APBRmetrics

Posted: **Sat May 14, 2011 8:28 am**

Along with their eWins, PER, and Win Shares, partitioned into 1st, 2nd, and 3rd teams.
Followed by next ten in descending order of votes/points. (Maximum is 595)

Code: Select all

votes  player           tm    eW     PER    WS
595   James,Lebron     Mia   17.7   27.3   15.6
593   Howard,Dwight    Orl   15.6   26.0   14.4
593   Rose,Derrick     Chi   15.4   23.5   13.9
551   Bryant,Kobe      LAL   14.0   23.9   10.3
492   Durant,Kevin     Okl   13.4   23.6   12.0

437   Nowitzki,Dirk    Dal   11.8   23.4   11.1
392   Wade,Dwyane      Mia   14.3   25.6   12.8
259   Gasol,Pau        LAL   12.8   23.3   14.7
258   Stoudemire,Amare NYK   11.4   22.7    8.0
184  Westbrook,Russell Okl   13.1   23.6    9.4

157   Paul,Chris       NOH   12.1   23.7   13.9
135  Aldridge,Lamarcus Por   12.0   21.5   11.1
106   Ginobili,Manu    SAS    9.9   21.7    9.9
_67   Randolph,Zach    Mem   11.9   22.6   10.5
_62   Horford,Al       Atl    9.9   20.7   10.1

note: Rondo is out of order = 'should have been' 3rd Team.

votes  player           tm    eW     PER    WS
68   Rondo,Rajon       Bos    7.8   17.1    6.6
55   Pierce,Paul       Bos   10.7   19.7   11.6
53   Anthony,Carmelo Den-NY  11.4   21.7    7.9
48   Love,Kevin        Min   12.1   24.3   11.4
43   Duncan,Tim        SAS    9.2   21.9    7.7

36   Griffin,Blake     LAC   13.5   21.9    9.8
27   Parker,Tony       SAS    9.6   20.4    8.2
22   Garnett,Kevin     Bos    9.6   20.6    9.0
19   Williams,Deron  Uta-NJ   8.0   21.1    7.3
17   Nash,Steve        Phx    9.3   20.8    8.0

Is it unprecedented that 2 all-NBA members in one year were not all-Stars? Both are forwards from the west.
Kendrick Perkins with a 2nd-team vote.

The All-NBA Teams were chosen by a panel of 119 sportswriters and broadcasters throughout the United States and Canada. The media voted for All-NBA First, Second and Third Teams by position with points awarded on a 5-3-1 basis.

http://www.nba.com/2011/news/05/12/all- ... ef:nbahpt2
Bosh had 2 votes, same as Bargnani. Josh Smith with none.

Posted: **Sat May 14, 2011 9:57 am**

Yeah the voters aren't doing so great lately.

Slowly I hope APBRmetrics get more integrated into this sport.

Posted: **Sat May 14, 2011 3:30 pm**

Slowly I hope APBRmetrics get more integrated into this sport.

I would bet slowly, reluctantly, and only when there's no alternative. There's already something of a backlash against apebear metrics in general and even specific planks of our platform (efficiency rather than PPG, the value of on/off stats) have their detractors/peanut galleries.

The very last people who will adapt are the most visible ones - members of the media who have no intention of changing how they do their job, what they write about, or adaptation of any sort.

Posted: **Sat May 14, 2011 4:44 pm**

It eventually happened in baseball, and it's slowly starting to happen in basketball. The Sloan Conference seems to be helping. ESPN and TNT now routinely mentions +/- stats (although haven't once seen adjusted +/-). I think it just takes time.

Posted: **Sat May 14, 2011 5:24 pm**

EvanZ wrote:It eventually happened in baseball, and it's slowly starting to happen in basketball. The Sloan Conference seems to be helping. ESPN and TNT now routinely mentions +/- stats (although haven't once seen adjusted +/-). I think it just takes time.

Yeah based on how awards are voted on in baseball, APBRmetrics could have a future in basketball.

Plus generations seem to mature as time passes, I believe they will be more nuanced eventually.

Posted: **Sat Jun 11, 2011 3:09 am**

Oh Only Miami Heat and Dallas mavericks remain.

Posted: **Fri Jul 29, 2011 4:30 pm**

wow!

