Minute Projections for players of the 2014 NBA DRAFT

[J.E.]

This is going to be a gigantic post, so I made a new thread instead of adding to the draft model thread

The purpose of this was to project 'Minutes played per NBA season since draft year' for players that have entered the 2014 NBA draft and have played at least one year of college. No combine data was used except for height and weight

Starting with 2003 I'm going to show what the model would have predicted back then, so people can get a feel for how good this works in reality. The predictions use absolutely no data other than what was available at the time of the draft. So, the model that 'predicts' the 2003 NBA draft only uses data from April 2003 and earlier, etc. As such, the model coefficients are going to be different for every year. Since this consitutes real out-of-sample-testing there are going to be a bunch of misses

2003

Code: Select all

╔══════╦════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name               ║ Actual MP ║ Projected MP ║
╠══════╬════════════════════╬═══════════╬══════════════╣
║ 1    ║ Carmelo Anthony    ║ 2687      ║ 1652         ║
║ 2    ║ Nick Collison      ║ 1544      ║ 1490         ║
║ 3    ║ Kyle Korver        ║ 1825      ║ 1346         ║
║ 4    ║ Kirk Hinrich       ║ 2457      ║ 1283         ║
║ 5    ║ Mike Sweetney      ║ 451       ║ 1152         ║
║ 6    ║ Keith Bogans       ║ 1621      ║ 1137         ║
║ 7    ║ Dwyane Wade        ║ 2567      ║ 1120         ║
║ 8    ║ Matt Bonner        ║ 1161      ║ 1058         ║
║ 9    ║ Josh Howard        ║ 1769      ║ 1019         ║
║ 10   ║ Chris Bosh         ║ 2701      ║ 994          ║
║ 11   ║ Troy Bell          ║ 4         ║ 976          ║
║ 12   ║ Reece Gaines       ║ 75        ║ 962          ║
║ 13   ║ David West         ║ 2145      ║ 886          ║
║ 14   ║ Chris Kaman        ║ 1832      ║ 884          ║
║ 15   ║ Brian Cook         ║ 671       ║ 792          ║
║ 16   ║ Jarvis Hayes       ║ 1237      ║ 792          ║
║ 17   ║ Mo Williams        ║ 2052      ║ 733          ║
║ 18   ║ Dahntay Jones      ║ 944       ║ 729          ║
║ 19   ║ Jerome Beasley     ║ 0         ║ 634          ║
║ 20   ║ Luke Ridnour       ║ 1946      ║ 630          ║
║ 21   ║ Steve Blake        ║ 1799      ║ 569          ║
║ 22   ║ Willie Green       ║ 1356      ║ 551          ║
║ 23   ║ Marcus Banks       ║ 696       ║ 503          ║
║ 24   ║ Luke Walton        ║ 1058      ║ 490          ║
║ 25   ║ T.J. Ford          ║ 1461      ║ 410          ║
║ 26   ║ Jason Kapono       ║ 1099      ║ 280          ║
║ 27   ║ Brandon Hunter     ║ 78        ║ 273          ║
║ 28   ║ James Jones        ║ 1057      ║ 231          ║
║ 29   ║ Travis Hansen      ║ 63        ║ 103          ║
║ 30   ║ Derrick Zimmerman  ║ 4         ║ -17          ║
╚══════╩════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.28
Non-zero coefficients in the sparse model: 4.
2 most important coefficients: PTS_tot*SOS, MOV*SOS
The minute projections are generally too low here because players from drafts a couple of years earlier haven't played much (until that point). It's far too low on Wade, Bosh and West; too high on Sweetney, Troy Bell and Reece Gaines

2004

Code: Select all

╔══════╦═══════════════════╦═══════════╦══════════════╗
║ Rank ║ Name              ║ Actual MP ║ Projected MP ║
╠══════╬═══════════════════╬═══════════╬══════════════╣
║ 1    ║ Luol Deng         ║ 2389      ║ 1757         ║
║ 2    ║ Emeka Okafor      ║ 2081      ║ 1649         ║
║ 3    ║ Delonte West      ║ 1480      ║ 1594         ║
║ 4    ║ Jameer Nelson     ║ 1860      ║ 1529         ║
║ 5    ║ Ben Gordon        ║ 2217      ║ 1462         ║
║ 6    ║ Devin Harris      ║ 1907      ║ 1284         ║
║ 7    ║ Chris Duhon       ║ 1838      ║ 1239         ║
║ 8    ║ Andre Emmett      ║ 9         ║ 1152         ║
║ 9    ║ Tony Allen        ║ 1151      ║ 1102         ║
║ 10   ║ Kirk Snyder       ║ 442       ║ 1082         ║
║ 11   ║ Andre Iguodala    ║ 2902      ║ 928          ║
║ 12   ║ Josh Childress    ║ 1288      ║ 887          ║
║ 13   ║ Kris Humphries    ║ 1099      ║ 875          ║
║ 14   ║ Rafael Araujo     ║ 198       ║ 871          ║
║ 15   ║ David Harrison    ║ 335       ║ 763          ║
║ 16   ║ Lionel Chalmers   ║ 54        ║ 742          ║
║ 17   ║ Matt Freije       ║ 73        ║ 690          ║
║ 18   ║ Jackson Vroman    ║ 138       ║ 550          ║
║ 19   ║ Luis Flores       ║ 9         ║ 545          ║
║ 20   ║ Trevor Ariza      ║ 1597      ║ 506          ║
║ 21   ║ Kevin Martin      ║ 1856      ║ 402          ║
║ 22   ║ Antonio Burks     ║ 98        ║ 386          ║
║ 23   ║ Royal Ivey        ║ 682       ║ 338          ║
║ 24   ║ Justin Reed       ║ 179       ║ 308          ║
║ 25   ║ Bernard Robinson  ║ 238       ║ 244          ║
║ 26   ║ Pape Sow          ║ 109       ║ -43          ║
╚══════╩═══════════════════╩═══════════╩══════════════╝

OOS R^2: 0.38
Non-zero coefficients in the sparse model: 7
2 most important coefficients: Age, MOV*SOS
Has Andre Emmet way too high; Iguodala, Trevor Ariza and Kevin Martin way too low. Otherwise pretty solid

2005

Code: Select all

╔══════╦═════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name                ║ Actual MP ║ Projected MP ║
╠══════╬═════════════════════╬═══════════╬══════════════╣
║ 1    ║ Andrew Bogut        ║ 1765      ║ 1903         ║
║ 2    ║ Luther Head         ║ 974       ║ 1729         ║
║ 3    ║ Sean May            ║ 233       ║ 1651         ║
║ 4    ║ Raymond Felton      ║ 2611      ║ 1599         ║
║ 5    ║ Marvin Williams     ║ 2068      ║ 1509         ║
║ 6    ║ Rashad McCants      ║ 629       ║ 1370         ║
║ 7    ║ Jarrett Jack        ║ 2129      ║ 1296         ║
║ 8    ║ Deron Williams      ║ 2616      ║ 1237         ║
║ 9    ║ Salim Stoudamire    ║ 334       ║ 1232         ║
║ 10   ║ Brandon Bass        ║ 1287      ║ 1218         ║
║ 11   ║ Chris Paul          ║ 2534      ║ 1206         ║
║ 12   ║ Channing Frye       ║ 1494      ║ 1184         ║
║ 13   ║ Chris Taft          ║ 18        ║ 1169         ║
║ 14   ║ Ryan Gomes          ║ 1678      ║ 1097         ║
║ 15   ║ Antoine Wright      ║ 714       ║ 1072         ║
║ 16   ║ Danny Granger       ║ 2128      ║ 1015         ║
║ 17   ║ Charlie Villanueva  ║ 1506      ║ 1002         ║
║ 18   ║ Ike Diogu           ║ 349       ║ 1000         ║
║ 19   ║ Nate Robinson       ║ 1549      ║ 986          ║
║ 20   ║ Wayne Simien        ║ 63        ║ 962          ║
║ 21   ║ Daniel Ewing        ║ 210       ║ 946          ║
║ 22   ║ Von Wafer           ║ 309       ║ 917          ║
║ 23   ║ Francisco Garcia    ║ 1321      ║ 903          ║
║ 24   ║ Julius Hodge        ║ 12        ║ 865          ║
║ 25   ║ David Lee '05       ║ 2340      ║ 836          ║
║ 26   ║ Hakim Warrick       ║ 1328      ║ 825          ║
║ 27   ║ Jason Maxiell       ║ 1251      ║ 764          ║
║ 28   ║ Joey Graham         ║ 764       ║ 683          ║
║ 29   ║ Dijon Thompson      ║ 11        ║ 680          ║
║ 30   ║ Bracey Wright       ║ 40        ║ 581          ║
║ 31   ║ Alex Acker          ║ 29        ║ 505          ║
║ 32   ║ Lawrence Roberts    ║ 143       ║ 400          ║
║ 33   ║ Linas Kleiza        ║ 1023      ║ 310          ║
║ 34   ║ Ronny Turiaf        ║ 924       ║ 265          ║
║ 35   ║ Robert Whaley       ║ 26        ║ 207          ║
║ 36   ║ Orien Greene        ║ 194       ║ 131          ║
║ 37   ║ Travis Diener       ║ 332       ║ 81           ║
╚══════╩═════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.2
Non-zero coefficients in the sparse model: 11
2 most important coefficients: DRB_tot*SOS, TS%
Definitely missed on Chris Paul

