Most Improved 2012-13

[Mike G]

Ranked by current eWins above and beyond what players got at last year's rates, in this season's minutes.
Season about 15% over.

Code: Select all

eW+    per36 rates    tm   Eff%   Sco   Reb   Ast   e484  2012 Eff%   Sco   Reb   Ast   e484
.58   Morris,Marcus  Hou  .543   14.9   7.6   1.3    .84      .345    7.9   3.9    .7   -.16
.55 Stephenson,Lance Ind  .591   14.0   4.2   4.0    .88      .397    7.0   3.7   3.6    .14
.54   Kidd,Jason     NYK  .763   15.1   4.6   5.2   2.07  Dal .521    8.7   5.4   7.3   1.23
.48   Dragic,Goran   Phx  .575   19.6   3.2   8.1   1.74  Hou .558   16.8   3.6   7.0   1.27
.47  Aminu,Al-Farouq NOH  .553   13.7   9.5   2.5   1.17      .471    9.5   7.7   1.3    .58

.45   Bledsoe,Eric   LAC  .568   21.4   5.6   4.5   1.37      .446    8.8   4.7   4.1    .52
.39 Varejao,Anderson Cle  .540   14.5  15.9   3.1   1.92      .533   12.9  14.1   1.9   1.54
.37   Landry,Carl    GSW  .637   22.0   9.4   1.4   1.47  NOH .564   19.3   7.9   1.1   1.05
.36   Morris,Darius  LAL  .503   11.3   3.0   4.1    .46      .495    8.4   2.9   3.4   -.16
.36   Mayo,O.J.      Dal  .626   25.3   3.5   3.0   1.22  Mem .507   17.0   4.3   3.3    .90

eW+    per36 rates    tm   Eff%   Sco   Reb   Ast   e484  2012 Eff%   Sco   Reb   Ast   e484
.35   Conley,Mike    Mem  .591   19.4   3.4   8.1   1.70      .516   14.6   2.9   7.0   1.31
.35   Metta W Peace  LAL  .580   15.9   5.7   2.1    .95      .472   10.5   4.7   2.7    .64
.35   Hayes,Chuck    Sac  .500    6.2   9.0   3.5   1.01      .454    4.9   7.4   2.2    .43
.34   Bryant,Kobe    LAL  .621   31.8   5.3   5.2   1.87      .517   28.5   5.4   4.0   1.61
.34   Felton,Raymond NYK  .524   19.7   2.9   8.5   1.37  Por .486   13.4   3.0   7.4    .95

.34   Teague,Jeff    Atl  .546   20.0   3.5   8.9   1.59      .544   15.9   2.9   5.5   1.20
.33 Parsons,Chandler Hou  .586   16.3   6.9   3.6   1.21      .509   12.5   6.5   2.8    .93
.32   Sanders,Larry  Mil  .586   12.9  10.8   1.3   1.33      .458    8.4   7.6   1.5    .76
.32   Mullens,Byron  Cha  .477   14.3   9.9   1.4   1.02      .468   13.4   7.9   1.1    .68
.31   Walker,Kemba   Cha  .488   17.7   4.0   5.3   1.15      .457   14.8   4.7   4.4    .86

eW+    per36 rates    tm   Eff%   Sco   Reb   Ast   e484  2012 Eff%   Sco   Reb   Ast   e484
.29   Gasol,Marc     Mem  .591   17.5   7.5   5.5   1.91      .530   16.3   9.7   3.3   1.63
.28   Davis,Glen     Orl  .446   17.5  10.7   2.4   1.11      .458   13.7   8.4   1.2    .79
.28   Williams,Mo    Uta  .498   16.6   2.6   7.2   1.07  LAC .526   16.9   2.5   3.5    .76
.28  Thabeet,Hasheem Okl  .792   11.0   9.4    .2   1.49  Por .518    7.7   9.7    .0    .59
.28   Holiday,Jrue   Phi  .509   19.0   3.9   8.0   1.30      .492   16.5   3.6   4.6   1.07

.28   Moore,E'Twaun  Orl  .512   14.8   3.7   4.2    .62  Bos .468   10.7   3.1   3.0    .26
.28  Webster,Martell Was  .594   15.0   5.9   2.2   1.04  Min .526    9.9   5.3   1.3    .44
.28   Lowry,Kyle     Tor  .573   23.5   7.0   7.3   2.08  Hou .547   17.4   5.5   7.2   1.56
.26   Pargo,Jeremy   Cle  .578   20.2   2.5   4.2    .94  Mem .385    8.8   2.8   4.1   -.04
.40    Reggie Evans  Brk  .608    6.1  14.9   1.0   1.61  LAC .484    4.2  11.8    .6    .68 

EDIT: R Evans slipped out, now reinserted at the bottom (out of order)
The formula is basically: eW+ = (2013e484 - 2012e484) * Min/484
Players with few minutes this season will not rank near the top or the bottom of the list.

