Page 1 of 1

Differences in WS/48, BPM, and PER

Posted: Mon Oct 12, 2015 2:41 pm
by Mike G
http://www.basketball-reference.com/lea ... anced.html

Paring the 2014-15 player list down to those (349) who played at least 600 minutes, I derived estimated win rates (per48) via PER and BPM, simply by minimizing avg difference from WS/48:

perW/48 = (PER-5.75)/92.5
bpmW/48 = (BPM+5.14)/51.4

Both these formulas also produce avg = .100
For PER, this is 15.0; for BPM, it's zero.

Biggest discrepancies between WS/48 and BPM:

Code: Select all

WS/48-bpmW  WS/48   BPM   bpmW       bpmW-WS48  WS/48   BPM   bpmW
Whiteside   .221    -.9   .082        Kobe      .006     .3   .106
Speights    .158   -2.6   .049       Westbrook  .222   11.0   .314
Valanciunas .189    -.5   .090       Josh Smith .048    1.0   .119
Kanter      .145   -2.1   .059     MC-Williams  .017    -.7   .086
Baynes      .166   -1.0   .081        G Harris -.046   -4.3   .016

Landry      .112   -3.6   .030      Stephenson -.026   -3.5   .032
Hansbrough  .156    -.5   .090        Rubio     .033    -.5   .090
Stoudemire  .138   -1.4   .073        Wroten    .014   -1.5   .071
Biyombo     .156    -.4   .092        Batum     .105    3.1   .160
D Howard    .140   -1.2   .077        T Evans   .082    1.9   .137

WS/48-bpmW  WS/48   BPM   bpmW       bpmW-WS48  WS/48   BPM   bpmW
Dedmon      .099   -3.3   .036        Rondo     .037    -.5   .090
T Zeller    .179     .9   .118        Millsap   .020   -1.6   .069
T Thompson  .148    -.6   .088        Cousins   .144    4.7   .191
Humphries   .134   -1.3   .075        Shumpert  .047    -.4   .092
Aldridge    .165     .3   .106        T Allen   .119    3.3   .164

Freeland    .122   -1.8   .065        LeBron    .199    7.4   .244
Pekovic     .069   -4.5   .012        Smart     .077    1.1   .121
T Black     .124   -1.6   .069        Amundson  .017   -2.0   .061
B Lopez     .159     .2   .104        E Payton  .044    -.7   .086
Ajinca      .159     .2   .104        K Walker  .101    2.2   .143
All those on the left are C and PF. On the right are a few 'bigs'.
BPM loves those who rebound and assist. WS loves high% shooters.

Top departures between BPM and PER:

Code: Select all

perW-bpmW   BPM   bpmW    PER   perW     bpmW-perW  BPM   bpmW    PER   perW
Whiteside   -.9   .082   26.2   .221      Dr Green  5.0   .197   16.4   .115
Pekovic    -4.5   .012   16.6   .117      Da Green  5.0   .197   16.5   .116
Speights   -2.6   .049   19.5   .149      Batum     3.1   .160   13.1   .079
Kanter     -2.1   .059   20.3   .157      M Barnes  1.6   .131   11.3   .060
Bargnani   -3.7   .028   16.6   .117      Bogut     4.1   .180   15.8   .109

S Muhammad -1.8   .065   19.9   .153      Roberson   .3   .106    9.1   .036
Stoudemire -1.4   .073   20.3   .157      T Allen   3.3   .164   14.6   .096
B Lopez      .2   .104   22.7   .183      Patterson 3.2   .162   14.6   .096
Boozer     -3.3   .036   16.4   .115      Korver    3.3   .164   14.8   .098
Aldridge     .3   .106   22.8   .184      Smart     1.1   .121   11.0   .057

perW-bpmW   BPM   bpmW    PER   perW     bpmW-perW  BPM   bpmW    PER   perW
Drummond    -.2   .096   21.4   .169      Matthews  3.9   .176   16.1   .112
B Gordon   -6.1  -.019   10.6   .052      Noah      3.4   .166   15.3   .103
Valanciunas -.5   .090   20.6   .161      Crabbe    -.3   .094    8.7   .032
Landry     -3.6   .030   14.9   .099      Ariza     1.9   .137   12.7   .075
D Howard   -1.2   .077   19.2   .145     Westbrook 11.0   .314   29.1   .252

J Hill     -2.6   .049   16.2   .113      Iguodala  1.6   .131   12.3   .071
Villanueva -1.8   .065   17.4   .126      Temple     .8   .116   10.9   .056
Sims       -3.4   .034   14.5   .095      H Walker -1.0   .081    8.0   .024
Vucevic      .5   .110   21.5   .170      Beverley   .8   .116   11.3   .060
N Young    -3.5   .032   14.2   .091      M Miller -3.0   .042    4.6  -.012
Note Whiteside topped the previous list also. And his PER-equivalent win rate is identical to his WS/48.