2006

Code: Select all

╔══════╦══════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name                 ║ Actual MP ║ Projected MP ║
╠══════╬══════════════════════╬═══════════╬══════════════╣
║ 1    ║ LaMarcus Aldridge    ║ 2557      ║ 1695         ║
║ 2    ║ Daniel Gibson        ║ 1164      ║ 1692         ║
║ 3    ║ Rudy Gay             ║ 2626      ║ 1670         ║
║ 4    ║ Tyrus Thomas         ║ 990       ║ 1468         ║
║ 5    ║ Jordan Farmar        ║ 1123      ║ 1424         ║
║ 6    ║ J.J. Redick          ║ 1307      ║ 1404         ║
║ 7    ║ Shelden Williams     ║ 698       ║ 1351         ║
║ 8    ║ Shawne Williams      ║ 535       ║ 1343         ║
║ 9    ║ Shannon Brown        ║ 905       ║ 1308         ║
║ 10   ║ Hilton Armstrong     ║ 422       ║ 1295         ║
║ 11   ║ David Noel           ║ 99        ║ 1286         ║
║ 12   ║ Kyle Lowry           ║ 1818      ║ 1278         ║
║ 13   ║ Marcus Williams '06  ║ 386       ║ 1259         ║
║ 14   ║ Ronnie Brewer        ║ 1442      ║ 1250         ║
║ 15   ║ Paul Davis           ║ 87        ║ 1223         ║
║ 16   ║ James Augustine      ║ 19        ║ 1208         ║
║ 17   ║ P.J. Tucker          ║ 560       ║ 1190         ║
║ 18   ║ Brandon Roy          ║ 1445      ║ 1159         ║
║ 19   ║ Renaldo Balkman      ║ 383       ║ 1138         ║
║ 20   ║ Steve Novak          ║ 674       ║ 1127         ║
║ 21   ║ Hassan Adams         ║ 68        ║ 1099         ║
║ 22   ║ Patrick O'Bryant     ║ 65        ║ 1092         ║
║ 23   ║ Rajon Rondo          ║ 2080      ║ 1075         ║
║ 24   ║ Quincy Douby         ║ 191       ║ 949          ║
║ 25   ║ Rodney Carney        ║ 575       ║ 932          ║
║ 26   ║ Dee Brown '06        ║ 88        ║ 925          ║
║ 27   ║ Paul Millsap         ║ 2162      ║ 919          ║
║ 28   ║ Randy Foye           ║ 1908      ║ 882          ║
║ 29   ║ Josh Boone           ║ 560       ║ 877          ║
║ 30   ║ Craig Smith          ║ 853       ║ 832          ║
║ 31   ║ Cedric Simmons       ║ 88        ║ 795          ║
║ 32   ║ Mardy Collins        ║ 346       ║ 736          ║
║ 33   ║ Bobby Jones '06      ║ 108       ║ 711          ║
║ 34   ║ Ryan Hollins         ║ 654       ║ 696          ║
║ 35   ║ Guillermo Diaz       ║ 2         ║ 571          ║
║ 36   ║ Maurice Ager         ║ 63        ║ 511          ║
║ 37   ║ Will Blalock         ║ 20        ║ 506          ║
║ 38   ║ Adam Morrison        ║ 409       ║ 496          ║
║ 39   ║ James White          ║ 72        ║ 493          ║
║ 40   ║ Solomon Jones        ║ 386       ║ 149          ║
║ 41   ║ Alexander Johnson    ║ 162       ║ 109          ║
║ 42   ║ Leon Powe            ║ 415       ║ -29          ║
╚══════╩══════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.23
Non-zero coefficients in the sparse model: 10
2 most important coefficients: Age, TS%
Too low on Rondo and Millsap, too high on Shelden/Shawne Williams. Has Adam Morrison at #38, got picked 3rd.

2007

Code: Select all

╔══════╦══════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name                 ║ Actual MP ║ Projected MP ║
╠══════╬══════════════════════╬═══════════╬══════════════╣
║ 1    ║ Kevin Durant         ║ 2959      ║ 2188         ║
║ 2    ║ Mike Conley          ║ 2386      ║ 1796         ║
║ 3    ║ Brandan Wright       ║ 637       ║ 1676         ║
║ 4    ║ Corey Brewer '07     ║ 1714      ║ 1619         ║
║ 5    ║ Jeff Green           ║ 2236      ║ 1614         ║
║ 6    ║ Javaris Crittenton   ║ 264       ║ 1544         ║
║ 7    ║ Julian Wright        ║ 435       ║ 1544         ║
║ 8    ║ Joakim Noah          ║ 2023      ║ 1510         ║
║ 9    ║ Taurean Green        ║ 10        ║ 1433         ║
║ 10   ║ Al Horford           ║ 2056      ║ 1388         ║
║ 11   ║ Thaddeus Young       ║ 2216      ║ 1363         ║
║ 12   ║ Josh McRoberts       ║ 999       ║ 1313         ║
║ 13   ║ Greg Oden            ║ 289       ║ 1286         ║
║ 14   ║ Jared Dudley         ║ 1798      ║ 1251         ║
║ 15   ║ Nick Fazekas         ║ 38        ║ 1150         ║
║ 16   ║ Daequan Cook         ║ 856       ║ 1078         ║
║ 17   ║ Demetris Nichols     ║ 8         ║ 1035         ║
║ 18   ║ Acie Law             ║ 340       ║ 1032         ║
║ 19   ║ Arron Afflalo        ║ 1956      ║ 989          ║
║ 20   ║ Spencer Hawes        ║ 1776      ║ 971          ║
║ 21   ║ D.J. Strawberry      ║ 38        ║ 968          ║
║ 22   ║ Marcus Williams '07  ║ 5         ║ 964          ║
║ 23   ║ Aaron Brooks         ║ 1337      ║ 914          ║
║ 24   ║ Derrick Byars        ║ 5         ║ 897          ║
║ 25   ║ Glen Davis           ║ 1423      ║ 865          ║
║ 26   ║ Gabe Pruitt          ║ 65        ║ 808          ║
║ 27   ║ Ramon Sessions       ║ 1723      ║ 793          ║
║ 28   ║ Al Thornton          ║ 1182      ║ 788          ║
║ 29   ║ Wilson Chandler      ║ 1623      ║ 708          ║
║ 30   ║ Alando Tucker        ║ 60        ║ 695          ║
║ 31   ║ Nick Young           ║ 1628      ║ 665          ║
║ 32   ║ Aaron Gray           ║ 550       ║ 658          ║
║ 33   ║ Rodney Stuckey       ║ 2006      ║ 621          ║
║ 34   ║ Sean Williams        ║ 282       ║ 584          ║
║ 35   ║ Carl Landry          ║ 1394      ║ 554          ║
║ 36   ║ Nathan Jawai         ║ 61        ║ 550          ║
║ 37   ║ Jason Smith          ║ 788       ║ 529          ║
║ 38   ║ Morris Almond        ║ 51        ║ 513          ║
║ 39   ║ Chris Richard        ║ 111       ║ 491          ║
║ 40   ║ Dominic McGuire      ║ 764       ║ 355          ║
║ 41   ║ Mike Taylor          ║ 110       ║ 299          ║
║ 42   ║ Jermareo Davidson    ║ 61        ║ 193          ║
║ 43   ║ Stephane Lasme       ║ 43        ║ -134         ║
╚══════╩══════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.22
Non-zero coefficients in the sparse model: 10
2 most important coefficients: Age, MOV*SOS
Durant gets the highest minute-projection yet. Oden is #13

