Top values on new metric called "Individual Player Values"

[talkingpractice]

Numbers for the 2014 season are up (through last night's games) for the top players in the league. This year, the number we're posting is solely an in-season number (no information is used from any prior seasons). So it's our SPM used as a prior for RAPM, using only 2013-2014 data. Just whole coefs and no off/def split (the splits mean less and less every year, anyway).

Rank Player IPV
1 Kevin Love 7.4
2 LeBron 5.5
3 Stephen Curry 5.1
4 Paul George 4.9
5 Blake Griffin 4.8
6 James Harden 4.4
7 Roy Hibbert 4.3
8 Dirk Nowitzki 4.3
9 Kevin Durant 4.1
10 Amir Johnson 4.0

[Crow]

Thanks for the small peek. Of course there is interest in the numbers for all players, if you are willing later. As you wish.

[AcrossTheCourt]

Why don't you have a prior at least for defense?

Where does Anthony Davis rank, by the way?

[bchaikin]

Where does Anthony Davis rank, by the way?

i'd be interested in knowing also, because what he's done so far is truly unprecedented. only 2 other players (hakeem olajuwon, david robinson) ever had seasons with 5.0+ ST/100min, 5.0%+ BS, and 14.0+ reb/48min (300+ minutes played), and davis is doing it while scoring 21 pts/g on a 55% ScFG% while committing just 1.3 TO/g...

[talkingpractice]

Davis is 15th (here are the next 5). I agree he's low. His rapm stats in that whopping 12 games or whatever are good but not great. The team he plays on is ofc a large part of that, and nearly everyone above him is on a (often much) better team.

Tony Parker 4.0
Jimmy Butler 4.0
Chris Paul 3.9
Isaiah Thomas 3.9
Anthony Davis 3.7

ATC we use all sorts of priors in our internal stuff, this version is just the one we're putting out there this year (in-season only). This version is solely descriptive, ie what has happened thus far, in this particular season.

The way Amir Johnson sticks like a magnet to the top of any sort of RAPM table, regardless of season/prior/type/universe, is uncanny. This is 2014 data only, not even through one month of the season, and hes right up near the top on the pure rapm stuff already. And fighting for minutes with Tyler Hansbrough.

[AcrossTheCourt]

Though you'd agree without a prior for defense it's pretty wonky, right? Are you using any non-traditional stats (things outside of the box score) for the statistical RAPM component?

And if I had to choose one more guy to peek at his IPV right now, that'd be Howard. Or maybe Aldridge or Marc Gasol.

[jbrocato23]

Though you'd agree without a prior for defense it's pretty wonky, right? Are you using any non-traditional stats (things outside of the box score) for the statistical RAPM component?

And if I had to choose one more guy to peek at his IPV right now, that'd be Howard. Or maybe Aldridge or Marc Gasol.

ATC, a bit off topic but I just saw that you put together npi rapm for '97. Did you make a post here already? If not, you should. Really cool stuff.

[talkingpractice]

Though you'd agree without a prior for defense it's pretty wonky, right?

Idunno. It is what it is, I guess? I surely agree that the numbers (particularly the plus/minus component) aren't as reliable after 14 games as they'll be after 40 games.

One caveat is that there is indeed a prior for defense. There's no *prior season* info used as a prior in the numbers we're displaying. But our in-season SPM coefs for the players are used as a prior for RAPM to create these values.

[talkingpractice]

Through last nights games (updated on our site), ie about 20% of the season. Again, it's 2014-only spm (whole coef, not off/def split) used as a prior for 2014-only rapm, no bells/whistles or prior year info used.

Rank Player IPV
1 Kevin Love 6.9
2 Paul George 5.9
3 Kevin Durant 5.1
4 Blake Griffin 4.8
5 LeBron 4.6
6 Dirk Nowitzki 4.6
7 Roy Hibbert 4.6
8 Chris Paul 4.6
9 Stephen Curry 4.5
10 Channing Frye 3.9

ATC -> Howard is around a 3, and he's around a 0 on pure npi rapm right now.

Interesting players with ultrahigh RAPM (npi) thus far = Redick, Frye, Reggie Jackson, Casspi. Interesting players with ultralow RAPM (npi) thus far = Mayo, Amare, Jack, Garnett.

We wont bump this thread going fw (will just autoupdate the site every few days).

[AcrossTheCourt]

Looks like Amir Johnson lost his starting job to Tyler Hansbrough. And last season he sat on the bench when Bargnani was healthy, although that was partly an organization trying to justify a number one pick. He's probably the best current representation of the no-stats all-star (although he does pick up a lot of stats) but he's rarely talked about in mainstream NBA circles.

[Bobbofitos]

Looks like Amir Johnson lost his starting job to Tyler Hansbrough. And last season he sat on the bench when Bargnani was healthy, although that was partly an organization trying to justify a number one pick. He's probably the best current representation of the no-stats all-star (although he does pick up a lot of stats) but he's rarely talked about in mainstream NBA circles.

Amir, it appears, has been hated by coaches for his entire career. Makes no sense to me, as he produces very well when he plays.

[talkingpractice]

Looks like Amir Johnson lost his starting job to Tyler Hansbrough. And last season he sat on the bench when Bargnani was healthy, although that was partly an organization trying to justify a number one pick. He's probably the best current representation of the no-stats all-star (although he does pick up a lot of stats) but he's rarely talked about in mainstream NBA circles.

To be fair to Casey, Amir's numbers have indeed fallen off over the past bunch of games, including his rapm numbers (now "down" to the 3's.... and his IPV is now down to the 2's). That said, this lineup change is of course hilarious. I dont think Casey appreciates a mean screen.

[HoopDon]

Does anybody have the IPV values with OFF/DEF splits for the 2013 regular season? I was under the assumption they used to be posted, but were taken down.

Also, would those values be calculated differently, then the current IPV being released in tid-bits for the 2013-2014 season?

Hoopdon.weebly.com

[talkingpractice]

Also, would those values be calculated differently, then the current IPV being released in tid-bits for the 2013-2014 season?

Hoopdon.weebly.com

Yeah, last years posted numbers were a best estimate of player value, involving prior year information, an aging curve, mean reversion, etc. This year the numbers we're giving our are an in-season prior-informed rapm model using our spm as prior, whole coefs only (no off/def). So think of last years posted values as predictive, and this years posted values as descriptive. I 'think' the latter provides more value to the community in terms of what is/isnt out there right now in the public domain.

[Mike G]

Looks like Amir Johnson lost his starting job to Tyler Hansbrough. And last season he sat on the bench when Bargnani was healthy, although that was partly an organization trying to justify a number one pick. He's probably the best current representation of the no-stats all-star (although he does pick up a lot of stats) but he's rarely talked about in mainstream NBA circles.

Amir, it appears, has been hated by coaches for his entire career. Makes no sense to me, as he produces very well when he plays.

Amir is a big enigma. He's always had better numbers off the bench.
http://www.basketball-reference.com/pla ... 01/splits/

For his career as a starter, his ORtg and DRtg are 115-106. As a reserve, they're 119-105.
Last year, he started 38 of 81 games. His O-D were 111-103 as a starter, and 120-105 off the bench.

Most years, he gets a lot more OReb and blocks (per minute) off the bench, and he shoots better.