Notes about RPM estimates

Home for all your discussion of basketball statistical analysis.
Post Reply
Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Notes about RPM estimates

Post by Crow » Wed Dec 19, 2018 5:58 pm

Just six players are plus 2 on both ORPM and DRPM estimates. Davis, Jokic, George, Vucevic, Giannis A. and Siakam. All 6-8 or better, all frontline.

J Butler and R Covington are measured at almost the exact same overall impact but are near polar opposites on the splits. Butler mostly offensive impact, Covington all defensive impact.

Of the 15 with 4 plus RPM only Harden and Lillard have a negative split, both on defense. Gasol with the weakest ORPM in the group.

32 over plus 3. At least ten won't make All-Star. Some legitimately because of reserve minutes. Some will deserve the recognition but won't get it because of being defenders or on lower profile teams or just not being big names like the guys who get in instead of them. Which mainstream media person is going to champion Vucevic, Siakam or Ingles to their colleagues, coaches and fans?

Of players not in the RPM top 40, which will still make All Star game? Westbrook, Simmons, KAT, Middleton, T Harris, Horford, Tatum, Draymond Green, DeRozan, Booker, Wall, A Gordon?

Knicks and Bulls have almost half of the bottom 15. Add Cavs & Suns and you have almost 75%. Not all are that bad, some of it is bad usage / coaching, tanking. But they are probably bad.

5 of the top 9 draft picks this summer are -3 or a lot worse. Aldridge is the most surprising big name / paycheck with a -3 estimate.

On average teams have 3 guys -2 or worse. OKC has 6. Not saying they are worst (haven't checked), just that this is notable.

tarrazu
Posts: 66
Joined: Mon Aug 04, 2014 5:02 pm

Re: Notes about RPM estimates

Post by tarrazu » Wed Dec 19, 2018 8:36 pm

LMA defense has dropped off a lot this year, but how weak is the prior that he is now a -3? He does have career high TO% and his on/off splits on defense have not been good.

Previous RPM seasons for reference:
2017-18: +0.96
2016-17: +3.14
2015-16: +2.15

Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Re: Notes about RPM estimates

Post by Crow » Wed Dec 19, 2018 8:39 pm

My understanding is that the box score prior is for season to date only now and not prior season.

Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Re: Notes about RPM estimates

Post by Crow » Wed Dec 19, 2018 8:40 pm

Among other top contenders, Raptors have 2 at -2 or worse. Celtics 1, Bucks 0, Warriors 2, Nuggets 1, Sixers 0.

They may play moderate to lower minutes and likely less in playoffs. But having a lot of these guys and almost all in regular season rotation does affect the Thunder.

DSMok1
Posts: 857
Joined: Thu Apr 14, 2011 11:18 pm
Location: Maine
Contact:

Re: Notes about RPM estimates

Post by DSMok1 » Thu Dec 20, 2018 1:04 pm

Crow wrote:
Wed Dec 19, 2018 8:40 pm
Among other top contenders, Raptors have 2 at -2 or worse. Celtics 1, Bucks 0, Warriors 2, Nuggets 1, Sixers 0.

They may play moderate to lower minutes and likely less in playoffs. But having a lot of these guys and almost all in regular season rotation does affect the Thunder.
This is just a guess, but I think the box score prior may be to blame for the OKC RPM distribution. If it has any of the issues that BPM has, it may over-credit Westbrook and George, who have high usage, and consider the other players worse than may actually be warranted.
Developer of Box Plus/Minus
APBRmetrics Forum Administrator
GodismyJudgeOK.com/DStats/
Twitter.com/DSMok1

shadow
Posts: 207
Joined: Fri May 29, 2015 3:38 am

Re: Notes about RPM estimates

Post by shadow » Thu Dec 20, 2018 7:19 pm

tarrazu wrote:
Wed Dec 19, 2018 8:36 pm
LMA defense has dropped off a lot this year, but how weak is the prior that he is now a -3?
His box score prior must be pretty bad because I have his RAPM at only -0.81 as of 12/18.

https://docs.google.com/spreadsheets/d/ ... ingle=true

Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Re: Notes about RPM estimates

Post by Crow » Thu Dec 20, 2018 9:07 pm

It is not what they use as RPM prior but fwiw Aldridge's BPM is -0.7 right now. He might be estimated a little low by RPM, based on past. Time will tell.

33.4 yrs old.

tarrazu
Posts: 66
Joined: Mon Aug 04, 2014 5:02 pm

Re: Notes about RPM estimates

Post by tarrazu » Fri Dec 21, 2018 3:22 pm

shadow wrote:
Thu Dec 20, 2018 7:19 pm
tarrazu wrote:
Wed Dec 19, 2018 8:36 pm
LMA defense has dropped off a lot this year, but how weak is the prior that he is now a -3?
His box score prior must be pretty bad because I have his RAPM at only -0.81 as of 12/18.

https://docs.google.com/spreadsheets/d/ ... ingle=true
I show -0.8 as well for LMA this season. Maybe I don't get what RPM is trying to do then? Showing a guy's impact THIS SEASON, while utilizing some data from prior seasons still?