2008

Code: Select all

╔══════╦════════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name                   ║ Actual MP ║ Projected MP ║
╠══════╬════════════════════════╬═══════════╬══════════════╣
║ 1    ║ Kevin Love             ║ 1988      ║ 2303         ║
║ 2    ║ Mario Chalmers         ║ 2013      ║ 1929         ║
║ 3    ║ Michael Beasley        ║ 1695      ║ 1801         ║
║ 4    ║ Ryan Anderson          ║ 1484      ║ 1474         ║
║ 5    ║ Derrick Rose           ║ 1763      ║ 1447         ║
║ 6    ║ Russell Westbrook      ║ 2488      ║ 1349         ║
║ 7    ║ Brandon Rush           ║ 1368      ║ 1316         ║
║ 8    ║ Chris Douglas-Roberts  ║ 713       ║ 1302         ║
║ 9    ║ Kyle Weaver            ║ 229       ║ 1251         ║
║ 10   ║ O.J. Mayo              ║ 2355      ║ 1226         ║
║ 11   ║ Donte Greene           ║ 709       ║ 1186         ║
║ 12   ║ Darnell Jackson        ║ 173       ║ 1124         ║
║ 13   ║ Darrell Arthur         ║ 943       ║ 1087         ║
║ 14   ║ Jerryd Bayless         ║ 1268      ║ 1081         ║
║ 15   ║ Kosta Koufos           ║ 854       ║ 1075         ║
║ 16   ║ Eric Gordon            ║ 1774      ║ 1070         ║
║ 17   ║ D.J. White             ║ 351       ║ 1055         ║
║ 18   ║ Roy Hibbert            ║ 1983      ║ 1051         ║
║ 19   ║ George Hill            ║ 2003      ║ 994          ║
║ 20   ║ D.J. Augustin          ║ 1785      ║ 989          ║
║ 21   ║ Jason Thompson         ║ 2059      ║ 883          ║
║ 22   ║ Marreese Speights      ║ 1106      ║ 870          ║
║ 23   ║ Joe Alexander          ║ 124       ║ 861          ║
║ 24   ║ Courtney Lee           ║ 1951      ║ 842          ║
║ 25   ║ Patrick Beverley       ║ 410       ║ 833          ║
║ 26   ║ Brook Lopez            ║ 1889      ║ 830          ║
║ 27   ║ Robin Lopez            ║ 1370      ║ 763          ║
║ 28   ║ Bill Walker            ║ 398       ║ 762          ║
║ 29   ║ Sonny Weems            ║ 472       ║ 747          ║
║ 30   ║ J.J. Hickson           ║ 1679      ║ 734          ║
║ 31   ║ Malik Hairston         ║ 78        ║ 688          ║
║ 32   ║ J.R. Giddens           ║ 41        ║ 675          ║
║ 33   ║ Sean Singletary        ║ 50        ║ 632          ║
║ 34   ║ DeAndre Jordan         ║ 1772      ║ 601          ║
║ 35   ║ Anthony Randolph       ║ 640       ║ 598          ║
║ 36   ║ Joey Dorsey            ║ 105       ║ 561          ║
║ 37   ║ Luc Mbah a Moute       ║ 1568      ║ 465          ║
║ 38   ║ Patrick Ewing '08      ║ 3         ║ 383          ║
║ 39   ║ JaVale McGee           ║ 1225      ║ 287          ║
║ 40   ║ Joe Crawford           ║ 3         ║ 270          ║
║ 41   ║ Walter Sharpe          ║ 3         ║ -42          ║
╚══════╩════════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.24
Non-zero coefficients in the sparse model: 6
2 most important coefficients: Age, MOV*SOS
Kevin Love gets an ever higher minute-projection than Kevin Durant. Has Ryan Anderson, who got picked 21st, 4th and absolutely nails his projection. Big misses with DeAndre Jordan and Javale McGee

2009

Code: Select all

╔══════╦════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name               ║ Actual MP ║ Projected MP ║
╠══════╬════════════════════╬═══════════╬══════════════╣
║ 1    ║ Ty Lawson          ║ 2056      ║ 1905         ║
║ 2    ║ Tyreke Evans       ║ 2198      ║ 1754         ║
║ 3    ║ Danny Green        ║ 1116      ║ 1718         ║
║ 4    ║ Terrence Williams  ║ 584       ║ 1673         ║
║ 5    ║ Stephen Curry      ║ 2389      ║ 1649         ║
║ 6    ║ Jrue Holiday       ║ 2186      ║ 1609         ║
║ 7    ║ James Harden       ║ 2327      ║ 1528         ║
║ 8    ║ DeJuan Blair       ║ 1331      ║ 1426         ║
║ 9    ║ Earl Clark         ║ 707       ║ 1423         ║
║ 10   ║ Wayne Ellington    ║ 1118      ║ 1398         ║
║ 11   ║ Blake Griffin      ║ 2193      ║ 1343         ║
║ 12   ║ Darren Collison    ║ 2157      ║ 1333         ║
║ 13   ║ Chase Budinger     ║ 1156      ║ 1190         ║
║ 14   ║ Jeff Teague        ║ 1807      ║ 1145         ║
║ 15   ║ Hasheem Thabeet    ║ 471       ║ 1139         ║
║ 16   ║ Jonny Flynn        ║ 747       ║ 1067         ║
║ 17   ║ Tyler Hansbrough   ║ 1166      ║ 976          ║
║ 18   ║ Patrick Mills      ║ 653       ║ 927          ║
║ 19   ║ Jodie Meeks        ║ 1728      ║ 906          ║
║ 20   ║ A.J. Price         ║ 721       ║ 875          ║
║ 21   ║ DaJuan Summers     ║ 164       ║ 862          ║
║ 22   ║ Austin Daye        ║ 757       ║ 858          ║
║ 23   ║ Eric Maynor        ║ 779       ║ 838          ║
║ 24   ║ Marcus Thornton    ║ 1778      ║ 814          ║
║ 25   ║ DeMarre Carroll    ║ 929       ║ 810          ║
║ 26   ║ Derrick Brown      ║ 510       ║ 782          ║
║ 27   ║ Byron Mullens      ║ 689       ║ 753          ║
║ 28   ║ DeMar DeRozan      ║ 2550      ║ 737          ║
║ 29   ║ Jordan Hill        ║ 851       ║ 725          ║
║ 30   ║ James Johnson      ║ 994       ║ 723          ║
║ 31   ║ Dante Cunningham   ║ 1422      ║ 718          ║
║ 32   ║ Jeff Pendergraph   ║ 366       ║ 713          ║
║ 33   ║ Sam Young          ║ 793       ║ 639          ║
║ 34   ║ Taylor Griffin     ║ 6         ║ 494          ║
║ 35   ║ Jermaine Taylor    ║ 157       ║ 433          ║
║ 36   ║ Jon Brockman       ║ 314       ║ 408          ║
║ 37   ║ Toney Douglas      ║ 1135      ║ 351          ║
║ 38   ║ Taj Gibson         ║ 1808      ║ 326          ║
║ 39   ║ Lester Hudson      ║ 107       ║ 231          ║
╚══════╩════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.26
Non-zero coefficients in the sparse model: 11
2 most important coefficients: Age, MOV*SOS
Misses big on DeMar DeRozan and Taj Gibson, and somewhat on Blake Griffin. Has Thabeet at #15, got picked 2nd. Interestingly has Danny Green at #3, who got picked 46th. Had Lawson, who got picked at 18, first and nailed his projection