I have to admit I would not have noticed Marcus Morris' emergence, without having generated this list. Last year, he appeared in just 17 games, totaling 126 minutes.
eWins suggests he has improved from a sub-replacement player (-.16) to almost average (.84)
Win Shares says he went from catastrophic (-.083) to 3rd-best player on a good team (.146)
His PER is from 3.8 to 16.0, also #3 on the team.

Here's the bottom of the list, with those at the bottom being most short of last year's eWin rate.

Code: Select all

eW+     per36 rates    tm   Eff%   Sco   Reb   Ast   e484  2012 Eff%   Sco   Reb   Ast   e484
-.28   George,Paul    Ind  .520   17.9   6.7   3.1   1.06      .547   16.5   7.2   3.3   1.37
-.28   Thompson,Klay  GSW  .470   15.9   4.0   2.3    .47      .541   18.0   3.5   2.6    .77
-.29   Hawes,Spencer  Phi  .464   13.2   9.8   2.3   1.05      .512   15.7  10.7   3.6   1.60
-.31   Garnett,Kevin  Bos  .548   20.5  10.6   2.0   1.44      .543   21.7  10.3   3.6   1.88
-.31   Lee,Courtney   Bos  .515    8.2   3.6   1.9    .12  Hou .530   13.7   3.4   1.8    .63

-.32   Curry,Stephen  GSW  .538   20.6   4.3   5.4   1.18      .599   20.4   4.5   6.2   1.52
-.34   Novak,Steve    NYK  .515    9.0   2.4    .6   -.04      .679   16.8   3.5    .4    .66
-.35   Brand,Elton    Dal  .398    8.2  10.0   2.6    .87  Phi .516   15.7   9.3   2.0   1.50
-.38   Ginobili,Manu  SAS  .504   17.4   3.7   6.0   1.06      .656   21.9   5.2   6.6   1.82
-.41   Dudley,Jared   Phx  .486   10.4   3.3   2.2    .39      .567   15.6   5.6   2.0    .97

-.41   Iguodala,Andre Den  .525   16.1   6.7   3.5   1.07  Phi .527   14.8   6.5   5.2   1.53
-.41  Bargnani,Andrea Tor  .501   19.6   5.1   1.6    .81      .528   23.4   6.6   1.9   1.28
-.44   Howard,Dwight  LAL  .559   19.8  11.2   2.1   1.79  Orl .550   22.1  14.9   2.0   2.27
-.45   Stuckey,Rodney Det  .415   11.1   3.7   4.2    .49      .536   19.5   3.5   4.1   1.11
-.45   James,Lebron   Mia  .582   27.3   9.6   6.7   2.13      .593   31.4   8.6   6.7   2.65

eW+     per36 rates    tm   Eff%   Sco   Reb   Ast   e484  2012 Eff%   Sco   Reb   Ast   e484
-.48   Johnson,Joe    Brk  .487   15.9   3.3   3.7    .74  Atl .551   22.2   4.1   4.0   1.34
-.49   Hibbert,Roy    Ind  .395   11.5  10.6   1.9    .98      .530   17.1  11.2   2.4   1.59
-.52   Gasol,Pau      LAL  .475   13.6   9.7   3.5   1.14      .539   18.5  10.7   3.5   1.68
-.54   Johnson,James  Sac  .379    7.7   4.3   2.1    .17  Tor .492   13.6   7.2   2.6   1.08
-.54 Aldridge,Lamarcu Por  .471   18.2   7.8   3.0   1.01      .551   23.2   8.7   2.5   1.63

-.57   Ilyasova,Ersan Mil  .352    8.3   8.1   2.1    .44      .568   17.6  11.5   1.5   1.52
-.58   Lawson,Ty      Den  .437   12.5   2.4   7.3    .83      .571   18.1   4.2   6.8   1.45
-.58   Wade,Dwyane    Mia  .511   20.4   5.2   5.4   1.27      .549   27.9   5.9   5.7   2.15
-.64   Lin,Jeremy     Hou  .432   10.8   5.1   7.0    .98  NYK .540   21.5   4.3   8.9   1.72
-.65   Smith,Josh     Atl  .447   16.9   8.0   3.8   1.08      .490   21.2  10.7   4.2   1.95

NBA average eWins/484 min is 1.00

[Crow]

Most of the top 10 improvements were surprises to me. Most of the biggest declines don't seem as surprising.

[Mike G]

That's the nature of the beast. Those at the bottom are still getting lots of minutes, despite massive drops in production. Those at the top include some main men playing noticeably better -- Varejao, Kidd, Conley -- but mostly they were near-nobodies last year, and now getting minutes and good numbers.

Felton just seems to play well when he's with the Knicks.

I am surprised no one seems to do such a comprehensive "improvement" list using Win Shares or other measure. When I have looked at WS changes of the leaders (and losers), they generally seem to be more extreme than with eWins.

[Crow]

Maybe Neil or someone else will do one if /when they feel like it.

I might construct a season to season XRAPM comparison later, perhaps deeper into the schedule.