DSMok1
Posts: 857
Joined: Thu Apr 14, 2011 11:18 pm
Location: Maine
Contact:

Re: Notes about RPM estimates

Post by DSMok1 » Fri Dec 21, 2018 3:45 pm

RPM, as shown on the ESPN website, does not include data from prior seasons AFAIK. Jerry does have a multiyear model for projections.

The RPM box score prior is quite strong early in the season, and that prior does not necessarily match Box Plus/Minus or any other box score metric.
Developer of Box Plus/Minus
APBRmetrics Forum Administrator
GodismyJudgeOK.com/DStats/
Twitter.com/DSMok1

Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Re: Notes about RPM estimates

Post by Crow » Fri Jan 11, 2019 1:16 am

After a long rise to about plus 2 Doncic's RPM estimate has fallen back to plus 0.28. The terrible overall team performance in last few weeks (probably including Doncic's contribution) has cancelled out the strong team & Doncic performance in the few weeks before that. So the young star or expected to become a star is back to basically a neutral impact estimate. 115th place. Probably not going to hear about that elsewhere.

It could bounce again. Too early to conclude much about impact. But better to watch and consider on that than not. His neighbors with near neutral impacts are generally not being talked about much. A few are getting criticism. It probably should vary by expectations or role / opportunity. If you are a star you should be expected to post star level RPM. Plus 2ish imo. So Doncic reached that threshold... and fell way back. Where will he be by season's end? We'll see.

On shooting / scoring / overall offense measures, Doncic is below avg. on efg%, above avg. in ts% by about as much. Below average on offensive rating but above average on OBPM. In between but positive on OPIPM. DBPM estimates positive defensive contributions. DPIPM and DRPM estimate negative defensive contributions.

Versatility bonus and positive offense bonus on DBPM may be causing BPM to "see" Doncic more positive than the other metrics. The extent to which all these metrics are adjusted to conform to eye test / expectations is an issue. May be appropriate but how much is the right amount and by what method?


Doncic riding the media / fan hype to a All-Star starter voting level. One of the 3 without the RPM corroboration. Doncic and Rose. Wade was a big negative last I saw. Want what you want... but me, I don't want Wade in All-Star game at all The other two, I might let slide but I wouldn't vote that way directly.

tarrazu
Posts: 66
Joined: Mon Aug 04, 2014 5:02 pm

Re: Notes about RPM estimates

Post by tarrazu » Fri Jan 11, 2019 5:37 am

tarrazu wrote:
Fri Dec 21, 2018 3:22 pm
shadow wrote:
Thu Dec 20, 2018 7:19 pm
tarrazu wrote:
Wed Dec 19, 2018 8:36 pm
LMA defense has dropped off a lot this year, but how weak is the prior that he is now a -3?
His box score prior must be pretty bad because I have his RAPM at only -0.81 as of 12/18.

https://docs.google.com/spreadsheets/d/ ... ingle=true
I show -0.8 as well for LMA this season. Maybe I don't get what RPM is trying to do then? Showing a guy's impact THIS SEASON, while utilizing some data from prior seasons still?
LMA is ok again? He is up a touch from his -3 earlier this year. RPM is essentially explanatory for this single season? I believe I was mistaken is thinking it was more than that.

Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Re: Notes about RPM estimates

Post by Crow » Tue Jan 15, 2019 1:13 am

Doncic back up to plus 2? That is a pretty big and quick bounce since 4 days ago...

Mike G
Posts: 4210
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Notes about RPM estimates

Post by Mike G » Wed Jan 16, 2019 5:53 pm

Doncic just had 2 of his 3 biggest games of the year, in GameScore.
https://www.basketball-reference.com/pl ... elog/2019/

Sorting thru his box scores, I see that in the 10 games he's played <30 minutes, the average scoring margin is 21 points: mostly blowouts, where scrubs from both teams get minutes.
In the 10 games he's played >36, avg margin is 7 points: He plays more in the close ones, presumably against better players.

Maybe it takes in-season RPM a while to stabilize everyone's "true" value.

Crow
Posts: 5695
Joined: Thu Apr 14, 2011 11:10 pm

Re: Notes about RPM estimates

Post by Crow » Sat Jan 19, 2019 2:39 am

Mavs are 10-3 when they go better than plus 5 with Doncic on court. 2-16 when they are negative with him on court. 7-5 when they are neutral to plus 5 with him on court. 1-0 without him.

Post Reply