Estimating the number of possessions
When estimating the number of possession, we have to estimate:
- The number of times a team went to the free throw line as a result of a foul, that wasn’t preceded by a shot. This is what I call a free throw trip.
- The number of times a possession continued because of an offensive rebound.
I have chosen to use an fta factor of 0.46, based on references of Oliver [1] and in-the-game.org [2].
Here are the formulas I use:
When accounting for second chance shots (after an OR), Oliver advices to use 0.44 instead of 0.40 to estimate the number of trips to the free throw line. I use 0.46:
FTtrips = ftaFactor*FTA # ftaFactor = 0.46
I decided to use 0.46 this after some graphical exploration, this is till arbitrary and I’m open to any feedback here.
Plays are then calculated:
plays = TO + FTtrips + (FG2A + FG3A)
And to calculate the number of possessions, we have to subtract the number of plays that occurred after an offensive rebound [1]:
# secondChanceFactor = 1.07
ps = plays - secondChanceFactor * (FG2A + FG3A - FG2M - FG3M) * OR / (OR + opp_DR)
The secondChanceFactor is taken from [1], but I must say I don’t really understand why it would be better than simply using the OR term. If any, I’d expect this term to be less than 1, because some offensive rebounds occur in a possession that had an OR before, thereby not continuing a possession. Thoughts are appreciated.
Footnotes
| [1] | (1, 2, 3) Basketball on Paper, D. Oliver |
| [2] | http://www.in-the-game.org/?page_id=10227 |