2010

Code: Select all

╔══════╦════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name               ║ Actual MP ║ Projected MP ║
╠══════╬════════════════════╬═══════════╬══════════════╣
║ 1    ║ John Wall          ║ 2393      ║ 2006         ║
║ 2    ║ Xavier Henry       ║ 704       ║ 1869         ║
║ 3    ║ Greg Monroe        ║ 2420      ║ 1767         ║
║ 4    ║ Evan Turner        ║ 2214      ║ 1749         ║
║ 5    ║ Gordon Hayward     ║ 2034      ║ 1629         ║
║ 6    ║ Wesley Johnson     ║ 1682      ║ 1513         ║
║ 7    ║ Patrick Patterson  ║ 1414      ║ 1491         ║
║ 8    ║ Derrick Favors     ║ 1724      ║ 1485         ║
║ 9    ║ Andy Rautins       ║ 6         ║ 1440         ║
║ 10   ║ DeMarcus Cousins   ║ 2211      ║ 1428         ║
║ 11   ║ Devin Ebanks       ║ 177       ║ 1410         ║
║ 12   ║ Luke Babbitt       ║ 469       ║ 1290         ║
║ 13   ║ Paul George        ║ 2273      ║ 1290         ║
║ 14   ║ Cole Aldrich       ║ 258       ║ 1272         ║
║ 15   ║ Elliot Williams    ║ 326       ║ 1240         ║
║ 16   ║ James Anderson     ║ 899       ║ 1226         ║
║ 17   ║ Eric Bledsoe       ║ 1318      ║ 1225         ║
║ 18   ║ Greivis Vasquez    ║ 1757      ║ 1224         ║
║ 19   ║ Jordan Crawford    ║ 1587      ║ 1215         ║
║ 20   ║ Quincy Pondexter   ║ 812       ║ 1180         ║
║ 21   ║ Ekpe Udoh          ║ 1109      ║ 1136         ║
║ 22   ║ Al-Farouq Aminu    ║ 1760      ║ 1136         ║
║ 23   ║ Darington Hobson   ║ 9         ║ 998          ║
║ 24   ║ Damion James       ║ 155       ║ 995          ║
║ 25   ║ Dominique Jones    ║ 185       ║ 980          ║
║ 26   ║ Landry Fields      ║ 1448      ║ 964          ║
║ 27   ║ Avery Bradley      ║ 1205      ║ 926          ║
║ 28   ║ Trevor Booker      ║ 1191      ║ 847          ║
║ 29   ║ Lazar Hayward      ║ 147       ║ 817          ║
║ 30   ║ Daniel Orton       ║ 135       ║ 808          ║
║ 31   ║ Lance Stephenson   ║ 1396      ║ 788          ║
║ 32   ║ Solomon Alabi      ║ 45        ║ 733          ║
║ 33   ║ Jarvis Varnado     ║ 99        ║ 724          ║
║ 34   ║ Larry Sanders      ║ 1009      ║ 697          ║
║ 35   ║ Armon Johnson      ║ 100       ║ 689          ║
║ 36   ║ Hassan Whiteside   ║ 27        ║ 638          ║
║ 37   ║ Luke Harangody     ║ 217       ║ 599          ║
║ 38   ║ Derrick Caracter   ║ 53        ║ 592          ║
║ 39   ║ Ed Davis           ║ 1430      ║ 580          ║
║ 40   ║ Dexter Pittman     ║ 86        ║ 551          ║
║ 41   ║ Craig Brackins     ║ 30        ║ 510          ║
║ 42   ║ Gani Lawal         ║ 0         ║ 504          ║
║ 43   ║ Willie Warren      ║ 33        ║ 455          ║
║ 44   ║ Jeremy Evans       ║ 526       ║ 367          ║
║ 45   ║ Jerome Jordan      ║ 27        ║ 333          ║
║ 46   ║ Ryan Reid          ║ 4         ║ 109          ║
║ 47   ║ Hamady N'Diaye     ║ 39        ║ -79          ║
╚══════╩════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.42
Non-zero coefficients in the sparse model: 11
2 most important coefficients: Age, MOV*SOS
Very good year for the model with a very high OOS R^2. Misses big with Xavier Henry and Andy Rautins. Not high enough on Paul George and DeMarcus Cousins

2011

Code: Select all

╔══════╦═══════════════════╦═══════════╦══════════════╗
║ Rank ║ Name              ║ Actual MP ║ Projected MP ║
╠══════╬═══════════════════╬═══════════╬══════════════╣
║ 1    ║ Kemba Walker      ║ 2421      ║ 1752         ║
║ 2    ║ Derrick Williams  ║ 1718      ║ 1713         ║
║ 3    ║ Kawhi Leonard     ║ 1755      ║ 1559         ║
║ 4    ║ Brandon Knight    ║ 2298      ║ 1445         ║
║ 5    ║ Jimmer Fredette   ║ 873       ║ 1433         ║
║ 6    ║ Jordan Hamilton   ║ 558       ║ 1399         ║
║ 7    ║ Alec Burks        ║ 1423      ║ 1350         ║
║ 8    ║ Darius Morris     ║ 388       ║ 1345         ║
║ 9    ║ Klay Thompson     ║ 2470      ║ 1234         ║
║ 10   ║ Kenneth Faried    ║ 1821      ║ 1231         ║
║ 11   ║ Marcus Morris     ║ 1150      ║ 1214         ║
║ 12   ║ Nolan Smith       ║ 276       ║ 1200         ║
║ 13   ║ Reggie Jackson    ║ 1257      ║ 1198         ║
║ 14   ║ Jimmy Butler      ║ 1694      ║ 1171         ║
║ 15   ║ Markieff Morris   ║ 1739      ║ 1127         ║
║ 16   ║ Tobias Harris     ║ 1209      ║ 1096         ║
║ 17   ║ JaJuan Johnson    ║ 99        ║ 1088         ║
║ 18   ║ Cory Joseph       ║ 530       ║ 1069         ║
║ 19   ║ Nikola Vucevic    ║ 1727      ║ 1052         ║
║ 20   ║ Tristan Thompson  ║ 2194      ║ 1025         ║
║ 21   ║ Kyrie Irving      ║ 2034      ║ 998          ║
║ 22   ║ Justin Harper     ║ 28        ║ 993          ║
║ 23   ║ Norris Cole       ║ 1621      ║ 992          ║
║ 24   ║ Jon Leuer         ║ 461       ║ 991          ║
║ 25   ║ Iman Shumpert     ║ 1554      ║ 984          ║
║ 26   ║ Travis Leslie     ║ 15        ║ 955          ║
║ 27   ║ Josh Harrellson   ║ 296       ║ 928          ║
║ 28   ║ Tyler Honeycutt   ║ 40        ║ 924          ║
║ 29   ║ Chandler Parsons  ║ 2448      ║ 918          ║
║ 30   ║ Isaiah Thomas     ║ 2091      ║ 916          ║
║ 31   ║ E'Twaun Moore     ║ 1173      ║ 879          ║
║ 32   ║ Jordan Williams   ║ 211       ║ 858          ║
║ 33   ║ Charles Jenkins   ║ 444       ║ 853          ║
║ 34   ║ MarShon Brooks    ║ 958       ║ 813          ║
║ 35   ║ Trey Thompkins    ║ 39        ║ 789          ║
║ 36   ║ Kyle Singler      ║ 1543      ║ 768          ║
║ 37   ║ Josh Selby        ║ 98        ║ 696          ║
║ 38   ║ Chris Singleton   ║ 868       ║ 637          ║
║ 39   ║ Shelvin Mack      ║ 895       ║ 597          ║
║ 40   ║ Andrew Goudelock  ║ 141       ║ 487          ║
║ 41   ║ Lavoy Allen       ║ 1121      ║ 481          ║
║ 42   ║ Malcolm Lee       ║ 177       ║ 377          ║
║ 43   ║ Keith Benson      ║ 3         ║ 375          ║
║ 44   ║ DeAndre Liggins   ║ 135       ║ 352          ║
║ 45   ║ Vernon Macklin    ║ 45        ║ -165         ║
╚══════╩═══════════════════╩═══════════╩══════════════╝

OOS R^2: 0.25
Non-zero coefficients in the sparse model: 10
2 most important coefficients: Age, STL_tot
'Total Steals' makes it into the top 2 coefficients for the first time.
Several misses all over the place: Darius Morris, Chandler Parsons, Jordan Hamilton and others. Has Kyrie Irving at #21, at around 1000 MP/season too low. Was extremely high on Kawhi Leonard, though (3rd), who got picked 15th. High projection for Kemba Walker, who has even surpassed it