...

keep moving dallas!!

Posted: **Wed Aug 17, 2011 7:36 pm**

eh, whatever.
Distilled to rankings of these top 25 vote-getters, the 3 metrics reviewed in terms of their absolute difference from the vote rankings:

Code: Select all

all-NBA voting and rank in these metrics, w abs. difference 
vote rk     rankings by:   eW   PER  WS    eW  PER  WS  Tot
48   19   Love,Kevin       11    4    9      8  15  10   33
36   21   Griffin,Blake     6   14   16     15   7   5   27
157  11   Paul,Chris       10    6    4      1   5   7   13
17   25   Nash,Steve       22   20   21      3   5   4   12
392   7   Wade,Dwyane       4    3    6      3   4   1    8

22   23   Garnett,Kevin    21   22   18      2   1   5    8
67   15   Randolph,Zach    13   13   12      2   2   3    7
19   24   Williams,Deron   24   19   24      0   5   0    5
27   22   Parker,Tony      20   23   19      2   1   3    4
259   8   Gasol,Pau         9   11    2      1   3   6    2

55   17   Pierce,Paul      17   24    8      0   7   9    2
595   1   James,Lebron      1    1    1      0   0   0    0
53   18   Anthony,Carmelo  16   16   22      2   2   4    0
593   2   Howard,Dwight     2    2    3      0   0   1   -1
43   20   Duncan,Tim       23   15   23      3   5   3   -1

184  10   Westbrook,Russe   8    8   17      2   2   7   -3
135  12   Aldridge,Lamarc  12   18   11      0   6   1   -5
492   5   Durant,Kevin      7    7    7      2   2   2   -6
62   16   Horford,Al       19   21   14      3   5   2   -6
593   2   Rose,Derrick      3    9    5      1   7   3   -11

551   4   Bryant,Kobe       5    5   13      1   1   9   -11
106  13   Ginobili,Manu    18   17   15      5   4   2   -11
437   6   Nowitzki,Dirk    14   10   10      8   4   4   -16
258   9   Stoudemire,Ama   15   12   20      6   3  11   -20
68   14   Rondo,Rajon      25   25   25     11  11  11   -33

             total absolute difference:    81  107  113

I've sorted these by most underrated by voters, relative to the 3 metrics. Most overrated at the bottom.
It seems all-NBA voters should approve of eWins more than either PER or Win Shares (bottom line).

Posted: **Mon Aug 29, 2011 4:45 pm**

Having found a few other all-encompassing type stats to compare with 'conventional wisdom' -- all-NBA voting -- here you may see them in one place.
Dan Statman's rankings were found here:
http://www.pointguardu.com/f136/statman ... post375888
Dan amazingly ranked all 25 top vote recipients as his top 25.

Then I found this by Neil Paine at b-r.com:

PER: Perhaps the most visible APBRmetric stat out there.
Win Shares per 48 minutes (WS48): Basketball-Reference's all-in-one rate stat of choice
Basketball-on-Paper Statistical +/- (bopSPM): An SPM variant explained at the bottom of this page; by using only MPG/ORtg/%Pos/DRtg, it is less biased toward PGs (or any other position) than ordinary SPM
Multiyear Regularized Adjusted +/- (RAPM): The best version of adjusted +/- currently available to the public.

I ran each boxscore metric through the Simple Projection System process, creating an expected rate for each player in 2012. For RAPM, I applied the SPS's aging adjustment to the predictive ranking found at Jeremias' site.

Each player was ranked according to each metric. I then (arbitrarily) applied the "Russian Judge Method" of discarding a player's high and low ranking, added the two remaining ranks, and re-ranked according to those totals.

http://www.basketball-reference.com/blog/?p=9780

So I'm including rapm, bopSPM (abbreviated bps), and WS/48, as well as the Russian Judge composite (calling it simply b-r)
The rankings by all these methods, of the top 25 all-NBA vote-getters:

Code: Select all

vote   rankings by:  eW   PER  WS  ws48 Dan  bps rapm  b-r
1    James,Lebron     1    1    1    1    1    1    1    1
2.5  Rose,Derrick     3    9    5   15    4    9   21   15
2.5  Howard,Dwight    2    2    3    3    2    2    8    2
4    Bryant,Kobe      5    5   13   14   10    6   10    8
5    Durant,Kevin     7    7    7    6   13    4   13  4.5