Finally, PER vs WS/48 differences:

Code: Select all

perW-WS/48   PER   perW   WS/48    WS/48-perW   PER   perW  WS/48
Kobe        17.6   .128   .006      Hansbrough 13.3   .082   .156
Wroten      14.9   .099   .014      Bonner     10.1   .047   .110
Wade        21.4   .169   .086      T Chandler 20.1   .155   .216
LaVine      11.3   .060  -.018      M Leonard  14.8   .098   .158
Josh Smith  12.3   .071  -.006      H Barnes   13.4   .083   .139

C Anthony   21.5   .170   .097      Baynes     15.9   .110   .166
MC-Williams 14.1   .090   .017      Crabbe      8.7   .032   .088
Rose        15.9   .110   .038      Biyombo    15.2   .102   .156
Rubio       15.2   .102   .033      Morrow     14.8   .098   .151
Cousins     25.2   .210   .144      Freeland   12.2   .070   .122

perW-WS/48   PER   perW   WS/48    WS/48-perW   PER   perW  WS/48
Bargnani    16.6   .117   .058      J Jones    11.1   .058   .110
Stephenson   8.8   .033  -.026      D Jordan   21.0   .165   .217
Early        8.4   .029  -.030      Paul       26.0   .219   .270
Hill        16.2   .113   .055      Bogut      15.8   .109   .159
Wiggins     13.9   .088   .034      Dorsey     11.1   .058   .108

Payne        7.7   .021  -.033      Korver     14.8   .098   .148
Boozer      16.4   .115   .062      Dr Green   16.4   .115   .163
MoWilliams  15.4   .104   .052      Curry      28.0   .241   .288
Ellis       16.5   .116   .065      Splitter   18.9   .142   .189
Josh Smith  14.9   .099   .048      Da Green   16.5   .116   .163

Re: Differences in WS/48, BPM, and PER

Posted: Wed Dec 02, 2015 5:19 pm
by Crow
The formulas at the top are useful.

Would be interested in this data for the new season.

If you have everyone's data on the 2 metrics matched up it would be great to see a simple blended metric.

Re: Differences in WS/48, BPM, and PER

Posted: Thu Dec 03, 2015 3:50 am
by Crow
Found this player rating comparison
(by cemitten)

http://thundermoneyball.weebly.com/
second article from top at moment.

Re: Differences in WS/48, BPM, and PER

Posted: Thu Dec 03, 2015 4:53 am
by Mike G
This took 15 minutes.

Code: Select all

xyz48  wins    Player           PER   BPM   WS/48  perW   bpmW
.346   4.71   Stephen Curry    34.4  12.9   .376   .310   .351
.299   3.87  Russell Westbrook 31.8  12.1   .280   .282   .335
.233   3.19   LeBron James     26.3   7.9   .223   .222   .254
.258   3.17   Kawhi Leonard    27.0   8.2   .285   .230   .260
.236   3.13   Kyle Lowry       24.1   9.0   .236   .198   .275

.241   3.07   Blake Griffin    28.2   7.0   .244   .243   .236
.208   2.93   Paul Millsap     24.0   5.9   .213   .197   .215
.242   2.90   Paul George      26.1   7.8   .255   .220   .252
.194   2.89   James Harden     25.3   5.4   .167   .211   .205
.185   2.48   Draymond Green   18.9   5.4   .208   .142   .205

xyz    wins    Player           PER   BPM   WS/48  perW   bpmW
.276   2.43   Kevin Durant     31.4   6.8   .318   .277   .232
.163   2.37   Damian Lillard   21.6   3.7   .147   .171   .172
.184   2.35   Kevin Love       21.8   3.9   .202   .174   .176
.177   2.32   Andre Drummond   25.2   1.7   .187   .210   .133
.196   2.26   Jimmy Butler     21.1   6.0   .204   .166   .217

.181   2.24   Eric Bledsoe     22.7   4.4   .175   .183   .186
.170   2.21   Al Horford       21.5   3.6   .169   .170   .170
.228   2.21   Derrick Favors   26.4   6.2   .239   .223   .221
.178   2.17   Kemba Walker     22.0   3.9   .182   .176   .176
.160   2.16   Brandon Knight   20.6   3.8   .147   .161   .174

xyz    wins    Player           PER   BPM   WS/48  perW   bpmW
.193   2.11   Chris Bosh       22.2   3.9   .226   .178   .176
.180   2.03   Dirk Nowitzki    21.9   3.2   .203   .175   .162
.163   1.96   DeAndre Jordan   19.5   2.8   .186   .149   .154
.201   1.96   Hassan Whiteside 25.9   3.0   .228   .218   .158
.167   1.92   Isaiah Thomas    21.6   2.4   .182   .171   .147

.157   1.92   Carmelo Anthony  21.2   2.6   .153   .167   .151
.191   1.86   Chris Paul       23.5   3.9   .204   .192   .176
.143   1.84   Brook Lopez      21.4   1.3   .134   .169   .125
.158   1.84   Greg Monroe      22.8   2.3   .146   .184   .145
.134   1.82   DeMar DeRozan    19.2    .5   .147   .145   .110
Ranked by wins = xyz*Min/48.
xyz is just the avg of the 3 per48 rates.
Since teams have played 16 to 21 games, the wins column isn't really equivalent for all players.