Advanced Statistical Plus/Minus on Website

[DSMok1]

I have now posted ASPM numbers for the current season on my website, at http://godismyjudgeok.com/DStats/aspm-a ... 2012-aspm/

I'll try to update them regularly...

I have revised and simplified the ASPM formula somewhat while still maintaining high R^2 onto 8 year RAPM--I'll write up the latest (and simplest) formula shortly. The Excel spreadsheet linked to on my site also has the latest formula.

[Crow]

Thanks, I'll give it a look.


P.S. I assume you saw this in the BPI thread from bbstats:

"Note to self: convince Daniel to include efficiency * minutes% to improve SPM's R^2 against rAPM to create world's best power rating system..."

P.P.S. I see few names I recognize in the college top 50 VORP list. All stats in college are not the same of course in the eyes of NBA evaluators and probably shouldn't be. A SOS adjusted VORP might be an interesting additional product (since college RAPM seems unlikely or unlikely to be that accurate).

[DSMok1]

ASPM and VORP are already SoS adjusted. Many of the top players in college are already seniors (have maxed out potential much more) or simply have already maximized their (limited compared to NBA standards) athletic ability.

As to the "efficiency * minutes% to improve SPM's R^2 against rAPM ", I'll ask Nathan what he means.

[Crow]

"ASPM and VORP are already SoS adjusted"

Ok, thanks for the clarification.

[Statman]

Thanks, I'll give it a look.


P.S. I assume you saw this in the BPI thread from bbstats:

"Note to self: convince Daniel to include efficiency * minutes% to improve SPM's R^2 against rAPM to create world's best power rating system..."

P.P.S. I see few names I recognize in the college top 50 VORP list. All stats in college are not the same of course in the eyes of NBA evaluators and probably shouldn't be. A SOS adjusted VORP might be an interesting additional product (since college RAPM seems unlikely or unlikely to be that accurate).

Where is this college VORP list?

[xkonk]

Where is this college VORP list?

http://godismyjudgeok.com/DStats/aspm-and-vorp/

[Statman]

Where is this college VORP list?

http://godismyjudgeok.com/DStats/aspm-and-vorp/

Thanks! Great stuff - plus the charts & such are so pretty....

[bbstats]

Daniel,

If you want to increase the R^2 value of any SPM against APM, doing Minutes%*TeamEffMgn increases it significantly on a year-by-year basis. Once you did that, your team adjustments would probably be smaller and therefore more intuitive/parsimonious.

Just a thought.

[DSMok1]

Daniel,

If you want to increase the R^2 value of any SPM against APM, doing Minutes%*TeamEffMgn increases it significantly on a year-by-year basis. Once you did that, your team adjustments would probably be smaller and therefore more intuitive/parsimonious.

Just a thought.

Right now, I have a MPG term, and then a team eff margin number added or subtracted at the end to adjust for opposition/stuff not measured. What exactly does the min%*TeamEff term do? I guess I could throw it into the regression and see if it would help my R^2.

[DSMok1]

I have finally gotten around to writing up the background and derivation of ASPM on my website, here: http://godismyjudgeok.com/DStats/aspm-and-vorp/

[bbstats]

Thanks for the writeup Daniel, it's pretty insightful! I wish more people took approaches like this.

My only real comment:
Why is AST% inside the usage% * () term? BB-ref's Usage% only includes FGA, FTA, and TOV. Unless you're using a different usage%, of course.

EDIT: NEVERMIND. I am a bad reader.

[DSMok1]

I just added an interactive graph of ASPM vs. RAPM for the dataset used. http://godismyjudgeok.com/DStats/aspm-and-vorp/

[deepak]

This is really cool. A small correction: Chuck Hayes should be listed as a 4 or 5. He is not a 3.

[huevonkiller]

http://godismyjudgeok.com/DStats/aspm-and-vorp/

I like the explanation in this link. Your graphs are very cool.

[DSMok1]

This is really cool. A small correction: Chuck Hayes should be listed as a 4 or 5. He is not a 3.

I generated a system to estimate the position a player plays based on height, weight, and statistics--I couldn't find an easy database that actually listed all of the player's positions, beyond Basketball-Reference's "G, F, C" nomenclature. Some people will be misidentified because of that. If you've got a better source, let me know.