A discussion of usage versus efficiency: specific players.

[acrossthecourt]

I recently updated the (somewhat) famous Eli study of usage and efficiency to include five years of lineup data:
http://ascreamingcomesacrossthecourt.bl ... iency.html

Summary of results:
Basically, you're looking at this at the lineup level (all 5 man combinations with possessions in the regular season.) You calculate the expected offensive efficiency of a lineup based on an individual player offensive efficiency metric (like b-ref's, and for good measure I used a really simple one that's just points, shot attempts, free throws, and turnovers.) Then you compare it to the actual offensive efficiency of the lineup. Also compute the average usage of each lineup. (Each player has a usage rate, and you just calculate the average of the five man lineup's with their respective season averages.) Do this thousands and thousands of times, and now you can find a statistically significant link between usage and efficiency.

The results were pretty similar to Eli's. A 0.8 to 1% change in average lineup usage (average usage is 20, by the way) translates to a 1-unit change in points per 100 possessions (or a 0.01 unit change in points per possession, which is what most of the numbers are in.)

http://4.bp.blogspot.com/-6d6bir24VtU/U ... ffusg2.png

I want to continue to refine this and study it further. My question is, how does this relate to specific players? What do *you* think? I gave a good example of Carmelo Anthony in there. On the Nuggets before Iverson, if you replace him with a crappy player who never shoots but his efficiency is the same, how do you expect the offense will respond? Isn't this method finding the value of things like double teams and defensive pressure? And that you need someone to take difficult shots before the clock expires because your play didn't work?

Which players "make the most" out of their usage? (That is, they raise the efficiency of everyone else indirectly.) Which players make the least out of it? (Guys who stop the offense cold and refuse to take difficult shots when the team needs them to.)

Thanks.

[penbeast0]

I moved this to statistical analysis . . . I've quoted the Eli article from APBR in discussions and I've found that the level of statistical sophistication on the Stat board is pretty high and much more likely to give you the help you are looking for.