APBRmetrics

Posted: **Wed Jul 01, 2015 6:20 pm**

Using J.E.'s recently released 15 year RAPM, I decided to look at which players are getting overrated/underrated in box score stats like BPM and WS/48.

Top 10 most underrated by BPM

Name RAPM BPM diff
Steve Nash 6.36 1.7 4.66
Michael Kidd-Gilchrist 3.22 -1.4 4.62
LaMarcus Aldridge 5.75 1.2 4.55
Corliss Williamson 1.47 -2.8 4.27
Omer Asik 3.25 -0.9 4.15
Eric Williams 2.14 -2 4.14
Othella Harrington 1.13 -2.8 3.93
Avery Bradley 1.61 -2.2 3.81
Jason Collins 1.66 -2 3.66
Derek Fisher 2.93 -0.7 3.63

Top 10 most overrated by BPM:

Name RAPM BPM diff
Hakim Warrick -8.74 -3.1 -5.64
Anthony Mason -4.4 0.9 -5.3
Ed Davis -4.24 1 -5.24
Jeff Green -6.14 -1 -5.14
Josh Childress -3.5 1.3 -4.8
Kyrie Irving -1.13 3.3 -4.43
J.J. Hickson -6.46 -2.3 -4.16
Chris Webber -1.45 2.6 -4.05
Dominic McGuire -5.6 -1.6 -4
Karl Malone 0.43 4.4 -3.97

Underrated by Offensive BPM

Name O RAPM O BPM Off Diff
Draymond Green 2.1 -1 3.1
Michael Kidd-Gilchrist 0.42 -2.4 2.82
Steve Nash 6.94 4.5 2.44
Jamario Moon 1.32 -1.1 2.42
Terrence Ross 2.28 0 2.28
Jason Thompson 0.56 -1.7 2.26
Brian Scalabrine -0.98 -3.2 2.22
Alonzo Mourning -0.68 -2.8 2.12
Melvin Ely -1.56 -3.6 2.04
Mike Conley 3.84 1.9 1.94

Overrated by Offensive BPM

Name O RAPM O BPM Off Diff
Kyrie Irving 1.05 4.6 -3.55
Jeff Green -3.88 -0.7 -3.18
Ron Mercer -5.34 -2.2 -3.14
Hakim Warrick -4.47 -1.5 -2.97
Kemba Walker -0.88 2 -2.88
Brandon Knight -1.95 0.9 -2.85
Michael Curry -6.39 -3.6 -2.79
Aaron Williams -4.92 -2.2 -2.72
Andrew Bynum -2.9 -0.3 -2.6
Ed Davis -3.03 -0.5 -2.53

Underrated by D BPM

Name D RAPM D BPM D Diff diff
Corliss Williamson 2.36 -1.4 3.76 4.27
Kevin Garnett 7.2 3.6 3.6 3.21
Othella Harrington 2.73 -0.4 3.13 3.93
Jason Collins 5.1 2 3.1 3.66
Casey Jacobsen 1.63 -1.3 2.93 1.62
Omer Asik 4.62 1.7 2.92 4.15
John Stockton 1.49 -1.4 2.89 2.36
Dikembe Mutombo 5.53 2.7 2.83 3.26
LaMarcus Aldridge 3.18 0.4 2.78 4.55
Brendan Haywood 3.96 1.2 2.76 2.04

Overrated by D BPM

Name D RAPM D BPM D Diff
Dominic McGuire -2.48 1.9 -4.38
Josh Childress -4.09 0.2 -4.29
Anthony Mason -2.38 1.2 -3.58
Trevor Booker -2.51 1 -3.51
Nicolas Batum -2.21 1.1 -3.31
Bismack Biyombo -1.11 1.8 -2.91
Troy Murphy -2.67 0.1 -2.77
JaVale McGee -1.56 1.2 -2.76
Damien Wilkins -3.15 -0.4 -2.75
Ed Davis -1.21 1.5 -2.71

Here are the players underrated by WS/48 in comparison to 15 year RAPM:

