Question regarding ESPN's RPM

Post #1 » by **jbk1234** » Wed Sep 9, 2020 5:19 pm

So I'm not saying this is deliberate, but it is, at a minimum, really sloppy on ESPN's part. They're clearly listing players out of position when adjusting the box numbers and they're not circling back to correct the error. My question is does the player position listed impact the adjustment?

For example, John Henson is listed as the 23 best PF, with Love and Nance immediately behind him according to their web site: http://www.espn.com/nba/statistics/rpm/_/position/6

As far as I know, John Henson has never played PF in his life. He's a center. He had the best box DBPM off all the Cavs big men last year.

Henson had an OBPM of -1.2, a DBPM of 1.8, and BPM of .6.

Kevin Love was correctly listed as a PF and had an OBPM of 2.6, a DBPM of -.5, and BPM of 2.0.

Nance Jr., who spent a good chunk of time at center alongside Love, was also listed as a PF and had an OBPM of 1.4, a DBPM of .4, and a BPM of 1.7.

TT was correctly listed as a center and had an OBPM of -.4, and DBPM of -1.3, and BPM of -1.5.

So ESPN categorizes Henson as PF (inexplicable) and Nance as a PF (arguable) and this is what happens to TT's adjusted numbers: ORPM -.89, DRPM 1.48, RPM -.92

This is what happens to Love's: ORPM .38, DRPM -.28, RPM .1

These *adjustments* represent huge shifts away from the box. Is it a result, in least in part, of listing Henson out of position?

Slax · Post #2 » by **Slax** » Wed Sep 9, 2020 5:48 pm

I'm not sure about RPM, but as far as I'm aware most of these factor-adjusted plus minus stats do not use position as one of the factors, but they do often include player height.

I do want to emphasize, this whole family of stats based on APM is super prone to statistical artifacts and can have large differences just based on which factors are included or emphasized when doing adjustments.

yoyoboy · Post #3 » by **yoyoboy** » Wed Sep 9, 2020 5:51 pm

RPM sucks anyways since they updated the formula halfway through the season to become way more dependent on the box score.

It's a shame RAPM numbers are nowhere to be found anymore.

Post #4 » by **jbk1234** » Wed Sep 9, 2020 5:54 pm

Slax wrote:I'm not sure about RPM, but as far as I'm aware most of these factor-adjusted plus minus stats do not use position as one of the factors, but they do often include player height.

I do want to emphasize, this whole family of stats based on APM is super prone to statistical artifacts and can have large differences just based on which factors are included or emphasized when doing adjustments.

Henson was the tallest guy on the team so if they're adjusting based on height, it still doesn't make any sense.