most improved players of 2011-12

[EvanZ]

Does this display that the 3 best players in the league are Beno, Manu, and Bonner?

It says they are the 3 players with the highest A4PM ratings on that plot.

Since all 3 had less than or not much more than 2000 possessions, one might assume they would eventually regress to the mean by some amount. Although it should be noted that Bonner has had consistently high RAPM ratings for several seasons now. And I think we would all agree Manu is one of the best players in the NBA, no?

[Mike G]

Mostly I just liked saying "Beno, Manu, and Bonner".
But I wondered if something was awry with the Bucks. I had Beno Udrih as their most diminished player, his e484 down from .93 last year to .61 this year, a loss of .69 eWins. Note that his minutes were also cut in half.
You might (correct if wrong) run a diagonal from the lower left corner to the upper right, indicating players that were just as effective in both seasons.
EDIT: someone edited my post, and I've un-edited it.

[EvanZ]

Beno's shooting was down 100 points. I do find that odd. Maybe his +/- is high because he played so much with Dunleavy? I don't know.

[Crow]

I wonder what the average tenure of players is in Milwaukee compared to league average in the last 5-10 years. I have a hunch it is less than average and might be one of the lowest in the league but haven't crunched the data. I wonder how much of that is because of the GM vs the owner. I would speculate that the owner is an active part of that churn.

[Mike G]

Changes in A4PM and RAPM (calculated by me) from 2011 to 2012:



Delta is A4PM12-A4PM11.

Evan, your list -- ranked by A4PM increase -- is topped by Goran Dragic, among players with [what looks like] at least 2500 possessions this year and also 1250+ last year.
His A4PM went from -3.19 to +.48, a net gain of 3.67
At the same time, his RAPM dropped from 1.86 to -.05, a loss of 1.91

What is so different about these measures, that some players could get such conflicting results?
In general, there seems to be a pretty good correlation.

[EvanZ]

What is so different about these measures, that some players could get such conflicting results?
In general, there seems to be a pretty good correlation.

One idea might be that some stats have more variation than others, and A4PM adds another level of smoothing in addition to the regularization. Or maybe it introduces more noise and is worse. I don't know.

I should take both sets of ratings and do a retrodiction based on actual possessions played this season, and see which one, if either, has a significant predictive advantage.