2012

Code: Select all

╔══════╦═════════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name                    ║ Actual MP ║ Projected MP ║
╠══════╬═════════════════════════╬═══════════╬══════════════╣
║ 1    ║ Anthony Davis           ║ 2102      ║ 2135         ║
║ 2    ║ Jared Sullinger         ║ 1466      ║ 1591         ║
║ 3    ║ Michael Kidd-Gilchrist  ║ 1763      ║ 1552         ║
║ 4    ║ Draymond Green          ║ 1429      ║ 1512         ║
║ 5    ║ Jae Crowder             ║ 1303      ║ 1506         ║
║ 6    ║ Bradley Beal            ║ 2137      ║ 1501         ║
║ 7    ║ Thomas Robinson         ║ 964       ║ 1442         ║
║ 8    ║ Dion Waiters            ║ 1914      ║ 1350         ║
║ 9    ║ Terrence Jones          ║ 1177      ║ 1339         ║
║ 10   ║ Kendall Marshall        ║ 1133      ║ 1319         ║
║ 11   ║ Will Barton             ║ 640       ║ 1263         ║
║ 12   ║ Tyshawn Taylor          ║ 245       ║ 1227         ║
║ 13   ║ Jared Cunningham        ║ 53        ║ 1200         ║
║ 14   ║ Tyler Zeller            ║ 1541      ║ 1183         ║
║ 15   ║ Jeremy Lamb             ║ 842       ║ 1175         ║
║ 16   ║ Tony Wroten             ║ 1018      ║ 1158         ║
║ 17   ║ Harrison Barnes         ║ 2131      ║ 1157         ║
║ 18   ║ Marquis Teague          ║ 417       ║ 1017         ║
║ 19   ║ Maurice Harkless        ║ 1962      ║ 994          ║
║ 20   ║ Quincy Miller           ║ 407       ║ 969          ║
║ 21   ║ Damian Lillard          ║ 3052      ║ 949          ║
║ 22   ║ Andre Drummond          ║ 1931      ║ 940          ║
║ 23   ║ Jeff Taylor '12         ║ 1068      ║ 925          ║
║ 24   ║ Perry Jones             ║ 522       ║ 922          ║
║ 25   ║ Doron Lamb              ║ 636       ║ 920          ║
║ 26   ║ Quincy Acy              ║ 594       ║ 912          ║
║ 27   ║ Meyers Leonard          ║ 780       ║ 886          ║
║ 28   ║ Darius Miller           ║ 708       ║ 867          ║
║ 29   ║ Arnett Moultrie         ║ 364       ║ 832          ║
║ 30   ║ Austin Rivers           ║ 1378      ║ 793          ║
║ 31   ║ Kim English             ║ 203       ║ 777          ║
║ 32   ║ Terrence Ross           ║ 1699      ║ 763          ║
║ 33   ║ Kris Joseph             ║ 27        ║ 752          ║
║ 34   ║ John Henson             ║ 1341      ║ 719          ║
║ 35   ║ Andrew Nicholson        ║ 1211      ║ 715          ║
║ 36   ║ Darius Johnson-Odom     ║ 10        ║ 640          ║
║ 37   ║ Mike Scott              ║ 929       ║ 632          ║
║ 38   ║ Fab Melo                ║ 18        ║ 552          ║
║ 39   ║ Orlando Johnson         ║ 505       ║ 456          ║
║ 40   ║ Kyle O'Quinn            ║ 913       ║ 408          ║
║ 41   ║ Khris Middleton         ║ 1467      ║ 383          ║
║ 42   ║ Robert Sacre            ║ 646       ║ 375          ║
║ 43   ║ Glen Rice '13           ║ 54        ║ 371          ║
║ 44   ║ Miles Plumlee           ║ 1009      ║ 355          ║
║ 45   ║ Kevin Murphy            ║ 26        ║ 312          ║
║ 46   ║ Festus Ezeli            ║ 560       ║ 218          ║
║ 47   ║ Bernard James           ║ 301       ║ -80          ║
╚══════╩═════════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.22
Non-zero coefficients in the sparse model: 9
2 most important coefficients: Age, STL_tot
Anthony Davis with the 3rd highest projection ever. Misses big with Jared Cunningham, Terrence Ross, Harrison Barnes. It's a lot higher on Sullinger than his actual pick number (3 vs 21) but I remember that his back was a red flag. 4th highest and spot-on projection for Draymond Green, who got picked 35th. Exact same situation with Crowder, who had the 5th highest projection (which was only a tad too high), and got picked 34th

2013

Code: Select all

╔══════╦═══════════════════════════╦═══════════╦══════════════╗
║ Rank ║ Name                      ║ Actual MP ║ Projected MP ║
╠══════╬═══════════════════════════╬═══════════╬══════════════╣
║ 1    ║ Victor Oladipo            ║ 2487      ║ 1676         ║
║ 2    ║ Cody Zeller               ║ 1416      ║ 1549         ║
║ 3    ║ Trey Burke                ║ 2262      ║ 1521         ║
║ 4    ║ Otto Porter               ║ 319       ║ 1492         ║
║ 5    ║ Michael Carter-Williams   ║ 2414      ║ 1448         ║
║ 6    ║ Ben McLemore              ║ 2187      ║ 1345         ║
║ 7    ║ Shane Larkin              ║ 489       ║ 1278         ║
║ 8    ║ Kentavious Caldwell-Pope  ║ 1583      ║ 1258         ║
║ 9    ║ Nerlens Noel              ║ 0         ║ 1232         ║
║ 10   ║ Kelly Olynyk              ║ 1400      ║ 1053         ║
║ 11   ║ Mason Plumlee             ║ 1275      ║ 1023         ║
║ 12   ║ Anthony Bennett           ║ 663       ║ 999          ║
║ 13   ║ Steven Adams              ║ 1197      ║ 993          ║
║ 14   ║ Nate Wolters              ║ 1309      ║ 924          ║
║ 15   ║ Reggie Bullock            ║ 395       ║ 903          ║
║ 16   ║ Andre Roberson            ║ 399       ║ 871          ║
║ 17   ║ Alex Len                  ║ 362       ║ 845          ║
║ 18   ║ Mike Muscala              ║ 215       ║ 845          ║
║ 19   ║ Archie Goodwin            ║ 533       ║ 835          ║
║ 20   ║ Erik Murphy               ║ 62        ║ 822          ║
║ 21   ║ Allen Crabbe              ║ 100       ║ 807          ║
║ 22   ║ Jamaal Franklin           ║ 161       ║ 788          ║
║ 23   ║ Peyton Siva               ║ 224       ║ 785          ║
║ 24   ║ Solomon Hill              ║ 226       ║ 778          ║
║ 25   ║ Jeff Withey               ║ 684       ║ 735          ║
║ 26   ║ Gorgui Dieng              ║ 818       ║ 729          ║
║ 27   ║ Shabazz Muhammad          ║ 290       ║ 640          ║
║ 28   ║ Tony Snell                ║ 1230      ║ 639          ║
║ 29   ║ Lorenzo Brown             ║ 224       ║ 609          ║
║ 30   ║ Ray McCallum              ║ 897       ║ 534          ║
║ 31   ║ Isaiah Canaan             ║ 252       ║ 461          ║
║ 32   ║ Carrick Felix             ║ 38        ║ 435          ║
║ 33   ║ C.J. McCollum             ║ 476       ║ 160          ║
╚══════╩═══════════════════════════╩═══════════╩══════════════╝

OOS R^2: 0.40
Non-zero coefficients in the sparse model: 10
2 most important coefficients: Age, STL_tot
Good year for the model with a OOS R^2. Too high on Porter, Noel, Shane Larkin. Had Anthony Bennett at #12 and he even played less than projected. Had Nate Wolters 14th, got picked 38th and seems to be decent. Lots of good projections all around

