APBRmetrics

http://stats-for-the-nba.appspot.com/PBP/2001np.html
http://stats-for-the-nba.appspot.com/PBP/2002np.html
http://stats-for-the-nba.appspot.com/PBP/2003np.html
http://stats-for-the-nba.appspot.com/PBP/2004np.html
http://stats-for-the-nba.appspot.com/PBP/2005np.html
http://stats-for-the-nba.appspot.com/PBP/2006np.html
http://stats-for-the-nba.appspot.com/PBP/2007np.html
http://stats-for-the-nba.appspot.com/PBP/2008np.html
http://stats-for-the-nba.appspot.com/PBP/2009np.html
http://stats-for-the-nba.appspot.com/PBP/2010np.html
http://stats-for-the-nba.appspot.com/PBP/2011np.html
http://stats-for-the-nba.appspot.com/PBP/2012np.html

Each file with year number X includes data for all years <= X. All years equally weighted. No playoffs. Lambda = 3000

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

Garnett on prior-informed RAPM for 2011-12 with playoffs: +5.6.
Garnett on non prior-informed RAPM for 2011-12: +3.0.

Garnett on newest RAPM: +9.2. No playoffs, prior-informed.

Elite on all but these are big differences in the estimates.

It appears Garnett performed a lot less well this season than previously and less well in the playoffs.


What if any advice would you give on how to compare the with playoffs to without? Anything further you can say about the approximate size of the average errors for each?

Westbrook without playoffs, estimated at near +4. With playoffs, a bit above +2. Does that suggest in the playoffs he was probably estimated neutral to negative, given the big difference in minutes between regular season and playoffs?

Taj Gibson

on prior-informed RAPM for 2011-12 with playoffs: +4.0 but in 16th place.
on non prior-informed RAPM for 2011-12: +4.3 and in first place.
on newest RAPM (No playoffs, prior-informed): +1.2 and around 135th place?

Was he off the charts fantastic in the playoffs or is there more to this?


Durant on new RAPM at +0.5?

Crow, there never was prior informed 2012 RAPM with playoffs.
This RAPM here isn't 'prior informed', it's just multiyear RAPM. If the filename says 20XY, then all years <=XY were used. So, 2004 uses 2001+2002+2003+2004 etc

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

shawn bradley appears to have one of the highest defensive ratings each year from 2001-2004. when you look at your ratings for all of 2001-2004 combined, does he have the best defensive rating?

In the 2004 file, he is 2nd best to Mutumbo by 0.1.

Thanks for the explanation. I misinterpreted, reacting to the explicit reference to "no playoffs" for this new version and wrongly inferring that the old ones had playoffs. Aaron's APM is with playoffs and I was apparently thinking about that too. So throw out what I said about with / without playoffs.

Garnett still got less good last season and Gibson still got a lot better. And Durant was estimated as bad his first 2 years as he was good thereafter.

Any intention to plot this data over time by player, similar to what you've done for the prior-informed data?

I don't think this is too useful for anything other than SPM, so no

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

Shaq #1 every year from 01-08. He must have been really good even past his prime.

Why do those values change significantly? Stockton was 6.1 in 2004 and 5.3 in 2012 despite not playing a game since then?


-JE, is it possible you do a 4 factors RAPM study or at least one for rebounding? Does your lineup data have that data in it or is it just the scores?

I suppose the players he played with when he was playing had some good years after he stopped playing. That would lower his rating
It's not in the lineup data but in the PBP. So, certainly possible, but why not use http://www.d3coder.com/thecity/a4pm/ ?

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

Jeremias, sorry if I missed the fine print somewhere, but is there any explanation for why the distribution of results at the top end for the non-prior-informed 2002 (http://stats-for-the-nba.appspot.com/ranking02), where Shaq is at the top of the table at 3.3 and only one other player is above 3.0, is so different from the non-prior-informed 2001, where Shaq tops things at 6.3 and sixteen players top 3.0?

Is there, in fact, a prior for 2001?

The old version contained data from ESPN only and was missing a lot of games.

The multiyear ratings listed here and the non-informed ratings I posted recently use data from bbr instead. The bbr data contains more games and less weird errors in the play by play files

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

Does anyone have RAPM data going back to 1989?

I'm working on a small project and was hoping to get access to those numbers somehow just for the Orlando Magic. I'm assuming it's not readily available anywhere, but would it take long to churn out the RAPMs of players in Magic franchise history all-time and/or single-season (1989-present)?

If it's possible, I'd be greatly appreciative of the help.