2015 APM/RAPM/etc.

[colts18]

Shaq seems to be the only center who's a clear positive on offence.

Yes. This is a great article that discusses it:

http://nyloncalculus.com/2015/06/24/towns-okafor-and-the-great-modern-center-debate/?utm_medium=twitter&utm_source=twitterfeed

Great excerpt

Turning to more RAPM data, this time provided by Layne Vashro, there were 121 seasons with an offensive RAPM of +5 or greater from 2007 to 2015. But only 14 of those seasons were from power forwards or centers. Referencing the players below, it’s a list comprised mainly of Dirk and other jump shooters with some exceptions. (McRoberts and Jamison both have three-point range for those who have forgotten.) Elton Brand wasn’t exclusively a low-post player, and he used his midrange shot heavily with the Clippers at his peak. Drummond has no range, but he doesn’t have much of a post game either; he’s no Okafor. He’s an athletic finisher, and some of those guys like DeAndre Jordan or Chandler look pretty good in RAPM because of their abilities in the pick and roll. And Shaq is, well, Shaq — he’s singular.
Season Player ORAPM (100 poss.)
2012 Dirk Nowitzki 7.9
2011 Dirk Nowitzki 7.7
2015 Dirk Nowitzki 7.4
2007 Antawn Jamison 7.1
2013 Channing Frye 6.9
2007 Elton Brand 5.9
2012 Channing Frye 5.5
2007 Dirk Nowitzki 5.4
2013 Dirk Nowitzki 5.3
2008 Antawn Jamison 5.1
2013 Andre Drummond 5.1
2010 Dirk Nowitzki 5.1
2015 Josh McRoberts 5.0
2007 Shaquille O’Neal 5.0

That equates to 11 out of the top 14 players on offense having three-point range, but the rate is much lower for everyone else. Depending on how you define position and three-point range, the rate is around 17% (using my own position data over the past decade.) Using Synergy data again, the top low-post big men by volume are in the first table below. Compare that with the second table: the three-point shooters are, again, outclassing them. Note that the top big men in the post-up table are the ones with guard-like skills with Aldridge’s jump-shooting and Griffin’s playmaking.

Table: top post up-big men 2015
Player Post-up plays ORAPM
Al Jefferson 659 -0.31
LaMarcus Aldridge 618 2.51
Marc Gasol 533 -0.29
Zach Randolph 511 1.44
Greg Monroe 478 0.29
DeMarcus Cousins 468 -0.45
Nikola Vucevic 430 -1.54
Pau Gasol 420 0.66
Roy Hibbert 418 -1.13
Blake Griffin 397 2.11

Table: top 3PA-big men 2015
Player 3PA ORAPM
Kevin Love 392 2.38
Ryan Anderson 359 2.48
Channing Frye 346 1.38
Draymond Green 329 1.88
Nikola Mirotic 313 1.20
Paul Pierce 303 3.65
Anthony Tolliver 290 0.78
Patrick Patterson 283 1.27
Dirk Nowitzki 274 5.21
Marvin Williams 265 -0.45

[Keller61]

Yeah, again I'm coming to not trusting the NPI very much. Seeing massive swings in 2 versions of the same stat is not really giving me much assurance there's too much validity. Just an example using 2 that really stood out:
JE has Manu at 2.64, Westbrook 2.96

Evan has Manu at 8.22, Westbrook at 1.00

So, yeah.... :dontknow:

The more I learn about RAPM, the more I think there's a lot of randomness to it. You're essentially trying to make a "best fit" equation with hundreds of variables. Some of them are bound to be way off.

[bondom34]

Yeah, again I'm coming to not trusting the NPI very much. Seeing massive swings in 2 versions of the same stat is not really giving me much assurance there's too much validity. Just an example using 2 that really stood out:
JE has Manu at 2.64, Westbrook 2.96

Evan has Manu at 8.22, Westbrook at 1.00

So, yeah.... :dontknow:

The more I learn about RAPM, the more I think there's a lot of randomness to it. You're essentially trying to make a "best fit" equation with hundreds of variables. Some of them are bound to be way off.

Exactly how I am/was. I got into it, loved it, then kept learning, and the more I learn, the less I really like and feel work.

[SideshowBob]

JE just posted 02-15 multi-year sets (which he says essentially work in place of PI RAPM). Lower and lower weight given to older seasons.

Link

[RSCD3_]

JE just posted 02-15 multi-year sets (which he says essentially work in place of PI RAPM). Lower and lower weight given to older seasons.

Link

The link is broken sir

[SideshowBob]

JE just posted 02-15 multi-year sets (which he says essentially work in place of PI RAPM). Lower and lower weight given to older seasons.

Link

The link is broken sir

Thanks. Fixed it

[RSCD3_]

I find it unsettling that the study points to 11 lebron as his peak and has 13 as his seventh best season hmmmm

[Dr Spaceman]

Honestly never watched a ton of him either, so I suppose that's part of it too, but better than Wade....

Fun fact: Steve Nash says Baron Davis was the most difficult player he guarded in his career

May I ask what you think of Baron and how he compares to other point guards (like Westbrook for example)?

Just saw this now.