2014

Code: Select all

╔══════╦══════════════════════════╦══════════════╗
║ Rank ║ Name                     ║ Projected MP ║
╠══════╬══════════════════════════╬══════════════╣
║ 1    ║ Kyle Anderson            ║ 2053         ║
║ 2    ║ Elfrid Payton            ║ 2050         ║
║ 3    ║ Jordan Adams             ║ 1891         ║
║ 4    ║ Julius Randle            ║ 1806         ║
║ 5    ║ Jabari Parker            ║ 1784         ║
║ 6    ║ T.J. Warren              ║ 1706         ║
║ 7    ║ Andrew Wiggins           ║ 1672         ║
║ 8    ║ Shabazz Napier           ║ 1637         ║
║ 9    ║ Jarnell Stokes           ║ 1632         ║
║ 10   ║ Delon Wright             ║ 1605         ║
║ 11   ║ Marcus Smart             ║ 1602         ║
║ 12   ║ Aaron Gordon             ║ 1581         ║
║ 13   ║ Gary Harris              ║ 1572         ║
║ 14   ║ Nik Stauskas             ║ 1531         ║
║ 15   ║ James Young              ║ 1489         ║
║ 16   ║ Noah Vonleh              ║ 1456         ║
║ 17   ║ Tyler Ennis              ║ 1414         ║
║ 18   ║ Bobby Portis             ║ 1399         ║
║ 19   ║ Aaron Harrison           ║ 1370         ║
║ 20   ║ Sam Dekker               ║ 1368         ║
║ 21   ║ Jordan Mickey            ║ 1367         ║
║ 22   ║ K.J. McDaniels           ║ 1365         ║
║ 23   ║ Briante Weber            ║ 1350         ║
║ 24   ║ Marcus Paige             ║ 1304         ║
║ 25   ║ Lamar Patterson          ║ 1268         ║
║ 26   ║ Joel Embiid              ║ 1251         ║
║ 27   ║ Doug McDermott           ║ 1248         ║
║ 28   ║ Kevin Pangos             ║ 1243         ║
║ 29   ║ Nick Johnson             ║ 1214         ║
║ 30   ║ Rondae Hollis-Jefferson  ║ 1176         ║
║ 31   ║ Joseph Young             ║ 1140         ║
║ 32   ║ Jarell Martin            ║ 1125         ║
║ 33   ║ Perry Ellis              ║ 1115         ║
║ 34   ║ Glenn Robinson '14       ║ 1113         ║
║ 35   ║ Yogi Ferrell             ║ 1099         ║
║ 36   ║ Johnny O'Bryant          ║ 1084         ║
║ 37   ║ Andrew Harrison          ║ 1054         ║
║ 38   ║ Branden Dawson           ║ 1047         ║
║ 39   ║ Montrezl Harrell         ║ 1044         ║
║ 40   ║ Shawn Long               ║ 1043         ║
║ 41   ║ Jordan Clarkson          ║ 1034         ║
║ 42   ║ Wayne Selden             ║ 1033         ║
║ 43   ║ Brandon Ashley           ║ 1029         ║
║ 44   ║ Willie Cauley            ║ 1017         ║
║ 45   ║ Kaleb Tarczewski         ║ 1015         ║
║ 46   ║ Przemek Karnowski        ║ 1015         ║
║ 47   ║ Khem Birch               ║ 1003         ║
║ 48   ║ Treveon Graham           ║ 994          ║
║ 49   ║ Michael Qualls           ║ 951          ║
║ 50   ║ Rodney Hood              ║ 937          ║
║ 51   ║ Zach LaVine              ║ 926          ║
║ 52   ║ Eric Mika                ║ 908          ║
║ 53   ║ Juvonte Reddic           ║ 895          ║
║ 54   ║ Semaj Christon           ║ 894          ║
║ 55   ║ Spencer Dinwiddie        ║ 893          ║
║ 56   ║ Ron Baker                ║ 889          ║
║ 57   ║ Shaq Goodwin             ║ 885          ║
║ 58   ║ Jerami Grant             ║ 845          ║
║ 59   ║ Mike Tobey               ║ 839          ║
║ 60   ║ Alec Brown               ║ 838          ║
║ 61   ║ Sindarius Thornwell      ║ 835          ║
║ 62   ║ Brice Johnson            ║ 815          ║
║ 63   ║ Mamadou Ndiaye           ║ 795          ║
║ 64   ║ Isaiah Austin            ║ 792          ║
║ 65   ║ Troy Williams            ║ 789          ║
║ 66   ║ Andre Hollins            ║ 788          ║
║ 67   ║ Deonte Burton            ║ 788          ║
║ 68   ║ Rasheed Sulaimon         ║ 785          ║
║ 69   ║ Devyn Marble             ║ 741          ║
║ 70   ║ Jahii Carson             ║ 730          ║
║ 71   ║ Dwight Powell            ║ 714          ║
║ 72   ║ A.J. Hammons             ║ 639          ║
║ 73   ║ Markel Brown             ║ 636          ║
║ 74   ║ Olivier Hanlan           ║ 629          ║
║ 75   ║ Fuquan Edwin             ║ 624          ║
║ 76   ║ C.J. Fair                ║ 618          ║
║ 77   ║ Terran Petteway          ║ 611          ║
║ 78   ║ Alex Poythress           ║ 606          ║
║ 79   ║ Russ Smith               ║ 603          ║
║ 80   ║ Patric Young             ║ 595          ║
║ 81   ║ Kasey Hill               ║ 590          ║
║ 82   ║ James McAdoo             ║ 582          ║
║ 83   ║ Tyler Haws               ║ 541          ║
║ 84   ║ Joe Harris               ║ 540          ║
║ 85   ║ Jerian Grant             ║ 530          ║
║ 86   ║ Keith Appling            ║ 503          ║
║ 87   ║ C.J. Wilcox              ║ 479          ║
║ 88   ║ Jordan McRae             ║ 479          ║
║ 89   ║ Mitch McGary             ║ 468          ║
║ 90   ║ Jabari Bird              ║ 456          ║
║ 91   ║ Aaron Craft              ║ 419          ║
║ 92   ║ Chris Obekpa             ║ 417          ║
║ 93   ║ Joe Jackson              ║ 415          ║
║ 94   ║ Adreian Payne            ║ 402          ║
║ 95   ║ Austin Nichols           ║ 393          ║
║ 96   ║ Cory Jefferson           ║ 382          ║
║ 97   ║ Anthony Barber           ║ 369          ║
║ 98   ║ Shannon Scott            ║ 345          ║
║ 99   ║ LaQuinton Ross           ║ 323          ║
║ 100  ║ LeBryan Nash             ║ 309          ║
║ 101  ║ Jakarr Sampson           ║ 283          ║
║ 102  ║ Roscoe Smith             ║ 273          ║
║ 103  ║ Cleanthony Early         ║ 224          ║
║ 104  ║ Alex Kirk                ║ 182          ║
║ 105  ║ Wayne Blackshear         ║ 43           ║
╚══════╩══════════════════════════╩══════════════╝

Final coefficients in the sparse model

Code: Select all

╔══════════════╦═════════════╗
║ Coeff_name   ║ Coeff_Value ║
╠══════════════╬═════════════╣
║ STL_tot      ║ 146         ║
║ MOV*SOS      ║ 99          ║
║ DRB_tot*SOS  ║ 90          ║
║ TS*SOS       ║ 72          ║
║ PTS_tot*SOS  ║ 70          ║
║ AST          ║ 53          ║
║ TS           ║ 47          ║
║ ORB_tot*SOS  ║ 18          ║
║ TOV_tot      ║ 8           ║
║ PF_by_AGE    ║ -42         ║
║ Age          ║ -248        ║
╚══════════════╩═════════════╝

For this draft it projects Kyle Anderson and Elfrid Payton to play lots of minutes (4th and 5th highest projection ever). Embiid is ranked 26th. Doug McDermott, mocked at #8, is ranked 27th. Zach Lavine, mocked at #13, is ranked 51st

[colts18]

Does this model take into account draft position and the record of the team they got drafted to? That would improve the projections.

[J.E.]

colts... the whole point of this is to predict future NBA minutes before the draft

[talkingpractice]

qualitatively, i think the 2012 and 2013 results are awesome, especially insofar as finding sleepers (Sullinger, Dray, Crowder, Wolters, etc). importantly though, in terms of sleepers, there seem to be virtually no false positives from those years.

[J.E.]

qualitatively, i think the 2012 and 2013 results are awesome, especially insofar as finding sleepers (Sullinger, Dray, Crowder, Wolters, etc). importantly though, in terms of sleepers, there seem to be virtually no false positives from those years.

Thanks. I'm definitely pleased with the 2013 results

I think it's also interesting to note that for 2014, 11 players have a higher minute projection than last year's #2 and 6 players have a higher projection than last year's #1

[J.E.]

Here are some similarity scores when it comes to expected minutes per season.
I've used the coefficients from the regression model to weigh each data point. So, a difference of 'X' in the STL_tot column will have a larger influence on the similarity score than, say, a difference of 'X' in the ORB column (because STL_tot had a larger absolute weight in the regression). Differences are taken after transforming everything to z-scores

Andrew Wiggins

Code: Select all

╔══════════════════════╦═══════════╗
║         NAME         ║ SIM_SCORE ║
╠══════════════════════╬═══════════╣
║ Quincy Miller        ║       448 ║
║ Harrison Barnes      ║       523 ║
║ Bradley Beal         ║       670 ║
║ DeMar DeRozan        ║       683 ║
║ Austin Rivers        ║       684 ║
║ Rudy Gay             ║       776 ║
║ Luol Deng            ║       784 ║
║ Tobias Harris        ║       807 ║
║ DerMarr Johnson      ║       853 ║
║ Marcus Williams '07  ║       857 ║
╚══════════════════════╩═══════════╝

Jabari Parker

Code: Select all

╔══════════════════════╦═══════════╗
║         NAME         ║ SIM_SCORE ║
╠══════════════════════╬═══════════╣
║ Tobias Harris        ║       304 ║
║ DeMar DeRozan        ║       425 ║
║ Quincy Miller        ║       455 ║
║ Harrison Barnes      ║       479 ║
║ Anthony Bennett      ║       496 ║
║ Luol Deng            ║       627 ║
║ Marcus Williams '07  ║       650 ║
║ Shabazz Muhammad     ║       679 ║
║ Chris Bosh           ║       705 ║
║ Chris Taft           ║       770 ║
╚══════════════════════╩═══════════╝

