APBRmetrics

http://stats-for-the-nba.appspot.com/to11

Lambda was 2000 for offense and defense. Might be lower when using more years.

I haven't tested whether adjusted or individual numbers predict team turnovers better. I would suppose a mix of both is best, as it was with team rebounding.

The code that creates the turnover matchup files from the PBPs might not be 100% correct, but should be close.

Chris Paul towers above everyone.
Tony Allen being a clear #1 on defense was kind of expected.
Garnett is one of the best at causing turnovers, while also being the clear #1 on adjusted opponent points per shot. He's just a defensive beast.
Deron Williams is #1 on offense.
Amir Johnson is at #4 overall, and he also ranks quite well on adjusted opponent points per shot. He's 24 years old. Things could get crazy once he plays for a team that is average without him.
Radmanovic is one of the best at causing turnovers, I did not expect this. Evan, you have any idea why he could rank that high?
Glen Davis looks good on defense. Wasn't he the leader in charges taken?
Rose looks bad at causing turnovers, Asik ranks pretty well
Aaron Brooks is dead last at causing opponent turnovers. No idea why
Other than him, lots of big guys at the bottom

Multiyear coming tomorrow

Excellent!

I look forward to the multi-year.

J.E. wrote: Radmanovic is one of the best at causing turnovers, I did not expect this. Evan, you have any idea why he could rank that high?

Hustle? Contract year?

Seriously, though, this is really interesting. In fact...if anyone remembers my post on "non-steal forced turnovers" inspired by Bob Chaikin's request in a probably now deleted thread a while back ("unrecorded useful stats"):

http://thecity2.com/2011/01/29/non-stea ... turnovers/

I later found that there was no year-to-year correlation, though. So the multi-year will be interesting.

The thread you refer to was recovered and currently sits on page 7.

http://sonicscentral.com/apbrmetrics/vi ... ?f=2&t=149





Thanks for this set of Factor data.

3 year analysis is online

http://stats-for-the-nba.appspot.com/to3yr

Slow, heavy big men fill out the bottom

Doing 2009/2010 now to see y-t-y correlation

Chris Paul is simply phenomenal

y-t-y correlation from 2009/2010 to 2010/2011 is 0.43 for players with >2000 possessions total (279 players)

That's R or R^2?

R

Thanks. I wonder if you only look at players that switched teams, how much lower the correlation becomes. Actually, that would be useful to know for all these adjusted factors. Have you thought about doing that?

EvanZ wrote:Thanks. I wonder if you only look at players that switched teams, how much lower the correlation becomes. Actually, that would be useful to know for all these adjusted factors. Have you thought about doing that?

I think I don't care enough about y-t-y correlation to look up everybody that has changed teams over the years

I put this back online with data from this season. See the full list here

It's multiyear with older seasons getting less weight, no priors used

Again, lots of heavy big men at the bottom. "Surprises" (non-heavy + non big) at the bottom are Goodwin, Muhammad and Schroeder

Lots of fast guards at the top, with Mike Conley leading the way. Oladipo, Kentavious Caldwell-Pope and Michael Carter-Williams looking good on the defensive side

Thanks. A full of the 4 RAPM factors would be really great. For players and coaches too.

Free Throw Shooting is included in "adjusted PPS", so I basically rolled 2 factors into one and I'm going to keep it that way

Adjusted Rebounding is coming soon.

Adjusted 4(3) Factors for coaches could be cool, but I think I have more interesting topics to follow for quite a while. I might do an aging curve for "adjusted PPS" though

Sounds good. Adjusted PPS works.

I will probably post something about Phil Jackson and the 4 factors shortly using raw data.