Estimated non boxscore RPM value

[Dr Positivity]

A little project I did. I take PER or WS/48 and created an "expected RPM", on a scale that makes 9 RPM equivalent to 30 PER and .300 WS/48. Then the real RPM is subtracted from it to create the difference. The attempt is to estimate how much a player's RPM is affected by non boxscore production.

For example Harden has .289 WS/48 and 29.8 PER. In my scale this makes his expected RPM 8.51 by WS/48 and 8.88 by PER. Since his real RPM is 6.71 the difference for his "non boxscore value" is -1.8 for the WS/48 and -2.17 for the PER. This number essentially reflects that the main reason his RPM is so high is that his statistical production is just that good. That RPM is less high on Harden than his box stats is also inherently obvious when seeing he leads the league in PER and WS/48 handily but is 2nd in RPM with a great but not as through the roof +6.71 (would be lower than 2nd most years). On the other hand Oladipo's WS/48 is .155 and PER is 23.1 making his expected RPM for those values 2.48 and and 4.86 respectively. Since his real RPM is 5.91 this makes him +3.43 on the WS/48 chart and +1.05 on the PER chart. This would suggest Oladipo is making a positive impact outside of his box stats in addition to all-star production there.

The top 30 in the stat

WS/48:

Code: Select all

Jones, Tyus	4.01
Holiday, Justin	3.94
Speights, Mareese	3.54
Chalmers, Mario	3.46
Oladipo, Victor	3.43
Ball, Lonzo	3.24
Bazemore, Kent	3.03
Ntilikina, Frank	3.03
Holiday, Jrue	3.01
Smart, Marcus	2.94
Green, Draymond	2.82
Ferrell, Yogi	2.71
Porzingis, Kristaps	2.66
Rodney, Purvis	2.64
Embiid, Joel	2.62
James, Mike	2.54
Korver, Kyle	2.52
Covington, Robert	2.48
Gasol, Marc	2.45
Mitchell, Donovan	2.38
Murray, DeJounte	2.33
Cousins, DeMarcus	2.31
Evans, Tyreke	2.29
Hardaway, Tim	2.28
Brooks, Dillon	2.22
Porter, Otto	2.21
Westbrook, Russell	2.19
Smith, Dennis	2.18
Vucevic, Nikola	2.14

PER:

Code: Select all

Covington, Robert	6.71
Jones, Tyus	5.32
Tucker, P.J.	5.23
Roberson, Andre	4.23
Ingles, Joe	3.86
Mbah a Moute, Luc	3.8
Aminu, Al-Farouq	3.8
Iguodala, Andre	3.69
Korver, Kyle	3.66
Dudley, Jared	3.64
Anunoby, OG	3.59
Holiday, Justin	3.58
Ariza, Trevor	3.58
Smart, Marcus	3.48
O'Neale, Royce	3.42
Green, Draymond	3.38
Ferrell, Yogi	3.25
Terry, Jason	3.04
Calderon, Jose	3.02
Porter, Otto	2.98
VanVleet, Fred	2.86
Tatum, Jayson	2.74
Chalmers, Mario	2.66
Johnson, Stanley	2.64
Ball, Lonzo	2.62
Anderson, Kyle	2.58
Patterson, Patrick	2.54
Galloway, Langston	2.48
Lowry, Kyle	2.48
Miller, Darius	2.46
Liggins, DeAndre	2.46
Huestis, Josh	2.45

Since there is a lot of players on that list who only rate well because their boxscore stats are bad, isolating it to only players over .100 WS/48

Code: Select all

Jones, Tyus	4.01
Oladipo, Victor	3.43
Holiday, Jrue	3.01
Green, Draymond	2.82
Embiid, Joel	2.62
Korver, Kyle	2.52
Covington, Robert	2.48
Cousins, DeMarcus	2.31
Evans, Tyreke	2.29
Porter, Otto	2.21
Westbrook, Russell	2.19
Vucevic, Nikola	2.14
Butler, Jimmy	1.98
Mirotic, Nikola	1.89
Rubio, Ricky	1.85
Bledsoe, Eric	1.83
Roberson, Andre	1.77
Gordon, Eric	1.76
Ingles, Joe	1.65
Dinwiddie, Spencer	1.59
Walker, Kemba	1.56
Nurkic, Jusuf	1.44
Brown, Jaylen	1.39
Millsap, Paul	1.27
Harris, Gary	1.22
Griffin, Blake	1.21
VanVleet, Fred	1.19
Olynyk, Kelly	1.18
Tatum, Jayson	1.16
Ariza, Trevor	1.07
Young, Thaddeus	1.04
Horford, Al	0.96
Lowry, Kyle	0.95