Joel Embiid

Code: Select all

╔═══════════════════╦═══════════╗
║       NAME        ║ SIM_SCORE ║
╠═══════════════════╬═══════════╣
║ Sean Williams     ║       579 ║
║ Byron Mullens     ║       746 ║
║ DeAndre Jordan    ║       831 ║
║ Ryan Hollins      ║       950 ║
║ Daniel Orton      ║       981 ║
║ Hilton Armstrong  ║      1058 ║
║ Patrick O'Bryant  ║      1061 ║
║ DerMarr Johnson   ║      1116 ║
║ Ed Davis          ║      1157 ║
║ Josh Childress    ║      1168 ║
╚═══════════════════╩═══════════╝

Noah Vonleh

Code: Select all

╔═══════════════════╦═══════════╗
║       NAME        ║ SIM_SCORE ║
╠═══════════════════╬═══════════╣
║ Quincy Miller     ║       804 ║
║ DeAndre Jordan    ║       887 ║
║ Steven Adams      ║      1015 ║
║ Nerlens Noel      ║      1083 ║
║ Thaddeus Young    ║      1211 ║
║ Daniel Orton      ║      1219 ║
║ DerMarr Johnson   ║      1228 ║
║ Patrick O'Bryant  ║      1271 ║
║ Steven Hunter     ║      1299 ║
║ Spencer Hawes     ║      1314 ║
╚═══════════════════╩═══════════╝

Marcus Smart

Code: Select all

╔═══════════════════════════╦═══════════╗
║           NAME            ║ SIM_SCORE ║
╠═══════════════════════════╬═══════════╣
║ Iman Shumpert             ║       532 ║
║ Jared Cunningham          ║       644 ║
║ Ronnie Brewer             ║       936 ║
║ Keyon Dooling             ║       994 ║
║ Paul George               ║       994 ║
║ Chris Paul                ║      1123 ║
║ Ray McCallum              ║      1150 ║
║ Kentavious Caldwell-Pope  ║      1151 ║
║ Kyle Lowry                ║      1157 ║
║ Shannon Brown             ║      1183 ║
╚═══════════════════════════╩═══════════╝

Julius Randle

Code: Select all

╔═════════════════════════╦═══════════╗
║          NAME           ║ SIM_SCORE ║
╠═════════════════════════╬═══════════╣
║ Jordan Williams         ║       769 ║
║ Michael Kidd-Gilchrist  ║       826 ║
║ Devin Ebanks            ║       870 ║
║ Chris Bosh              ║       939 ║
║ Alex Len                ║       993 ║
║ Brandon Bass            ║      1021 ║
║ Nikola Vucevic          ║      1062 ║
║ DeMar DeRozan           ║      1077 ║
║ Tristan Thompson        ║      1088 ║
║ Tobias Harris           ║      1089 ║
╚═════════════════════════╩═══════════╝

Doug McDermott

Code: Select all

╔═════════════════╦═══════════╗
║      NAME       ║ SIM_SCORE ║
╠═════════════════╬═══════════╣
║ Dan Langhi      ║       585 ║
║ Justin Harper   ║      1109 ║
║ Nick Young      ║      1311 ║
║ Jon Leuer       ║      1361 ║
║ Maurice Ager    ║      1516 ║
║ Adam Morrison   ║      1571 ║
║ Luke Babbitt    ║      1602 ║
║ Casey Jacobsen  ║      1671 ║
║ Ryan Anderson   ║      1742 ║
║ Desmond Mason   ║      1751 ║
╚═════════════════╩═══════════╝

Nik Stauskas

Code: Select all

╔═════════════════╦═══════════╗
║      NAME       ║ SIM_SCORE ║
╠═════════════════╬═══════════╣
║ Tony Snell      ║      1120 ║
║ Arron Afflalo   ║      1130 ║
║ Doron Lamb      ║      1139 ║
║ Reggie Jackson  ║      1200 ║
║ Jeremy Lamb     ║      1217 ║
║ Jerryd Bayless  ║      1276 ║
║ Ben McLemore    ║      1282 ║
║ Malcolm Lee     ║      1340 ║
║ Roger Mason     ║      1410 ║
║ Guillermo Diaz  ║      1438 ║
╚═════════════════╩═══════════╝

Aaron Gordon

Code: Select all

╔═════════════════════════╦═══════════╗
║          NAME           ║ SIM_SCORE ║
╠═════════════════════════╬═══════════╣
║ Quincy Miller           ║       986 ║
║ DeMar DeRozan           ║      1088 ║
║ Chris Wilcox            ║      1148 ║
║ Luol Deng               ║      1154 ║
║ Steven Adams            ║      1194 ║
║ Michael Kidd-Gilchrist  ║      1197 ║
║ Thaddeus Young          ║      1226 ║
║ Rudy Gay                ║      1273 ║
║ Devin Ebanks            ║      1294 ║
║ Josh McRoberts          ║      1297 ║
╚═════════════════════════╩═══════════╝

Elfrid Payton

Code: Select all

╔════════════════════╦═══════════╗
║        NAME        ║ SIM_SCORE ║
╠════════════════════╬═══════════╣
║ Julius Hodge       ║       656 ║
║ John Wall          ║       876 ║
║ Derrick Rose       ║       932 ║
║ Tony Wroten        ║       991 ║
║ Joseph Forte       ║      1069 ║
║ Kenny Satterfield  ║      1076 ║
║ Tyreke Evans       ║      1091 ║
║ Russell Westbrook  ║      1094 ║
║ Frank Williams     ║      1168 ║
║ Kyle Weaver        ║      1193 ║
╚════════════════════╩═══════════╝

Zach Lavine

Code: Select all

╔══════════════════╦═══════════╗
║       NAME       ║ SIM_SCORE ║
╠══════════════════╬═══════════╣
║ Cory Joseph      ║      1091 ║
║ Avery Bradley    ║      1141 ║
║ Von Wafer        ║      1150 ║
║ Gabe Pruitt      ║      1701 ║
║ Malcolm Lee      ║      1743 ║
║ Josh Selby       ║      1750 ║
║ Thaddeus Young   ║      1831 ║
║ Khris Middleton  ║      1888 ║
║ Jamal Crawford   ║      1937 ║
║ Tony Snell       ║      1999 ║
╚══════════════════╩═══════════╝

Kyle Anderson

Code: Select all

╔═════════════════╦═══════════╗
║      NAME       ║ SIM_SCORE ║
╠═════════════════╬═══════════╣
║ Andre Iguodala  ║       826 ║
║ Jamison Brewer  ║      1029 ║
║ Kyle Weaver     ║      1038 ║
║ Chase Budinger  ║      1092 ║
║ Rajon Rondo     ║      1101 ║
║ Josh McRoberts  ║      1104 ║
║ John Salmons    ║      1110 ║
║ Otto Porter     ║      1170 ║
║ Darius Morris   ║      1218 ║
║ Ramon Sessions  ║      1253 ║
╚═════════════════╩═══════════╝

[mystic]

Final coefficients in the sparse model

Code: Select all

╔══════════════╦═════════════╗
║ Coeff_name   ║ Coeff_Value ║
╠══════════════╬═════════════╣
║ STL_tot      ║ 146         ║
║ MOV*SOS      ║ 99          ║
║ DRB_tot*SOS  ║ 90          ║
║ TS*SOS       ║ 72          ║
║ PTS_tot*SOS  ║ 70          ║
║ AST          ║ 53          ║
║ TS           ║ 47          ║
║ ORB_tot*SOS  ║ 18          ║
║ TOV_tot      ║ 8           ║
║ PF_by_AGE    ║ -42         ║
║ Age          ║ -248        ║
╚══════════════╩═════════════╝

For this draft it projects Kyle Anderson and Elfrid Payton to play lots of minutes (4th and 5th highest projection ever). Embiid is ranked 26th. Doug McDermott, mocked at #8, is ranked 27th. Zach Lavine, mocked at #13, is ranked 51st

Couple of questions regarding this:

1. MOV: Is that the scoring margin by the team overall or just during the minutes the respective player was on the court?
2. SOS: May I assume correctly that this is a factor ranging from 0 to 1? How did you calculate that or where did you get that info from?
3. PF_by_AGE: What does that mean? An adjusted PF value based on age I assume, but how was that adjusted?

