> A couple things...

Despite having taken several years of calc so far, Ive actually never

>

> 1. The band getting narrower with more poss simply expresses that you

> have more variance with fewer attempts, which is what statistically

> should happen. Variance of efficiency is theoretically inversely

> proportional to how many possessions you take.

>

> 2. In Basketball on Paper, I generate functions that show how

> players' efficiencies vary with possession percentage (percentage of

> the team's possessions, with an average of 1 out of 5 or 20%). It's

> one of the most useful things I do. It suggests whether a player can

> use more possessions and still be efficient. It suggests how to

> optimize an offense. It says why Allen Iverson is valuable even if he

> is inefficient. You should take a look at that -- Chapter 19. (I

> know, the axes are backwards on my plots. It's a relic of how I had

> to originally generate them years ago.)

>

> 3. You are definitely right in your theory that efficiency should go

> down with poss used. Part of that theory should also incorporate how

> good teammates are. That's the tough one to work with...

>

> DeanO

taken stats, in highschool or college, so unfortunately all I have is

what I can explain logically to myself, I can't compute the other

stuff. I feel like I have a pretty good grasp of most things

conceptually, but when it comes to things like variance etc, I dont

actually know how to find it. That said...

In response to 1, this is OP/M as opposed to OP. Unless I

misunderstand something, the narrowing of the band is not due to

variance in OP/M, as a low number of OP over an even lower number of

minutes can still produce a high OP/M number, and yet the variance is

only shown to the left of the graph, and not to the right.

2. Yes, there is a subtle difference between % of team OP and OP/M, as

you could probably argue that all of the players on a team like Dallas

are "unfairly" getting a boost to their OP/M, but on the other hand

using % of OP of total OP would "unfairly" reward players on slow

tempo/defensive type teams, like say Rip on Detroit. I dont remember

if I ever got to chapter 19, I put it down at some point and decided I

wanted to know more about computing statistics to be able to look over

the formulas myself before reading further. I might go back and look

it over now :)

3. Yes, but I believe the effects of playing with offensively good

teammates is overstated a bit. Part of that effect overlaps with the

OP/M effect. Good offensive teams have many offensive players, which

means any given player is more likely to have a lower OP/M than he

would on an average team, thereby indirectly raising his OE. The

converse is also true, where bad offensive teams usually have very few

good offensive players and a player's OP/M will likely be higher (See

McGrady, Orlando).- Also, here are the top 20 players in adjusted offensive efficiency

from the formula used in the article:

1. Brent Barry

2. Peja Stojakovic

3. Brian Cardinal

4. James Posey

5. Corey Maggette

6. Shaquille Oneal

7. Yao Ming

8. Antawn Jamison

9. Steve Nash

10. Fred Hoiberg

11. Ray Allen

12. Mark Blount

13. Reggie Miller

14. Antonio Daniels

15. Kobe Bryant

16. Sam Cassell

17. Elton Brand

18. Jarron Collins

19. Dirk Nowitzki

20. Richard Jefferson