[Mike G]

I think there have been some who just show per-minute improvement, with some arbitrary minute cutoff. This is unfortunate, since players are arbitrarily included or excluded.
The beauty of an individual Wins assessment is that it's productivity X minutes. Whether absolute or relative to something (like last season), you never have to resort to any minimum # of minutes.

[Crow]

Both methods have rationales but yeah the latter is probably not used enough.

[DSMok1]

I think there have been some who just show per-minute improvement, with some arbitrary minute cutoff. This is unfortunate, since players are arbitrarily included or excluded.
The beauty of an individual Wins assessment is that it's productivity X minutes. Whether absolute or relative to something (like last season), you never have to resort to any minimum # of minutes.

Win Shares or VORP would do this without any issue. I looked at both methods together last year: http://godismyjudgeok.com/DStats/aspm-a ... omparison/

[Crow]

More good stuff to re-visit.

[talkingpractice]

Thinking qualitatively about what "Most Improved" really implies, I think it mostly implies that the improvement was something that was speculated a priori (such as a large/discontinuous leap in performance for a young player), so that the performance improvement is more likely to be real rather than just variance/noise. Thinking of it in that limited sense, comparing our pre-season prior to each players in-season performance, the answer to the question would be

- Eric Bledsoe (2013 = 1.7, Prior = -1.3)
- Serge Ibaka (2013 = 2.9, Prior = 0.3)
- DeAndre Jordan (2013 = 2.8, Prior = 0.2)
- JaVale McGee (2013 = 0.7, Prior = -1.7)
- Tristan Thompson (2013 = 0, Prior = -2.2)

Ibaka's prior feels low to me now, if I'm playing Monday morning qb.

Other players who have way outplayed their prior but probably haven't really improved per se would be

- Reggie Evans
- Maxiell
- Delfino
- etc

[Crow]

Of the 5 players in your first grouping, only Bledsoe played significantly more minutes than the previous season and it was still under his rookie season level. All did take more shots per minute; by team design or not, I don't know.

[talkingpractice]

Two more now (that I forgot before).... I noticed them when we updated numbers.

- Kosta Koufos (2013 = 2.27, Prior = 0.65)
- Larry Sanders (2013 = 0.29, Prior = -1.99)

xDRAPM likes Koufos too.

[v-zero]

Here's the top twenty most improved so far according to the SEBF model, the left rating is their season-end rating last year, the right is up to now.

Code: Select all

Lance Stephenson 	-3.73	-1.30
Eric Bledsoe 	-0.43	1.75
DeMarre Carroll 	-0.41	1.48
Larry Sanders 	0.61	2.47
Andray Blatche 	-1.02	0.52
Jason Kidd 	0.59	2.07
Byron Mullens 	-3.32	-2.06
Carl Landry 	-1.48	-0.27
J.J. Hickson 	-0.61	0.55
Tristan Thompson 	-1.64	-0.69
Luke Babbitt 	-2.96	-2.01
Jordan Hill 	0.47	1.42
Kemba Walker 	-2.36	-1.43
Ryan Hollins 	-1.54	-0.67
Chris Duhon 	-1.56	-0.70
Reggie Jackson 	-1.41	-0.56
Quincy Pondexter 	-1.00	-0.16
Nikola Vucevic 	0.52	1.36
Antawn Jamison 	-1.42	-0.60
Raymond Felton 	-0.53	0.29

And the worst:

Code: Select all

Taj Gibson 	1.73	0.86
Ersan Ilyasova 	1.63	0.74
Isaiah Thomas 	-0.88	-1.77
Kevin Seraphin 	-0.21	-1.13
Rodrigue Beaubois 	-0.12	-1.08
Hakim Warrick 	-1.97	-2.94
Cartier Martin 	-1.38	-2.39
Wilson Chandler 	-0.26	-1.33
Ian Mahinmi 	0.21	-0.87
Brendan Haywood 	1.27	0.18
Jason Smith 	0.00	-1.08
James Johnson 	0.38	-0.73
Omer Asik 	2.18	0.94
Gustavo Ayon 	1.93	0.58
Jannero Pargo 	-0.89	-2.31
Keith Bogans 	-0.70	-2.24
Jared Jeffries 	1.30	-0.24
C.J. Miles 	-1.58	-3.13
John Lucas 	-0.23	-1.89
Greg Stiemsma 	2.90	1.00

There's a cutoff of about 250 minutes for last season, so I imagine there are some odd names bandying about.

And yes, I'm sorry about the ugly tables.

[DSMok1]

To get tables formatted correctly, save through Excel as a "Formatted Text (Space Delimited)*.prn" file, and then open that file in any text editor to copy in here.

[v-zero]

Thanks, unfortunately I don't use Excel (and couldn't even if I wanted to) - usually I'd format is properly in python, but this was just a messy pull from LibreOffice.

[talkingpractice]

These sorts of lists always confuse me as I don't really get what "improved" means in this context, ie do we really think that Kidd or Jamison have "improved" this season. I guess my lists (young guys who have gotten better) are a lot more boring though...