I had to head out the door before getting to this last thing...
> As for correlation with team success, this is definitely an
> intriguing avenue; I feely admit, when I total a team's talent (as
> measured by me), I don't get a team ranking compatible with their
> success rate (Denver looks better than NY).
> So I blame it on coaching. Easy!
Your system and just about every "linear weights" system (as I label
all these that add positives and subtract negatives) do tend to have
this problem. They don't capture the defensive teams very well, which
is NY (and Miami and probably Philly), and overrate the offensive
teams (I really don't think too highly of Antonio McDyess). Even
Heeren's system, which compensates for pace, doesn't do defense
justice, as far as I'm concerned. Defense is just underweighted in
the formula because there just aren't as many defensive stats kept
(dreb+stl+blk is often less than pts+oreb+ast-errors). Defense is the
biggest thing that players like Duncan, Mourning, Webber, Garnett
provide (though they all provide some O, too). Defense is why Mutombo
and Ratliff are in the league (and they do it very well). But that
contribution is hard to measure. I feel like my stuff captures it a
little better, if just because it doesn't combine offense and defense.
But it is really hard to say how many points per game Mutombo or
Ratliff defense is worth. And, frankly, defense is due a lot to
Also, I would add that I have tried hard and will continue to try hard
to determine the trade-off between "production" and "efficiency". I
have looked to see whether increased player production means lower
efficiency. We believe it must happen, but it's hard to see. For
instance, I would have expected David Robinson's efficiency to be
helped by not having to score as much due to Duncan's presence.
Hasn't been the case. Ron Harper, on the other hand, was made more
efficient by playing with Jordan. Heck, I think everyone who played
with Jordan got more efficient -- in part because they took smarter
shots or only smart shots. But it has really been hard to see the
effect with stars or even pretty good players. Shouldn't every player
shoot a little better and pass a little better when they're surrounded
by better players? I'd like to show that "efficiency" rise with the
simultaneous "production" decline, but it is hard to do. My methods
are complex, but if someone wants to just look at FG% as a function of
% of team shots while on the court -- useful stuff. (Hmm, maybe
Shawn Kemp was an example of getting less efficient because of losing
his good teammates. But Baker didn't exactly get better by going to
Journal of Basketball Studies