Code: Select all

Name	WS 48	RAPM	exp RAPM	diff
Ricky Rubio	0.085	4.74	-0.9	5.7
Baron Davis	0.108	5.35	0.1	5.3
Draymond Green	0.122	5.95	0.7	5.2
Kevin Garnett	0.201	9.11	4.2	4.9
Jrue Holiday	0.076	3.21	-1.3	4.5
Avery Bradley	0.043	1.61	-2.8	4.4
Iman Shumpert	0.061	2.3	-2.0	4.3
Metta World Peace	0.099	3.9	-0.3	4.2
Michael Kidd-Gilchrist	0.084	3.22	-1.0	4.2
Tony Allen	0.111	4.26	0.2	4.0

Overrated by WS/48:

Code: Select all

Name	WS 48	RAPM	exp RAPM	diff
Hakim Warrick	0.092	-8.74	-0.6	-8.1
Ed Davis	0.14	-4.24	1.5	-5.8
J.J. Hickson	0.088	-6.46	-0.8	-5.7
Aaron Williams	0.113	-5.02	0.3	-5.3
Marreese Speights	0.118	-4.58	0.5	-5.1
Anthony Mason	0.119	-4.4	0.6	-5.0
Jordan Hill	0.098	-5.23	-0.4	-4.9
Jeff Green	0.077	-6.14	-1.3	-4.9
Troy Murphy	0.121	-4	0.7	-4.7
Brandan Wright	0.184	-1.15	3.5	-4.6

A lot of the same names on multiple lists. Hakim Warrick is an interested name. By conventional box score stats he is average (15.5 PER, 0.092 WS/48). BPM has him at replacement level (-3.1). But 15 year RAPM has him as the worst player in the league over that span.

The players that get underrated by box score stats seem to be defensive players like Jason Collins or Omer Asik whose contributions on defense don't get measured on the stat sheet. On the offensive side of the ball, it's hard to pick out a trend on what kind of players get underrated or overrated. Nash is definitely the outlier on offense. Irving is the outlier on the opposite end of the spectrum.

By WS/48, defense is the one that gets underrated. By my count, 9 of the top 10 on the underrated by WS/48 list are known as defensive players with the exception of Baron Davis.

Posted: **Wed Jul 01, 2015 6:45 pm**

Chris Webber is on the overrated by BPM list. But he is another interesting name. Using Dsmok1's spreadsheet on BPM, I compared players WS/48 and BPM values.

Here are the top 10 seasons that would be underrated if you used WS/48 instead of BPM.

Year Tm Player WS/48 BPM exp Diff
1995 WSB Chris Webber 0.093 4.5 -0.3 4.8
2008 CLE LeBron James 0.242 10.4 5.7 4.7
2009 MIA Dwyane Wade 0.232 9.9 5.3 4.6
1982 NYK Micheal Ray Richardson 0.086 4.0 -0.6 4.6
1986 PHI Charles Barkley 0.176 7.6 3.0 4.6
1983 GSW Micheal Ray Richardson 0.006 0.7 -3.8 4.5
1992 MIL Alvin Robertson 0.071 3.3 -1.2 4.5
1987 PHI Charles Barkley 0.21 8.9 4.4 4.5
1986 SAS Alvin Robertson 0.143 6.2 1.7 4.4
2000 ORL Bo Outlaw 0.14 5.8 1.6 4.3

95 Webber was below average by WS/48 but BPM had him as one of the top players in the league.

Top 10 seasons overrated by WS/48:

Year Tm Player WS/48 BPM exp Diff
2005 SEA Danny Fortson 0.197 -2.5 3.9 -6.4
1986 GSW Geoff Huston 0.072 -6.1 -1.1 -5.0
1994 SEA Ricky Pierce 0.188 -1.3 3.5 -4.8
2013 IND Tyler Hansbrough 0.154 -2.6 2.2 -4.7
1997 DAL Chris Gatling 0.147 -2.7 1.9 -4.6
2002 DEN Avery Johnson 0.114 -4.0 0.5 -4.5
2002 CHH Jamaal Magloire 0.177 -1.4 3.1 -4.5
2005 HOU Dikembe Mutombo 0.186 -1.0 3.4 -4.5
1983 LAL Bob McAdoo 0.167 -1.7 2.7 -4.4
2014 CHA Bismack Biyombo 0.144 -2.5 1.8 -4.3

These lists tell me that BPM seems to be a better stat than WS/48. The top list is much better than the players on the bottom list.

