RAPM request thread

[permaximum]

Is there any level of RAPM from the MJ era? I noticed awhile back JE you said you didn't have the data to get this done from say 89. What is the earliest year RAPM we have?

It goes back to 1996-97.
Here's 1996-97.

Reading through this thread over at realgm I was surprised to find that we have NPI RAPM for 94-96 as well. Albeit with a few difference such as not included playoff data or players who switched teams. Credit to colts18 for doing the regressions. That said, I don't know where the data to fit the models was found.

94
95
96

AFAIK it's not possible to have that kind of data. Even official NBA.com has PBP data going back to 1996-97. I suspect he made estimates and predicted RAPM for those years like J.E. did for all of 90's once.

[AcrossTheCourt]

It goes back to 1996-97.
Here's 1996-97.

Reading through this thread over at realgm I was surprised to find that we have NPI RAPM for 94-96 as well. Albeit with a few difference such as not included playoff data or players who switched teams. Credit to colts18 for doing the regressions. That said, I don't know where the data to fit the models was found.

94
95
96

AFAIK it's not possible to have that kind of data. Even official NBA.com has PBP data going back to 1996-97. I suspect he made estimates and predicted RAPM for those years like J.E. did for all of 90's once.

Those realgm numbers are raw plus/minus, and there's some information that goes back to the 80's. There are no publicly released pbp files before 1997.

And yeah, the earliest MJ RAPM we have is the one I did for 1997. He did pretty well there.

[tarrazu]

Single https://docs.google.com/spreadsheets/d/ ... sp=sharing
Multi https://docs.google.com/spreadsheets/d/ ... sp=sharing (less weight to older seasons, 3 seasons total)

How did you determine optimal weights for older seasons?

[tarrazu]

Single https://docs.google.com/spreadsheets/d/ ... sp=sharing
Multi https://docs.google.com/spreadsheets/d/ ... sp=sharing (less weight to older seasons, 3 seasons total)

How did you determine optimal weights for older seasons?

If anyone else knows or has possible ideas for approaching this, it is an interesting problem.

[permaximum]

Single https://docs.google.com/spreadsheets/d/ ... sp=sharing
Multi https://docs.google.com/spreadsheets/d/ ... sp=sharing (less weight to older seasons, 3 seasons total)

How did you determine optimal weights for older seasons?

If anyone else knows or has possible ideas for approaching this, it is an interesting problem.

Ridge regression.

Predictor
Y-1 RAPM
Y-2 RAPM
Y-3 RAPM

Response
Next season's RAPM (Y)

He possibly found the lambda by 3-fold cross validation. I use and suggest 10-times repeated 10-fold cross-validation to minimize both variance and bias.

[shadow]

I ran a ridge regression on 13/14/15 RAPM to predict 16 RAPM and found the following weights using 10 fold CV:

2015: 45%
2014: 38%
2013: 17%

FWIW, this regression only had an R^2 of 0.26