Wednesday, December 08, 2004

Basketball Statistics Could Be So Much Better

Responding to Ken's post...

Before we consider the relative difficulty of passes counted as assists, we could take the intermediate step of counting all the points derived from good passes. I've always thought it silly that a passer gets no credit for a pass that forces the defense to foul a shooter rather than allow a lay-up. I'll speculate that giving credit in an assist-type stat (Points Created?) for free throws made following a good pass would separate passers to a greater degree than differentiating between setting up 2- or 3-point shots. Consistently getting the ball to a good free throw shooter with positional advantage has better than value than, say, consistently getting the ball to Nick Collison on April 7, 2003.

On the flip side, turnovers that merely end a possession (traveling, three seconds, offensive foul, shot clock violation, etc.) are not quite as bad as being stripped of the ball and immediately surrendering a fast break bucket.

You'd probably need a baseline of points per possession and points allowed per possession in order to normalize both the value of Points Created and Points Surrendered. In fact, in an ideal model you'd differentiate between points and points allowed per half-court and fast break possessions. That, in itself might sort out some of the degree of difficulty with regard to passes. Converting a 3-on-1 break is a generally available skill. The act of breaking one up is extremely valuable, akin to Torii Hunter reaching over the fence to bring back a home run. (Granted, preventing two points in a basketball game is less valuable in context than preventing 1 to 4 runs in a baseball game.)

No comments: