APBRmetrics

Getch · Joined: 03 Dec 2006 Posts: 1

In Dean Oliver's book, he gives the following formula as an estimate of team scoring possessions:

Team Scoring Possessions = Field Goals Made + (1-(1-FT%)^2) x FTA x 0.4

Can some one please explain to me how this formula was derived, as well as what each component means? (i.e. why is (1-(1-FT%)^2) or 0.4 included in this formula)

Finally, why do people calculate scoring possessions? What does it tell you? Does scoring possession alone (as opposed to its use in the Field% formula) have any utility?

Thanks for the help!

milt · Posted: Sun Dec 03, 2006 7:02 pm Post subject:

(1-(1-FT%)^2) looks like the probability that a trip to the line results in at least one point, assuming there were two free throw attempts.

0.4 is the number of possessions per free throw attempt. It's not exactly 0.5 (i.e. 2 FTAs = 1 possession) because it accounts for and-1 free throws, as well as fouls on 3-point attempts.

Ed **Küpfer** · Joined: 30 Dec 2004 Posts: 785 Location: Toronto

Because under this scenario possessions have a binary score/no score result, Floor% follows a binomial distribution, making variance and confidence interval calculations much easier. Multinomial variance, for things like EFG% or TS%, is freaking complicated (I outlined the method somewhere on this site) -- it's much easier to go with FL% * (1 - FL%) / POSS.
_________________
ed