APBRmetrics

[metalloids]

On page 28, footnote 3 of "Basketball on Paper" there is a technical definition of possessions:

Possessions = FGA - OREB/(OREB + DDREB) x (FGA-FGM) x 1.07 + TOV + 0.4 x FTA

The simpler version given earlier is: Possessions = FGA - OREB + TOV + 0.4 x FTA.

I am trying to understand the difference between the two, and its justification. Here's how I see it: only shots that are missed can be rebounded, but not all missed shots are rebounded (e.g. block going out of bounds, loose ball foul on the defense fighting for rebound). Any equation that estimates possessions should take these exceptions into account by not adding an extra possession for them (unless the "plays" definition of possession is used, but let's ignore that).

It seems like these exceptions, however, are already covered by the factor (FGA -FGM)/(OREB + DDREB). This is because the sum (OREB + DDREB) is usually close to the difference (FGA-FGM) but always less, due exactly to these exceptional cases.

So, my question is: why is the 1.07 factor there?

Thanks for your help.

[kjb]

I'll leave Dean to explain the 1.07 factor in the formula he presented in BoP.

There is a tweaking of Hollinger's basic form that a group reported in a recent study:

.976 x (FGA + .44 x FTA - OREB + TOV)

[HoopStudies]