6    Nowitzki,Dirk   14   10   10    7    9    8    2  6.5
7    Wade,Dwyane      4    3    6    5    3    3    6    3
8    Gasol,Pau        9   11    2    4    7    7   17  6.5
9    Stoudemire,Ama  15   12   20   21   19   14   24   21
10   Westbrook,Russe  8    8   17   25   11   21   15   22

11   Paul,Chris      10    6    4    2    5    5    7  4.5
12   Aldridge,Lamarc 12   18   11   17   15   16   11   14
13   Ginobili,Manu   18   17   15    8   17   13    5  9.5
14   Rondo,Rajon     25   25   25   23   25   23   20   25
15   Randolph,Zach   13   13   12   16   12   17   14   16

16   Horford,Al      19   21   14   13   18   22   22   18
17   Pierce,Paul     17   24    8   10   16   15   12   12
18   Anthony,Carmelo 16   16   22   24   20   11   18   23
19   Love,Kevin      11    4    9    9    6   19   25   11
20   Duncan,Tim      23   15   23   11   21   12    9  9.5

21   Griffin,Blake    6   14   16   20    8   10   19   17
22   Parker,Tony     20   23   19   22   24   24   23   24
23   Garnett,Kevin   21   22   18   12   14   20    3   13
24   Williams,Deron  24   19   24   19   22   18   16   19
25   Nash,Steve      22   20   21   18   23   25    4   20

And the departures from the voting rank, for each stat:

Code: Select all

rk     rankings by:  eW  PER   WS  ws48  Dan  bps  b-r rapm
1     James,Lebron    0    0    0    0    0    0    0    0
2.5  Rose,Derrick    .5   6.5  2.5 12.5  1.5  6.5 12.5 18.5
2.5  Howard,Dwight   .5    .5   .5   .5   .5   .5  0.5  5.5
4     Bryant,Kobe     1    1    9   10    6    2    4    6
5     Durant,Kevin    2    2    2    1    8    1    .5   8

6     Nowitzki,Dirk   8    4    4    1    3    2    .5   4
7     Wade,Dwyane     3    4    1    2    4    4    4    1
8     Gasol,Pau       1    3    6    4    1    1   1.5   9
9     Stoudemire,Ama  6    3   11   12   10    5   12   15
10    Westbrook,Russe 2    2    7   15    1   11   12    5

11    Paul,Chris      1    5    7    9    6    6   6.5   4
12    Aldridge,Lamarc 0    6    1    5    3    4    2    1
13    Ginobili,Manu   5    4    2    5    4    0   3.5   8
14    Rondo,Rajon    11   11   11    9   11    9   11    6
15    Randolph,Zach   2    2    3    1    3    2    1    1

16    Horford,Al      3    5    2    3    2    6    2    6
17    Pierce,Paul     0    7    9    7    1    2    5    5
18    Anthony,Carmelo 2    2    4    6    2    7    5    0
19    Love,Kevin      8   15   10   10   13    0    8    6
20    Duncan,Tim      3    5    3    9    1    8  10.5  11

21    Griffin,Blake  15    7    5    1   13   11    4    2
22    Parker,Tony     2    1    3    0    2    2    2    1
23    Garnett,Kevin   2    1    5   11    9    3   10   20
24    Williams,Deron  0    5    0    5    2    6    5    8
25    Nash,Steve      3    5    4    7    2    0    5   21
     
rk     rankings by:  eW  PER   WS  ws48  Dan  bps  b-r rapm
total (abs)          81  107  112  146  109   99  128  172
average  (abs)      3.2  4.3  4.5  5.8  4.4  4.0  5.1  6.9

The new bopSPM seems to jive more with conventional wisdom than either WS or WS/48.
Some of these, of course, are per-minute summary stats.

Posted: **Tue Aug 30, 2011 7:37 am**

Mike G wrote:Having found a few other all-encompassing type stats to compare with 'conventional wisdom' -- all-NBA voting -- here you may see them in one place.
Dan Statman's rankings were found here:
http://www.pointguardu.com/f136/statman ... post375888
Dan amazingly ranked all 25 top vote recipients as his top 25.