Not related to those coefficients:
I assume Colts actually didn't want to ask that you adjust the values for the not-yet-drafted players to a specific scenario, but rather try to find a normalised minutes value for the players we have the information already how many minutes they played. Meaning, if player A gets drafted by a team having already filled the position with better players, he simply would get less minutes not based on his own quality/numbers. What might improve the prediction is a minute value adjusted for that; in essence predicting the minutes of a player, if he would get drafted into a fictional average situation just based on the information available before the draft. Hope, that makes some sense ... so, I understand that the model is basically doing that by just assuming that in the future players will get drafted into similar good/bad situations as in the past. Question would be: Does an adjustment would improve the results? Probably not by much, so maybe that wouldn't be worth the effort. But nonetheless it could done somewhat by using prior draft results where you predict minutes on draft position, then compare that model result to the result of your current minute prediction model and maybe find some significant coefficient there to adjust ... in that way you would only use information available before the draft.

[J.E.]

1. MOV: Is that the scoring margin by the team overall or just during the minutes the respective player was on the court?

I don't think there's reliable PBP for College, definitely not for 2000-2010, so it's the former

2. SOS: May I assume correctly that this is a factor ranging from 0 to 1? How did you calculate that or where did you get that info from?

It's VJL's dataset, so I'm not entirely sure. I suppose he got it from KenPom or Sagarin. It was originally in points/game but I shifted it by league average team strength (+72)

3. PF_by_AGE: What does that mean? An adjusted PF value based on age I assume, but how was that adjusted?

Sorry, I should have been more clear. It's PF divided by Age

Not related to those coefficients:
I assume Colts actually didn't want to ask that you adjust the values for the not-yet-drafted players to a specific scenario, but rather try to find a normalised minutes value for the players we have the information already how many minutes they played. Meaning, if player A gets drafted by a team having already filled the position with better players, he simply would get less minutes not based on his own quality/numbers. What might improve the prediction is a minute value adjusted for that; in essence predicting the minutes of a player, if he would get drafted into a fictional average situation just based on the information available before the draft. Hope, that makes some sense ... so, I understand that the model is basically doing that by just assuming that in the future players will get drafted into similar good/bad situations as in the past. Question would be: Does an adjustment would improve the results? Probably not by much, so maybe that wouldn't be worth the effort. But nonetheless it could done somewhat by using prior draft results where you predict minutes on draft position, then compare that model result to the result of your current minute prediction model and maybe find some significant coefficient there to adjust ... in that way you would only use information available before the draft.

That sounds interesting and might help a little bit, but it's probably non-trivial to come up with an algorithm that can figure out a team's depth at certain positions - for one, you'd need injury data, which I don't have

What obviously should be done, by people who have more time/interest than I have, is try to predict player impact on top of player minutes. Optimally, your metric that measures impact is free of any situational factors

[Statman]

What obviously should be done, by people who have more time/interest than I have, is try to predict player impact on top of player minutes. Optimally, your metric that measures impact is free of any situational factors

This is exactly what I'm trying to do with my model - which I posted already in your sister thread, but I'll post the link here too for fun.

http://hoopsnerd.com/?p=600

[J.E.]

I went and checked whether actual pick number or my model give better "prediction" results when it comes to minutes played in the NBA. My model has a little bit of a disadvantage here as teams sometimes give a player minutes, despite him being awful, because they drafted him early, probably hoping he'll turn it around. Anyway

Scoring was done the following way: Rank all players of one draft by actual minutes played (1), sum up the squared differences between (1) and a) Actual Pick Rank, and b) Predicted Minutes Rank.
As an example, Lillard has played the most minutes of his draft class. My model had him ranked 21st, for a total error of 20^2. He got picked 6th, giving the "Actual Pick Rank" an error of 5^2

As you can see from the table below, the model did better in some of the years (2003, 2006, 2009, 2013), but "actual pick number" did better overall when predicting rank in minutes played (the model did pretty awful in 2007 and 2012)

Code: Select all

╔══════╦═════════════╦═══════════════════╦═════════════╗
║ Year ║ Model Error ║ Actual Pick Error ║ Comb. Error ║
╠══════╬═════════════╬═══════════════════╬═════════════╣
║ 2003 ║        2246 ║              3006 ║        2506 ║
║ 2004 ║        1192 ║              1101 ║         894 ║
║ 2005 ║        4375 ║              3150 ║        3213 ║
║ 2006 ║        6389 ║              7278 ║        5925 ║
║ 2007 ║        7968 ║              5008 ║        4780 ║
║ 2008 ║        5748 ║              4202 ║        3742 ║
║ 2009 ║        5368 ║              5954 ║        4866 ║
║ 2010 ║        6140 ║              4688 ║        4066 ║
║ 2011 ║        7595 ║              6440 ║        6177 ║
║ 2012 ║        9090 ║              5282 ║        4878 ║
║ 2013 ║        3011 ║              4132 ║        2869 ║
║ SUM  ║       59122 ║             50241 ║       43916 ║
╚══════╩═════════════╩═══════════════════╩═════════════╝

What's interesting is that if you combine the model and "actual pick number" predictions and form a new "prediction" ("Comb." in the table above), this prediction method would have outperformed both other methods every year (except 2003 and 2005) and also has the lowest error sum.
This new method would have performed best if you had weighed Actual Pick Number with 65%, and the model's prediction with 35%

As such, here are the predictions from the combined model

Code: Select all

╔════════════════╦══════════════════════════════════╗
║      Name      ║ Predicted Rank in Minutes Played ║
╠════════════════╬══════════════════════════════════╣
║ Jabari Parker  ║                                1 ║
║ Andrew Wiggins ║                                2 ║
║ Julius Randle  ║                                3 ║
║ Aaron Gordon   ║                                4 ║
║ Marcus Smart   ║                                5 ║
║ Elfrid Payton  ║                                6 ║
║ Joel Embiid    ║                                7 ║
║ Nik Stauskas   ║                                8 ║
║ T. J. Warren   ║                                9 ║
║ Noah Vonleh    ║                               10 ║
║ Doug McDermott ║                               11 ║
║ Jordan Adams   ║                               12 ║
║ James Young    ║                               13 ║
║ Zach LaVine    ║                               14 ║
║ Gary Harris    ║                               15 ║
║ Kyle Anderson  ║                               16 ║
║ Tyler Ennis    ║                               17 ║
║ Shabazz Napier ║                               18 ║
║ Adreian Payne  ║                               19 ║
║ Mitch McGary   ║                               20 ║
║ Rodney Hood    ║                               21 ║
║ C. J. Wilcox   ║                               22 ║
╚════════════════╩══════════════════════════════════╝

Now that the draft is over, this list represents my "best guess" when it comes to predicting these players' rank in "minutes played in their NBA career"

[Statman]

Double Post - where's the delete option?

[Statman]

I went and checked whether actual pick number or my model give better "prediction" results when it comes to minutes played in the NBA. My model has a little bit of a disadvantage here as teams sometimes give a player minutes, despite him being awful, because they drafted him early, probably hoping he'll turn it around.

I really think this is a pretty big detriment to trying to figure out the validity of your metric, especially the first few NBA seasons. Obviously Kyle Anderson would have played a ton more minutes if he landed with the Sixers than he will with the Spurs.

I think your metric got hosed by Kyle Anderson going so late - a player with his special combination of offensive skillsets (while being 6'9 w/ a 9' standing reach) deserved to be lottery - I don't care how methodical ("Slo-Mo") he looks when he takes the ball up court.

[J.E.]

I went and checked whether actual pick number or my model give better "prediction" results when it comes to minutes played in the NBA. My model has a little bit of a disadvantage here as teams sometimes give a player minutes, despite him being awful, because they drafted him early, probably hoping he'll turn it around.

I really think this is a pretty big detriment to trying to figure out the validity of your metric, especially the first few NBA seasons. Obviously Kyle Anderson would have played a ton more minutes if he landed with the Sixers than he will with the Spurs.

I think your metric got hosed by Kyle Anderson going so late - a player with his special combination of offensive skillsets (while being 6'9 w/ a 9' standing reach) deserved to be lottery - I don't care how methodical ("Slo-Mo") he looks when he takes the ball up court.

I'm not trying to prove anything here, I'm just trying to make good minute projections. Remember that I'm forecasting career "MP per season", not just the first season. If Anderson really is that good, it will show eventually. Obviously I have my doubts that he'll play the most minutes of all players picked, but 16th seems reasonable

[colts18]

Good call on Zach Lavine. #51 is appropriate for him considering how bad he is.

Did you run a projection for the 2015 draft?

[J.E.]

Did you run a projection for the 2015 draft?

Unfortunately I didn't have the chance to do so.

As for past predictions I am most proud about having Draymond Green, who just signed a $85 million deal, 4th in the 2012 draft. He got picked 35th