Friday, November 17, 2017

Free Throw % Splits Based on Shot Type

I've always been a solid free throw shooter: I shot 75% over 4,250 attempts during the insane three-year stretch where I tracked every single shot I took on the basketball court (and 85% over 630 attempts in 2016)). But when I played organized basketball when I was younger I would always brick free throws when I shot technical foul shots and was alone at the line (although never as badly as my intramural teammate who had an 0 for 34 stretch shooting free throws). Was this just an abnormal anecdotal observation, or does NBA play-by-play data back it up?

It's been previously established that free throw shooters improve from their first shot to their second/third. Nylon Calculus has a great database on this phenomenon from 1999-2016, and TrueHoop suggested in 2011 that the first attempt is like getting to practice free throws in the middle of the game. But what about other splits like when you're alone at the line, or when you can take only one shot (such as an and-1)?

I pulled down NBA play-by-play data for the full 2016-17 season from BigDataBall (a great source at a relatively cheap price for all play-by-play data back to 2006) and looked at every shot attempt and found the same thing previous studies had:


Free throw shooters get better on their second/third attempts, to a very high degree of statistical significance. Comparing the shooting percentage between the first vs second/third shot gives a z-score of 13.56, which has an associated p-value of 1. Even comparing first vs second attempts gives a z-score of 12.54, which has an associated p-value of 1 as well.

But what about my earlier considerations regarding being alone at the line (with no rebounders around you)? I looked at "normal" free throws (on an and-1, two-point shot, or three-point shot) compared with technical/flagrant/clear path foul shots where the shooter is "alone":


3% of all free throw attempts last season occurred where the shooter was "alone", and players shot significantly better in this case, going against my hypothesis. But these "alone" attempts include technical foul shots, which are taken by the best free throw shooter on the floor (the shooting team gets to choose who shoots them). If I remove technical foul shots I get a different picture:


picture that illustrates no significant difference in make percentage (z-score of 0.35 with an associated p-value of 0.64, which is inconclusive).

My final look focused on only the first free throw taken in a set: does the shooter perform better or worse if they know they're getting additional attempts? I.E. Is an and-1 different from the first shot from a set of 2 or 3 attempts?



As before, I filtered out technical shots, and as before, there's no significant difference (z-score of 0.29, p-value of 0.61).

All in all it seems my experience was abnormal: NBA free throw shooters do improve after their first attempt, but make their shots regardless of whether there are other players around them. This makes sense, since they are professional athletes and I am not.

No comments:

Post a Comment