Page 1 of 1

Player pairs that might have collinearity issues on rpm

Posted: Tue Oct 07, 2014 9:43 pm
by Crow
Anybody have a database and technique to identify say top 50-100 player pairs by frequency (% of total minutes) they play together? I wonder about it, think it would be useful to know (so as to have more caution interpreting their rpm estimates) but don't have a database set up for easy use of lookup function of total player minutes at the moment. Anyone interested in topic and providing assist?

I noticed Perkins played with Durant for all but 30 of 1200ish regular season minutes. That might noticeably affect their rpm estimates, especially Perk's. I know a few such examples have been noted before by Lewin and others but a fuller list might give more perspective on size of issue / problem.

Re: Player pairs that might have collinearity issues on rpm

Posted: Wed Oct 08, 2014 6:37 am
by bondom343
Yep, BBR's lineup finder. I just ran it for the top 2 man lineups last season, but it goes back to 2001, and can do 2, 3, 4, or 5 man groups:

http://bkref.com/tiny/D4Ahd

Wall/Ariza 1st, almost 2500 minutes. Durant/Ibaka first for OKC.

Re: Player pairs that might have collinearity issues on rpm

Posted: Wed Oct 08, 2014 3:33 pm
by Crow
Yeah, I looked at that very page and it gives clues, but I wanted easy path to calculate the % of total time that the pair played together.