Here are the correlation values between 15 year RAPM and certain stats:

BPM 0.768
VORP 0.680
WS/48 0.651
WS 0.597
OWS 0.557
OBPM 0.545
DWS 0.545
PER 0.514
ORtg 0.466
MP 0.427
DBPM 0.393
STL% 0.286
AST% 0.217
USG% 0.173
FTr 0.163
DRB% 0.143
TRB% 0.104
BLK% 0.089
3PAr 0.088
ORB% 0.022
TOV% -0.036
DRtg -0.381

Stl% and AST% seem to be the most important component stats. Rebounding has little correlation to RAPM. Total Minutes played has a stronger correlation than any other component.

Posted: **Wed Jul 01, 2015 10:49 pm**

Thanks for the analysis.

Posted: **Thu Jul 02, 2015 2:10 pm**

Nash is probably the most interesting player on the "underrated" list for me. He's got a skillset that if you were summarizing would lead you to think would be well captured by RAPM. It's not like he's setting a lot of picks or something. The best guess I've seen theorized have is that his assists are of a higher quality than most point guard assists (big man assists get credit via AST%*TRB%).

He seems like a guy who suffers from BPM not having John Stockton to train itself on.

Posted: **Thu Jul 02, 2015 5:24 pm**

kmedved wrote:Nash is probably the most interesting player on the "underrated" list for me. He's got a skillset that if you were summarizing would lead you to think would be well captured by RAPM. It's not like he's setting a lot of picks or something. The best guess I've seen theorized have is that his assists are of a higher quality than most point guard assists (big man assists get credit via AST%*TRB%).

He seems like a guy who suffers from BPM not having John Stockton to train itself on.

BPM was trained on a set including Nash and his outlier RAPM. Just... it's an outlier.

Posted: **Thu Jul 02, 2015 5:31 pm**

The Suns in the Nash era are known to have had a scorekeeper who was very stingy in awarding assists. All visiting players got fewer than normal assists in Phx, and Nash got considerably more (per Phx FG) on the road. So you might just adjust his assists upward, relative to league norms, in any boxscore metric.

If you also give extra credit for assisted 3-point FG -- since in fact that's better than a 2-pt FG -- Nash gets even more credit for his assists.

Boxscore rates aren't lacking so much when you make these readily available adjustments.

Posted: **Thu Jul 02, 2015 9:03 pm**

Shot defense is the main problem issue, with any metric or non-use of metrics. RPM and player tracking plus minus make the strongest efforts at capturing player shot defense impact. A blend of some kind is probably best.

Posted: **Thu Jul 02, 2015 10:24 pm**

Agreed re: defense. But there are elements of box scores -- at the team level-- which could and should be utilized, to improve most all box score metrics.
Most aggravating is the lack of counted team rebounds -- those not assigned to individuals, thus assumed to not exist, or to be equal for both teams. Not those silly missed-first-FT "team offensive rebounds", but actual possessions gained or retained by contesting missed shots.

These numbers could be gleaned from pbp, and assigned proportionally to individual rebounds, as padding. Of course this creates fractional 'equivalent rebounds'; but over the course of a season, who cares?
Easier still would be having official 'team rebounds' without the phantom ones.

Posted: **Thu Jul 02, 2015 10:42 pm**

Don't recall what Evan does on that subject with EZPM. Is it still the only PBP based statistical metric? I think so. Could be worthy of further use, study, discussion, adjustment.

APBRmetrics

Players that get underrated by Box Score stats

Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats

Re: Players that get underrated by Box Score stats