. Modern)

[JimmyNaismith]

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.

How valuable are Points Per Possession figures, and which (Dean vs. Modern) is better at estimating Winning Percentage?

I evaluated the 3,133 team seasons since 2015 that had complete data sets. From this I ran Excel's Correlation function and found that all of these stats were good ones. Here are Points Per Possession with Smith Method (PPS) and Modern Method for Offense (uppercase) and Defense (lowercase):

• PPS: 0.689
• PPM: 0.771
• pps: -0.671
• PPM: -0.708

For these figures correlation ranges from 0.0-1.0, so all have decent correlation with Winning Percentage. However we see significantly higher figures for the Modern Method and the Smith Method. This is because the Smith Method includes no information about Rebounding, while the Modern Method, because it subracts out Offensive Rebounds, has more information and is more accurate.

If we create multiple regression analyses, however, we get a slightly different picture. Here are the resulting equations (R-squared):

1. W% = 1.977*PPM - 1.535*ppm (0.986)
2. W% = 2.294*PPS - 1.748*pps (0.980)
3. W% = 2.304*PPS - 1.764*pps + 0.000648*POSDIF (0.986)

The R-squared value in parenthesis ranges from 0.0-1.0, so the closer we are to 1.0, the more perfect fit we have. Here we see that the equation that uses Smith's PPP numbers (#2) only isn't quite as good of a fit as the one that uses the Modern Definition (#1). However if we add in the Possession Differential (#3), we get a fit that is just as good as with the Modern Method.

The real point of all of this is that the Modern Method is more accurate because with two variables it explains the Winning Percentage as well as Smith's Method using 3 variables. That said, the differences in accuracy of all of these is insignificant, so they are all good, and that is news to me.

Next up: Points Per Possession against Winning Percentage (Smith vs. Modern)