2013 RAPM up to the All Star break

[v-zero]

This is basically a data dump as a result of incomplete work I'm doing - I know people are missing xRAPM and IPV (and most largely the lack of a good defensive metric) so hopefully this will help. It's built from 2000-01 onwards, with the playoffs, and with each year's ratings as the prior for the next. There is no box-score information involved, and new players to the league receive a prior according to their percentage of possible possessions played - so in 2000-01 everybody got a prior based on their percentage playing time, basically. Lambdas were found via cross-validation until it was clear that lambda settled (i.e. that the ratio of useful information contained in the prior to useful information contained in the in-season measurements became constant, which more or less happened in 2003-04). The ratings are such that possession weighted offensive ratings sum to zero, and possession weighted defensive ratings sum to zero, so that these can be interpreted with zero as the origin for for defence and offence.

Single Year Prior Informed RAPM:


Single Year Non Prior Informed RAPM:


N.B. I'm not suggesting anybody have too much faith in these, given that they aren't my end goal, but they should be as good as the estimates JE made with RiRAPM.


EDIT:

These have been discussed somewhat at gamefaqs and possibly elsewhere, I don't post there so I'll post here: Don't trust these ratings as prescriptive. They're a work in progress. Specifically intelligently decaying 'old' information (data from prior years) and weighting priors by sample size (in a basic sense) is yet to be properly done, so when you see KG and Amir ridiculously high on D you should assume the number is too high, but that they're pretty excellent. As for guards being underrated on D: size matters, guards are nothing like as useful as bigs on the interior, and since shooting% is roughly inversely proportional to distance it's fair to assume that great interior D beats great perimeter D. However, guys like Battier whose awesome perimeter D doesn't show up in the box score will show up here.

EDIT 30/03/2013:

Updated with most up to date pbp data and a vast improvement on the priors.

[deepak]

Thanks. Are you planning on updating this for games after the all-star break?

[v-zero]

I probably can do if people would like. I should be posting a superior version to this later today if all goes well - superior in that the priors are better weighted.

[deepak]

I probably can do if people would like. I should be posting a superior version to this later today if all goes well - superior in that the priors are better weighted.

That would be great. Could you also add a column for team and #poss in the spreadsheet?

[v-zero]

I can do, but the possession number will be misleading as players have a rating based on a prior as well as their possessions in season. But if people want to see it then that's no problem.

[Crow]

Is the "superior version" the one that is up there right now or is it still coming later? No rush, just seeking clarification. Thanks.

Will any GM give Milsap a max deal, in part supported by very high / max salary RAPM values? I assume he will get somewhat less despite his RAPM value. Age is of some concern. Did Daniel do a graph of average RAPM change from age 1 to age 2, etc.? I'd like to see that in general and for big minute players by position or role.

[v-zero]

This is still the bog standard version, as I'm taking my time doing cross-validation for the improved edition. The key difference between the two is to do with the implied confidence in the prior (the weight vector for the penalty vector).

Millsap should be getting a max deal, but I doubt he will - he doesn't have the flash required. However he has consistently been extremely good by most (reasonable) box score measures, so maybe.

I don't think RAPM is very suitable for creating an age curve, I think SPM is the be-all for that, and I believe Daniel has graphed that, though not by position.

[deepak]

Amir Johnson always seems to be rated really highly by these adjusted +/- stats. Even going back to his Detroit days, if I remember correctly. Can anyone comment on how good he really is?

[bchaikin]

Amir Johnson always seems to be rated really highly by these adjusted +/- stats. Even going back to his Detroit days, if I remember correctly. Can anyone comment on how good he really is?

he makes in salary about 1/2 what andrea bargnani makes, and about 1/3 of what rudy gay gets. but despite playing just 28 min/g he leads the team in - by far - rebounding, shot blocking, and is their best overall shooter. he's their most efficient scorer, yet commits just 1.4 TO/g. constant foul trouble limits his playing time, but i can't imagine they'd have anywhere near the 25 wins they have without him...

[v-zero]

His affect beyond what the box score captures, his defensive impact by RAPM has been consistently strong with less than star-level minutes, which is a very good indicator that he is actually an extremely good defender, as few players with middling minutes have consistent defensive impact by one year RAPM. If I was a GM I'd be looking to sign him to a max contract - he'd really suit somewhere like Houston where the guard play of Lin and Harden are the main offensive threat rather than that being in the forwards, so he wouldn't displace any potent offensive assets. Plus the big man pairing of Asik and Johnson would be seriously formidable on both ends.

However, his impact is not as high as these ratings suggest.

[Bobbofitos]

His affect beyond what the box score captures, his defensive impact by RAPM has been consistently strong with less than star-level minutes, which is a very good indicator that he is actually an extremely good defender, as few players with middling minutes have consistent defensive impact by one year RAPM. If I was a GM I'd be looking to sign him to a max contract - he'd really suit somewhere like Houston where the guard play of Lin and Harden are the main offensive threat rather than that being in the forwards, so he wouldn't displace any potent offensive assets. Plus the big man pairing of Asik and Johnson would be seriously formidable on both ends.

However, his impact is not as high as these ratings suggest.

As a GM I'd look to sign him for considerably less than a max.

[v-zero]

Yeah, I really meant I'd go all the way to max if needed.

[Mike G]

Amir Johnson has always had much better stats coming off the bench, compared to his stats when starting. In particular, his OReb and Blk rates are magnified as a sub.

This in spite of the fact that players tend to be promoted to the starting lineup when they're playing well.
After all these years, he's just 25. And as a starter, he's averaged 35.5 mpg this year.

After 4 years in Detroit, he was traded to Milwaukee for Fabricio Oberto.
Without playing for the Bucks, he was traded to Toronto for Carlos Delfino, who was not in the league at the time.

[talkingpractice]

I don't think RAPM is very suitable for creating an age curve, I think SPM is the be-all for that.

Yeah, that's what we do on offense. And something for n of MP in the league (ie high school guys age diff than 4yr college guys). Position too. We tested it by various components and found that while players do tend to age differently in different areas, it evens out enough to just use a complete SPM model for making the curve on offense.

But on defense, I think you don't have much of a choice but to use RAPM for the aging/experience curve, no?

[v-zero]

But on defense, I think you don't have much of a choice but to use RAPM for the aging/experience curve, no?

Agreed, though the only sensible way to do that would be using 1-year uninformed RAPM values and tracking year-to-year, so the battle is getting enough players to use to avoid the fuzziness of 1-year uninformed RAPM. It helps that defence is generally more stable game-to-game than offence.

It would be interesting to compare an RAPM D curve to and SPM D curve, as I think it might well show the 'a big part of defence is experience' notion to be true, while the SPM D curve would likely show a player degrading more quickly on D.