With draft season upon us, my spreadsheets are working overtime. All of the great work vjl110 has been doing on the draft apparently couldn't take away all of the nostalgia for the MDSD. So I thought I'd provide my (tweaked) version of it and a summary of other statistics for the college players in Draft Express' top 100.
The MDSD (Madison Dan Standard Deviation) score is an ad hoc means of determining which players are "doing stuff." This year, I've changed the categories up a bit. In the past, I've given no credit to scoring volume (just efficiency, via TS%). Now I'm including it as points scored per 40 minutes times TS%. I've consolidated steals and blocks into one category that is the sum of the two. I've consolidated assists and turnovers by including the difference between them as a category. Rebounds continue to be included as its own category. (All of these are pace-adjusted per 40 minutes.)
For each category and player, I calculate: (player's stat - average for position) / standard deviation for position.
Position assignments are taken from DX. Some players (e.g., McCollum) are cross-listed. The MDSD score is simply the sum across the categories. A score of zero indicates that a player is average for his position, relative to the players included in DX's database.
This is frankly pretty dumb because it assumes that all categories carry equal weight. Still, I find it to be a useful way of finding "active" players and seeing who is good at what. (I look at each category's score, though I haven't included them here for compactness.) Note that you can't compare these to MDSD scores from past years (different data, methods, and samples), nor is it a good idea to compare across positions.
ASPM is a step up from the MDSD because it is based on "weights" derived from regressions of NBA RAPM scores on box score stats. So ASPM could be considered a predictor of NBA success if the player puts up the same stats in the NBA that he put up in college.
That's still not great, though, because they probably won't do that. That's where vjl110's work is really cool: he looks at how college stats and player characteristics relate to NBA success based on the historical relationships between the two. He also accounts for strength of schedule and age, which are missing from both MDSD and ASPM. Neat!
I've also thrown in WS/48 and PER, which are familiar (if flawed) measures of performance. To the tables!
|Nate Wolters||South Dakota St||5.04||9.10||0.307||33.3|
|Lorenzo Brown||N.C. State||3.15||6.02||0.143||20.9|
|Erick Green||Virginia Tech||1.89||9.39||0.259||32.3|
|Shane Larkin||Miami FL||1.65||8.52||0.240||21.0|
|Matthew Dellavedova||Saint Mary's||0.60||6.42||0.209||23.1|
|Isaiah Canaan||Murray State||0.49||4.24||0.221||24.0|
McCollum, Wolters, and MCW all look better than Burke according to the MDSD score, but Burke comes out on top by ASPM. No real shockers here compared to vjl110's work: the same guys all look good.
|Jamaal Franklin||San Diego State||5.06||7.44||0.297||24.0|
|Carrick Felix||Arizona State||3.07||5.33||0.223||21.7|
|Erick Green||Virginia Tech||2.27||9.39||0.259||32.3|
|Matthew Dellavedova||Saint Mary's||1.35||6.42||0.209||23.1|
|Isaiah Canaan||Murray State||0.99||4.24||0.221||24.0|
|Michael Snaer||Florida State||-0.34||3.37||0.152||18.3|
|Tim Hardaway Jr.||Michigan||-0.59||5.72||0.219||19.6|
Jamaal Franklin's rebounding pushes him up the SG list. Oladipo, KCP, and McCollum all look good. McLemore and Hardaway, not so much. Though notice how ASPM sees a lot more good in McLemore than MDSD does. Is Erick Green someone we should pay more attention to?
|James Ennis||Long Beach State||2.27||6.44||0.263||26.9|
|Reggie Bullock||North Carolina||2.18||7.70||0.270||23.9|
|C.J. Leslie||N.C. State||0.29||3.57||0.203||21.0|
|Deshaun Thomas||Ohio State||0.22||7.06||0.242||25.7|
|Will Clyburn||Iowa State||-0.02||4.23||0.215||21.1|
|Tony Snell||New Mexico||-0.91||3.57||0.173||15.1|
As expected, Porter and Oladipo top the SF list. Check out Oladipo's crazy high ASPM. WANT!
MDSD hates Muhammad, as does its creator.
|Richard Howell||N.C. State||2.84||7.29||0.212||25.8|
|Jackie Carmichael||Illinois State||1.90||6.83||0.303||28.5|
|Robert Covington||Tennessee State||1.48||5.40||0.227||28.1|
|Tony Mitchell (North Texas)||North Texas||0.04||1.34||0.192||21.0|
|Kenny Kadji||Miami FL||-0.29||5.00||0.263||20.7|
|Deshaun Thomas||Ohio State||-0.74||7.06||0.242||25.7|
|Will Clyburn||Iowa State||-1.18||4.23||0.215||21.1|
|C.J. Leslie||N.C. State||-1.38||3.57||0.203||21.0|
These are some strange results. DJ Stephens is the star of the group? Bennett is mediocre? Hmmm.
|Jack Cooley||Notre Dame||3.78||7.79||0.308||30.6|
|Colton Iverson||Colorado State||2.51||7.31||0.362||26.7|
Muscala is an MDSD superstar. Noel looks very good. In perhaps the biggest departure from vjl110's work, Zeller does not look that great. (If you were wondering why I've run hot and cold on him, there you go.) Interestingly, ASPM really likes Zeller.
So there you have it, folks. I doubt I'll look into the performance versus top 100 teams as I did last year. The samples tend to be too small, and it takes some work to put it all together. I might weigh in with some case-by-case analyses, but this one seems pretty clear to me: McCollum, KCP, or Zeller at 9 (in that order?); and Muscala, Wolters, or Bullock at 26 (depends on the pick at 9).
It's a fine year to grab front court depth in the second round, with Kazemi, Mbakwe, and Roberson all looking good. DJ Stephens could be a very entertaining flyer pick. He's been compared to Hoopus favorite Jeremy Evans. Don't screw it up, Flip!
(Revision: I initially forgot to credit DSMok1 for his fine work producing the ASPM stats. I also took the PER and Win Shares values from his spreadsheet. All of this can be found here.)