2001-2014 RAPM

[colts18]

I compiled all the RAPM's from 2001-2014 and created a list of players with the most points above average from 2001-2014 and the best RAPM's per 100 possessions (offense and defense). I weighed each year's RAPM by possessions played to get the 01-14 list. I also created a RAPM above replacement level stat. I calculated that the average 8th-9th man on the bench is worth around -1 points. I wanted to compare players above that threshold because teams rarely play beyond the 8th-9th man in meaningful possessions. I'm willing to listen to other values that I should use as the replacement threshold. I was going to add 1997 and 1998 RAPM into this list but I couldn't figure out a proper way to estimate how many possessions each player was on the court.

Code: Select all

Rank    Name    Points Above Avg
1	Tim Duncan	4112.4 points above average
2	LeBron James	3933.5
3	Dirk Nowitzki	3686.0
4	Kevin Garnett	3452.9
5	Kobe Bryant	3105.8
6	Dwyane Wade	2613.8
7	Steve Nash	2446.2
8	Manu Ginobili	2439.7
9	Shaquille O'Neal	2232.0
10	Jason Kidd	1903.7

Best RAPM per 100 possessions from 01-14:

Code: Select all

rank 	Name      Ovr/100
1	LeBron James	5.64
2	Tim Duncan	5.40
3	Manu Ginobili	5.07
4	Kevin Garnett	4.67
5	Dwyane Wade	4.58
6	Dirk Nowitzki	4.52
7	Shaquille O'Neal	4.44
8	Chris Paul	3.88
9	Kobe Bryant	3.59
10	Amir Johnson	3.55

Best offensive RAPM per 100 from 01-14:

Code: Select all

Name	     Off/100        	Poss
LeBron James	4.63	69749.0
Dwyane Wade	4.31	57051.0
Steve Nash	4.10	73342.0
Manu Ginobili	3.79	48097.5
Kobe Bryant	3.75	86610.0
James Harden	3.53	22481.0
Dirk Nowitzki	3.51	81563.0
Shaquille O'Neal	3.36	50296.5
Chris Paul	3.31	42397.0
Baron Davis	2.80	55669.0
Blake Griffin	2.56	19418.0

site for every player from 2001-2014:

https://sites.google.com/site/rapmstats/rapm-above-avg

Spreadsheet:

https://docs.google.com/spreadsheet/ccc ... XRkE#gid=0

[J.E.]

Cool stuff. Your estimates seem to be closer to 0 than mine. What penalty value are you using?

[colts18]

Cool stuff. Your estimates seem to be closer to 0 than mine. What penalty value are you using?

I think you misunderstood what I did. What I did was compiled the RAPM list from each season, then found out how many points the player was above average in that season. Then I added up each season for each player to get a career 01-14 value. It wasn't an actual 01-14 RAPM where I calculated RAPM in that span. I compared players to the year they played because my goal was to find the RAPM season which compiled the most value after adjusting for volume. I don't have the matchup files to do an actual 01-14 RAPM. For example, LeBron is worth -0.5 RAPM in 2014. I multiplied that by the 5981 possessions he played that season to get -29.9 points above average. LeBron is worth +640 points in 2010 so that gets added up too. I added each season up to get the career points above average. For the RAPM/100, I weighed each season by possessions played.

[J.E.]

I see. Thanks for the clarification

[jbrocato23]

Very cool. To be clear, are you using npi rapm? Or something else (maybe rapm informed rapm)?

[colts18]

I see. Thanks for the clarification

J.E., I'm curious about your 2006 prior informed RAPM. For some reason I couldn't find some of the players you listed on it while searching in B-R. A few had only last names and some of those shared the same last name.

http://web.archive.org/web/201110030504 ... /ranking06

[J.E.]

I see. Thanks for the clarification

J.E., I'm curious about your 2006 prior informed RAPM. For some reason I couldn't find some of the players you listed on it while searching in B-R. A few had only last names and some of those shared the same last name.

http://web.archive.org/web/201110030504 ... /ranking06

I guess I was using basketballvalue.com's matchupfiles back then. They were kind of messed up for '06

[colts18]

Very cool. To be clear, are you using npi rapm? Or something else (maybe rapm informed rapm)?

01-06 NPI RAPM. 07-14 Prior informed. From what I understand the prior informed from 2002-2006 is missing a lot of data (like 1/3 of the data) while the NPI of 01-06 is full season. Obviously there are flaws to this method because NPI values are smaller than the PI values.