APBRmetrics

So, I think the MIP discussions are superannoying, since in almost all cases, the increased observed performance for the player is much more about context than about improvement. Hence, the MIS Award (Most Improved Situation) should go to lots of the people I've been seeing on most MIP lists. To try to separate the two, we took all players and calced their IPV minus PRIOR. Here's the 10 big deltas ->
We then wave a pen and decide qualitatively as to the few who may deserve the big fat MIP designation, vs the many guys with a very improved situation. Clearly, Big Al owes a lot of thanks to Clifford for his "improvement" (due to the defensive scheme he put in largely to deal with Al's deficiencies), and Dragic/Frye owe a lot of their improvement to Hornacek, as well as to having each other as pnr/pnp partners. DeAndre owes a lot to Doc, and Mills sorta doesn't count due to Pop. So that leaves the below as the 5 actual Most Improved Players, once you remove all the guys who primarily entered much better situations ->

NAME IMPROVEMENT (IPV-PRIOR)
DeMarcus Cousins +3.95
Kemba Walker +3.42
Markieff Morris +2.45
Reggie Jackson +2.24
Timofey Mozgov +2.23

This matches my eyes, which think the answer should be a major motion picture called Boogie by a Mile.

Sorry for the formatting today, and not clipping off all the extra decimals, etc.

EDIT: fixed formatting. http://www.sensefulsolutions.com/2010/1 ... table.html and the CODE tag is your friend -DSMok1

No Stephen Curry?

Mike G reads minds lol. We had a prior of +3.4 for Steph, and IPV this year of +5.4. He'd be next on the list if I had went one guy deeper.

DSMok1 that formatting rocks, and I will ofc now use this going forward .

TP, is there any way that you could evaluate the deltas that you calculated versus the jump you'd expect from each player? For instance, Cousins was in his 4th season and 23 years old and Dragic was in his 6th season at 27 years old. Is there a way you could spit out a benchmark for how much you'd expect each player to improve based on their age/experience? Maybe career minutes played would also be an appropriate way to do this.

There are a million different ways to define any award, but personally, I think MIP should be (performance - expected change based on XP and age), and I think this might be a way of doing that.

Pellican wrote:TP, is there any way that you could evaluate the deltas that you calculated versus the jump you'd expect from each player? For instance, Cousins was in his 4th season and 23 years old and Dragic was in his 6th season at 27 years old. Is there a way you could spit out a benchmark for how much you'd expect each player to improve based on their age/experience? Maybe career minutes played would also be an appropriate way to do this.

There are a million different ways to define any award, but personally, I think MIP should be (performance - expected change based on XP and age), and I think this might be a way of doing that.

I think that's exactly what he did. His "prior" column seems to be last year's performance + expected bump in performance based on age/exp. So actual performance (IPV) - prior = exactly what you're talking about

Hmm.. I thought age/experience did not factor into IPV and that IPV was a way of keeping past data from over-influencing the metric and creating biases. Maybe I am misunderstanding it.

If it does not take those things into consideration, it would seem that young players would greatly benefit from this calculation (IPV - prior), yes? Their priors would be low because they are still feeling the effects of the rookie/young player penalization in RAPM. But if IPV does not weigh age/XP into the calculation, great young players would have high IPVs and low RAPMs, thus creating a large difference and making their "jumps" seem more substantial than those of older players.

Then again, if IPV does weigh those things, then I guess you're right and TP did exactly what I was saying without me realizing it.

Great points brought up by Pellican, as obv I did an awful job of explaining what the numbers meant. I know him from twitter, and he rather obviously knows a ton about NOP (pick his brain about them and itll become obvious), and is in the process of slowly dominating the rest of us in analytic power.

1, What James said was correct, and the numbers we used for prior include an age/experience component.
2, Our IPV numbers indeed do not include any age-based prior, and are in-season only. The priors we have here are used by us in other stuff (not in IPV), namely to have player values for the first x games of the season. It's those priors that we used here.
3, The priors used here have no rookie/young player penalization in them at all. They're based solely on prev year IPVs (which themselves have no rookie penalty), and aging/experience.
4, ftr, in hindsight, I think our prior for Jefferson sucked and was too low. his defense had just been so historically atrocious.
5, Another potential limitation is that all these MIP/MIS guys may just be guys we had bad (too low) priors for. But I 'think' that all 10 of the ones listed here (other than maybe Al) are pretty solid. Maybe Reggie was a bit too low too, considering how he looked in the playoffs last year.

Ah, that makes total sense. Looking back, it seeems obvious that you'd be using last year's IPV as a prior. I don't know why that didn't occur to me.

I was doing a mental run-through of the potential MIP candidates and Dragic/Gerald Green/Davis/Stephenson were the ones I was focusing on.. I didn't even stop to think of DeMarcus Cousins, which seems almost criminal now. I wonder how much Mike Malone played into his surge this season. Malone was uber-respected over in New Orleans and I believe at GS as well.

Finally, thanks for the kind words and for referring me to this forum, TP. Looking forward to learning from everyone here