That is very weird - I never realized my top 25 was the same as the top 25 vote getters.

Thanks for putting my metric in with the others here - it's interesting to how they vary from each other.

It appears my ratings liked the youngins (Rose, Love, & Griffin) more than most of the other metrics.

Posted: **Wed Aug 31, 2011 6:39 pm**

Thought I would add mine to the mix (last column):

Code: Select all

vote   rankings by:  eW   PER  WS  ws48 Dan  bps rapm  b-r	ezPM
1    James,Lebron     1    1    1    1    1    1    1    1     2
2.5  Rose,Derrick     3    9    5   15    4    9   21   15    10 
2.5  Howard,Dwight    2    2    3    3    2    2    8    2     1
4    Bryant,Kobe      5    5   13   14   10    6   10    8    12
5    Durant,Kevin     7    7    7    6   13    4   13  4.5    21

6    Nowitzki,Dirk   14   10   10    7    9    8    2  6.5    24
7    Wade,Dwyane      4    3    6    5    3    3    6    3     4
8    Gasol,Pau        9   11    2    4    7    7   17  6.5    14
9    Stoudemire,Ama  15   12   20   21   19   14   24   21    25
10   Westbrook,Russe  8    8   17   25   11   21   15   22     9

11   Paul,Chris      10    6    4    2    5    5    7  4.5     3
12   Aldridge,Lamarc 12   18   11   17   15   16   11   14    20
13   Ginobili,Manu   18   17   15    8   17   13    5  9.5     7
14   Rondo,Rajon     25   25   25   23   25   23   20   25     8
15   Randolph,Zach   13   13   12   16   12   17   14   16    11

16   Horford,Al      19   21   14   13   18   22   22   18    18
17   Pierce,Paul     17   24    8   10   16   15   12   12    13
18   Anthony,Carmelo 16   16   22   24   20   11   18   23    19
19   Love,Kevin      11    4    9    9    6   19   25   11     6
20   Duncan,Tim      23   15   23   11   21   12    9  9.5    16

21   Griffin,Blake    6   14   16   20    8   10   19   17    23
22   Parker,Tony     20   23   19   22   24   24   23   24    17
23   Garnett,Kevin   21   22   18   12   14   20    3   13    22
24   Williams,Deron  24   19   24   19   22   18   16   19    15
25   Nash,Steve      22   20   21   18   23   25    4   20     5

Posted: **Wed Aug 31, 2011 8:33 pm**

Thanks, Evan. Here's how ezpm's ranks differ from the rank by all-NBA votes:

Code: Select all

rankings by:   eW  PER   WS  ws48  Dan  bps  b-r rapm ezpm
total (abs)    81  107  112  146  109   99  128  172  170
average (abs) 3.2  4.3  4.5  5.8  4.4  4.0  5.1  6.9  6.8

All measures had Amare' lower than his #9 (2nd team) vote total, but ezpm has him weakest among the 25 players.

Posted: **Thu Sep 01, 2011 1:03 pm**

I'm not sure whether it's better to be closer to the voters or farther away. ezPM is closest to RAPM fwiw.

Posted: **Thu Sep 01, 2011 7:52 pm**

I noticed that by RAPM, Kevin Love is ranked #320 in the league. He's just 25th among these 25.
RAPM and ezPM have some stark contrasts:

Code: Select all

vote  rankings by:  ezpm  rapm
6    Nowitzki,Dirk   24     2
19   Love,Kevin       6    25
23   Garnett,Kevin   22     3
14   Rondo,Rajon      8    20
2.5  Rose,Derrick    10    21
12   Aldridge,Lamarc 20    11
...

So it's not as though these two metrics have very similar conclusions, just that they are similarly distant from popular opinion.

Conceivably, there's a metric that ranks players very close to what their votes ranked them.
EFF, that sum of per-game stats, is just about average among these stats, at predicting the vote -- 4.8 abs dev. The bottom 6 EFF are Duncan, Parker, Ginobili, Pierce, Rondo, and Garnett, among the 25.
Apparently voters understand that elite defensive teams tend to have understated per-game averages.

Posted: **Fri Sep 02, 2011 2:02 am**

Maybe I'll run a rank correlation between each metric.

APBRmetrics

2011 all-NBA teams

2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams

Re: 2011 all-NBA teams