I like Davis’ game a lot. He was a much more effective offensive player than the stats would indicate. He had a smooth dribble game kind of reminiscent of Irving. Could shoot 3s off the bounce and while he wasn’t a terrifying specimen he was crazy strong which allowed him to get good looks in the 3-7 foot range despite him not being able to get all the way to the rim in the half court. Westy is a pretty good comp, although I think Baron was much more willing to scale down his own scoring and he was a decent playmaker. I think he’d be debatably better than anyone but Paul and Curry as a PG.

[SideshowBob]

I find it unsettling that the study points to 11 lebron as his peak and has 13 as his seventh best season hmmmm

Actually its pretty understandable, given the weight put on prior years. 09/10 is the best two-year stretch of Lebron's career, and 2011 probably gives those years the greatest weight. 2013 on the other hand follows up on 11 and 12, with 11 being by far the weakest year of his prime and 12 being 2nd or 3rd weakest IMO. If I just take a three year average (n, n-1, n-2) for each year of my own ratings, I get similarly wonky results.

This is why I hoped we'd get a PI RAPM instead. I don't quite understand how a PI set would be essentially the same as a multi-year set, that doesn't really make sense to me. There is no weighting of past seasons there, its more just that the priors act as corroborating data, whereas here we've literally expanded the data.

[Dipper 13]

It would be nice to see the prior informed RAPM for 2001, and to what extent (if at all) it would affect all the prior informed data from 2001-2015.

Can the 2001 data be created using the 2000 PI data?

http://ascreamingcomesacrossthecourt.blogspot.com/2014/03/2000-rapm-non-prior-and-prior-informed.html

[lorak]

JE just posted 02-15 multi-year sets (which he says essentially work in place of PI RAPM). Lower and lower weight given to older seasons.

Link

What "multiyear" does mean here? For 2002 it's only 2001, but for 2003 it's 2001+2002 and so on to 2015 where prior is everything from 2001 to 2014? If so doesn't it make previous seasons less accurate (because of less prior years used)?

[SideshowBob]

JE just posted 02-15 multi-year sets (which he says essentially work in place of PI RAPM). Lower and lower weight given to older seasons.

Link

What "multiyear" does mean here? For 2002 it's only 2001, but for 2003 it's 2001+2002 and so on to 2015 where prior is everything from 2001 to 2014? If so doesn't it make previous seasons less accurate (because of less prior years used)?

Yeah I would assume so, but this would have been an issue with the old PI RAPM as well. I don't know if I'd say "less accurate" because the weight of the older seasons is probably minimal. I doubt there's much influence moving the needle beyond the 2 prior years.

Also, JE noted that he loosened the spread on this set relative to the old multiyear ones.

[SideshowBob]

JE's notes

Some observations:

2002
- The Lakers of early 2000 are mostly seen as Shaq/Kobe's team, yet Derek Fisher is actually rated higher than Kobe
- Steve Francis, who many back then saw as a mindless chucker (that's at least how I remember it) ranks in the top 10
- Horace Grant, at age 36, ranks in the top 30. Stockton ranks 3rd at age 39

2003
- Duncan with with one of the highest ratings (in pure RAPM) ever, at 9.1. He was certainly seen as a 2-way player by RAPM, ranking 6th for offense, 1st for defense. Soon, his offense started to decline, but his defense has remained very good

2004
- KG with the highest rating in those years analysed with a +10, ranking 2nd on offense and 3rd on defense. The TWolves were playing like an elite team with him ON the court: +10, and like a lottery team with him OFF: -11
- A couple of not-so-well-knows/surprises near the top: Bo Outlaw, Najera, Eric Williams, Jason Collins, Desmond Mason, Derek Anderson, Brian Cardinal

2006
- Manu Ginobili comes in 1st for a change (from KG, Duncan). Ron Artest is right behind him
- Kirilenko 3rd. He had been near the top for many years in a row
- This year's finals MVP, Iguodala, comes in ~25th

2007
- Ginobili, again, 1st
- Nash near the top for the ~3rd year in a row, just due to his elite offense (his defense is slightly below average)
- Arenas makes an appearance in the top 20

2008
- KG's 1st year in Boston brings him back into form

2010
- The age of LeBron as undisputed #1 starts

2012
- Ekpeh Udoh with a surprise appearance in the top 10. His +/- numbers have since fallen off a cliff

2013
- Nowitzki takes the #1 spot from LeBron for a brief moment
- Chris Paul in the top 5
- Nick Collison in the top 10
- Asik in the top 15

2014
- Iguodala 3rd, Curry 4th

A note on 2015: these are a tiny bit "wider" than the ones I posted earlier. This is because I used a lower penalization value here (2500 v 3000), which I remembered gives slightly better OOS projections when there's more than 1 season of data. It's certainly not a big difference, by any means

[lorak]

Top 5 DRAPM seasons - all KG.
Nash 5 seasons in top 8 ORAPM, including all in top 3. LeBron 6 times in top 13.

EDIT
How good was John Stockton? So good that even when he didn't play in the NBA his RAPM was positive: +2.2 in 2004 and +0.4 in 2005

[WhateverBro]

Top 5 DRAPM seasons - all KG.
Nash 5 seasons in top 8 ORAPM, including all in top 3. LeBron 6 times in top 13.

EDIT
How good was John Stockton? So good that even when he didn't play in the NBA his RAPM was positive: +2.2 in 2004 and +0.4 in 2005

That post-career impact.