Only players over 15 PER:

Code: Select all

Green, Draymond	3.38
Porter, Otto	2.98
VanVleet, Fred	2.86
Tatum, Jayson	2.74
Anderson, Kyle	2.58
Lowry, Kyle	2.48
Horford, Al	2.33
Udoh, Ekpe	2.29
Rubio, Ricky	2.15
Holiday, Jrue	2.05
Hilario, Nene	1.99
Gordon, Eric	1.87
Dinwiddie, Spencer	1.86
Murray, DeJounte	1.68
Satoransky, Tomas	1.66
Olynyk, Kelly	1.66
Paul, Chris	1.35
Butler, Jimmy	1.17
Rozier, Terry	1.16
Thompson, Klay	1.15
Mitchell, Donovan	1.13
Oladipo, Victor	1.05
Bell, Jordan	1.01
Harris, Gary	0.95
Mirotic, Nikola	0.93
Johnson, James	0.89
Nowitzki, Dirk	0.87
Sefolosha, Thabo	0.85
Wright, Delon	0.75
Davis, Ed	0.73
George, Paul	0.68
Walker, Kemba	0.45

This is not a list of best players. For example on Toronto Valanciunas rates as the worst on the team by a lot (-3.51, 2nd worst is Derozan at -1.94) while Ibaka has a credible +0.84. However Valanciunas leads the team in both WS/48 (.210 WS/48) and PER (22.5) to Ibaka's .117 and 14.7 respectively. Overall RPM is 0.99 for Valanciunas and 0.66 for Ibaka it's just they get there in different ways by Valanciunas putting up big numbers but then taking a lot of his impact away in non boxscore value, while Ibaka puts up average stats but does not take as much off the table, for an impact rated close to the same in RPM. Teams need players to put up boxscore stats so there is a place for a Valanciunas contribution to a team.

Could a list like this be used to predict breakout players? Perhaps what you want is a young player who has appears to have a high baseline of non box value like Tyus Jones and then hope their boxscore production improves to make them a legitimate star in value.

[Crow]

Or you could put a good boxscore guy in less negative / more constructive non-boxscore context.

To be clear, all these lists are the non-boxscore residuals right? A label to that effect might be good.

Would be interested in seeing full league database, perhaps using BPM or PIPM. PER and WS/48 are perhaps purer boxscore but also probably less well calibrated.

I have done similar projects before. My conversion rates might have been a little different. I don't remember offhand.

Thanks for doing this and I hope you even more. I am always in favors on knowing as much as possible about the splits of valuation. Offense / defense is the starting point, not the end. Boxscore / nonboxscore is very important to track and think about. Player development timing? Priorities? Realistic expectations? That all moves analysis toward action / improvement.

[DSMok1]

I am working with a 20 year RAPM dataset (1997-2017) to develop a new BPM formulation.

With a pure linear statistical plus minus, here are the residuals:


A negative residual would indicate the player produced more than the box score would indicate (non-box-score contribution) while a positive residual indicates the player produced less than the box score would indicate.

What I see is that post players are the hardest to measure with box score statistics. Much of their contribution comes from post defense--or lack thereof. More generally, defense is not shown in the box score. Players elite on defense are underrated with box scores and players bad on defense are overrated (particularly if they are post players). Jason Collins and Nick Collison vs. Hassan Whiteside and Carlos Rogers.

[Dr Positivity]

I am working with a 20 year RAPM dataset (1997-2017) to develop a new BPM formulation.

With a pure linear statistical plus minus, here are the residuals:
https://docs.google.com/spreadsheets/d/ ... sp=sharing

A negative residual would indicate the player produced more than the box score would indicate (non-box-score contribution) while a positive residual indicates the player produced less than the box score would indicate.

What I see is that post players are the hardest to measure with box score statistics. Much of their contribution comes from post defense--or lack thereof. More generally, defense is not shown in the box score. Players elite on defense are underrated with box scores and players bad on defense are overrated (particularly if they are post players). Jason Collins and Nick Collison vs. Hassan Whiteside and Carlos Rogers.

Yes it was generally the same on my list. Kanter, Monroe, Biyombo, Capela, Valanciunas, Harrell, McGee, Drummond, Whiteside, etc all a near the bottom of the list, especially on the WS/48 list. It just backs up that centers playing near the basket really hurts teams on offense, especially if they are not great defenders