APBRmetrics

deepak wrote:Are you computing offensive and defensive ratings separately, and then adding together for the final IPV rating? If so, could you also post the offensive and defensive ratings?

We do compute them separately, but for now we're keeping that part proprietary (sorry), and just publishing the total values.

Updated the numbers today.

It seems that there hasn't been an update to http://stats-for-the-nba.appspot.com/ in a while now...

No updates in over two weeks. I think it's looking like JE got picked up.

AND 2013 data is down.

Shoulda screenshotted that junk.

bbstats wrote:AND 2013 data is down.

Shoulda screenshotted that junk.

I've got it all... in case he is or has been hired.

v-zero wrote:No updates in over two weeks. I think it's looking like JE got picked up.

If this turns out to be true, then we'll update IPV every 1-2 days so that there's a rapm-based model out there in the public domain in the interim. I know that ours isn't open source and hence not as valuable to you guys, but hopefully it could be a tad of a stopgap until someone else puts a pure rapm out there.

If it's permanently down, I'll make an effort to get RAPM and A4PM up in the next couple of months. I have the data to do it. Just a matter of finding the time.

talkingpractice wrote: If this turns out to be true, then we'll update IPV every 1-2 days so that there's a rapm-based model out there in the public domain in the interim. I know that ours isn't open source and hence not as valuable to you guys, but hopefully it could be a tad of a stopgap until someone else puts a pure rapm out there.

Would you consider releasing off/def splits?

rshea wrote: Would you consider releasing off/def splits?

We're discussing this over the next few days.

Numbers updated through last night's games. We also added a tad more detail on the model since many people were asking. As of now, we think it's not a good idea to give the offense vs defense splits, but we're still discussing it and it's reasonably possible that we put up the splits after the break. From now on (I promise this time), we're going to update every 2-3 days.

Briefly, here's everyone we have over 2.0 on D as of tonight, in order from highest downwards.

• Kevin Garnett
  Tim Duncan
  Dwight Howard
  Omer Asik
  Andre Iguodala
  Larry Sanders
  Josh Smith
  Joakim Noah
  Taj Gibson
  Paul George
  Marc Gasol
  Tony Allen
  Nene Hilario
  Gerald Wallace
  Tyson Chandler
  Ekpe Udoh
  Amir Johnson
  Serge Ibaka
  Kosta Koufos
  Paul Millsap
  Luol Deng

Updated through the Break.

Additionally, we are considering adding one more person to our analytical team. Our models are based primarily on Bayesian methods and/or statistical signal processing. Experts in either discipline who may be interested in joining our team can send a resume and CV to the email address on our site. Applicants should be proficient in R, OpenBUGS, MySQL, and Python.

Thanks for the update. I do hope at some point you're able to post the offense/defense splits. There is a lack of good defensive metrics available, and especially with xRAPM out of action yours would be a great stat to refer to.

Dirk is a +4 on IPV despite the lowest TS% and EFG% of his career after his rookie season. He got an .098 on winshares /48.

Nene at +3.

Amare a tiny fraction below neutral.

Jeff Green still -3 and in the bottom 20%. Austin Rivers in bottom 3%.

talkingpractice wrote:Numbers updated through last night's games. We also added a tad more detail on the model since many people were asking. As of now, we think it's not a good idea to give the offense vs defense splits, but we're still discussing it and it's reasonably possible that we put up the splits after the break. From now on (I promise this time), we're going to update every 2-3 days.

Briefly, here's everyone we have over 2.0 on D as of tonight, in order from highest downwards.

• Kevin Garnett
  Tim Duncan
  Dwight Howard
  Omer Asik
  Andre Iguodala
  Larry Sanders
  Josh Smith
  Joakim Noah
  Taj Gibson
  Paul George
  Marc Gasol
  Tony Allen
  Nene Hilario
  Gerald Wallace
  Tyson Chandler
  Ekpe Udoh
  Amir Johnson
  Serge Ibaka
  Kosta Koufos
  Paul Millsap
  Luol Deng

For comparison here is everybody I have above 2 on D on my (freshly baked) SEBF V2 model:

Tim Duncan
Larry Sanders
Kevin Garnett
Marcus Camby
Greg Stiemsma
Andre Drummond
Joakim Noah
Dwight Howard
Josh Smith
Jermaine O'Neal
Paul George
Emeka Okafor
Ivan Johnson
Lamar Odom
Taj Gibson
Nene Hilario
Roy Hibbert
Marc Gasol

Crow wrote:Dirk is a +4 on IPV despite the lowest TS% and EFG% of his career after his rookie season. He got an .098 on winshares /48.

I imagine that's due to his prior. I have Dirk as neutral defensively, and -0.7 offensively. That's mostly due to his crap performance this season, plus his prior in my model wasn't that great to begin with.

v-zero wrote:For comparison here is everybody I have above 2 on D on my (freshly baked) SEBF V2 model:

Tim Duncan
Larry Sanders
Kevin Garnett
Marcus Camby
Greg Stiemsma
Andre Drummond
Joakim Noah
Dwight Howard
Josh Smith
Jermaine O'Neal
Paul George
Emeka Okafor
Ivan Johnson
Lamar Odom
Taj Gibson
Nene Hilario
Roy Hibbert
Marc Gasol

How do you rate Chandler, Iguodala, and Asik defensively?