I'm working on a project to collect RAPM from 1997-2014 and compiling how many points above average a player was in that time period. So its a volume based stat that will reward excellence with higher minutes compared to lower minute high RAPM guys (Collison, Johnson, etc.). I'm using 97-00 NBA.com pbp, 01-04 non-prior informed RAPM, 05-14 prior informed. I have a few questions.
-Should I compile how many points a player is above 0? Above -1? -2? other?
-How can I estimate possessions on court for players from 1997-2000?
-From 2001-2014, I'm using exact possession on court from B-R. Should I multiply a players offensive RAPM by the number of offensive possessions on court or the average of his offensive and defensive possessions on court?
RAPM above average
Re: RAPM above average
You'd have to estimate replacement level to figure out how many points below average to count.
I did it with xRAPM here: http://public.tableausoftware.com/share ... _count=yes
I found something like -1.75 on O and 0 on D.
I did it with xRAPM here: http://public.tableausoftware.com/share ... _count=yes
I found something like -1.75 on O and 0 on D.
Re: RAPM above average
Who has made 1997-2000 RAPM public?colts18 wrote:How can I estimate possessions on court for players from 1997-2000?
If you've got the PBP you could potentially create your own matchupfiles with a parser, then figure out the possessions that way. The easy way would be to look up player minutes for those seasons (maybe also team pace) and with that you can estimate possessions. It's something like player_minutes*4
Re: RAPM above average
http://ascreamingcomesacrossthecourt.bl ... -rapm.htmlJ.E. wrote:Who has made 1997-2000 RAPM public?colts18 wrote:How can I estimate possessions on court for players from 1997-2000?
If you've got the PBP you could potentially create your own matchupfiles with a parser, then figure out the possessions that way. The easy way would be to look up player minutes for those seasons (maybe also team pace) and with that you can estimate possessions. It's something like player_minutes*4
http://ascreamingcomesacrossthecourt.bl ... d-rpi.html
A poster named acrossthecourt was nice enough to compile the 97 and 98 RAPM and posted them on his site. I might skip the 99 and 00 RAPM's and add them to this project if he ever posts the 99 and 00 RAPM.