Page 2 of 2
Re: Front Porch (for casual chat)
Posted: Thu Dec 11, 2014 4:49 pm
by sndesai1
i'm a big warriors fan...started posting at realgm around 2002, but i rarely post there now, just read some of the discussions and game threads. i wish evanz would go back to posting there but i guess there's too many dummies to deal with
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 12:34 pm
by Hoopalytics
Crow wrote:Who here is a big college hoop head? What do you think of Kentucky, especially the lesser praised offense? Main challenger?
Major college hoop head here. I currently work/have worked with some D1 and D2 coaching staffs and helped to develop HoopLens.com with Jeff Haley, the founder of Hoop-math.com.
As for UK's offense, as long as they are creating that many transition 1st looks off Blocks (25.9%) and Steals (12.5%), and getting that many 2nd chances (46.3% OR), they won't need to generate good 1st looks out of their halfcourt sets. Holding teams to 38% at the rim gives you plenty of leeway.
Their biggest challenge could be complacency. Texas had enough big bodies to throw at them for a while. A pack line team like Wisconsin or Virginia will force them to play out of their comfort zone. If they catch them sleepwalking early and shorten the game to ~50 possessions....maybe.
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 1:08 pm
by DSMok1
Can anyone explain why Anthony Davis's PER and WS/48 are greater than Russell Westbrook's? Looking at their stats, I would consider Westbrook's numbers better, and indeed Box Plus/Minus has Westbrook well above Davis. Is it due to the difference in blocks? BPM doesn't consider blocks particularly valuable.
Code: Select all
╔═════╦════════════════════╦══════╦══════╦══════╦════╦═════╦══════╦═══════╦═══════╦═══════╦═══════╦═══════╦═══════╦═══════╦═══════╦═══════╦═══════╦═══════╦══════╦══════╦═════╦════════╦═══════╦═══════╦══════╦══════╗
║ Rk ║ Player ║ Pos ║ Age ║ Tm ║ G ║ MP ║ PER ║ TS% ║ 3PAr ║ FTr ║ ORB% ║ DRB% ║ TRB% ║ AST% ║ STL% ║ BLK% ║ TOV% ║ USG% ║ OWS ║ DWS ║ WS ║ WS/48 ║ OBPM ║ DBPM ║ BPM ║ VORP ║
╠═════╬════════════════════╬══════╬══════╬══════╬════╬═════╬══════╬═══════╬═══════╬═══════╬═══════╬═══════╬═══════╬═══════╬═══════╬═══════╬═══════╬═══════╬══════╬══════╬═════╬════════╬═══════╬═══════╬══════╬══════╣
║ 1 ║ Russell Westbrook ║ PG ║ 26 ║ OKC ║ 8 ║ 231 ║ 30.6 ║ 0.579 ║ 0.151 ║ 0.496 ║ 5.5 ║ 15.2 ║ 10.5 ║ 53.9 ║ 2.9 ║ 0 ║ 16.7 ║ 39.8 ║ 0.8 ║ 0.4 ║ 1.2 ║ 0.24 ║ 10.4 ║ 1.1 ║ 11.5 ║ 2.9 ║
║ 2 ║ Stephen Curry ║ PG ║ 26 ║ GSW ║ 21 ║ 691 ║ 26.9 ║ 0.629 ║ 0.449 ║ 0.265 ║ 2.2 ║ 13.7 ║ 8.3 ║ 39 ║ 2.8 ║ 0.5 ║ 15.1 ║ 28.6 ║ 2.7 ║ 1.5 ║ 4.2 ║ 0.293 ║ 8 ║ 1.2 ║ 9.2 ║ 7.7 ║
║ 3 ║ James Harden ║ SG ║ 25 ║ HOU ║ 22 ║ 825 ║ 25.4 ║ 0.581 ║ 0.373 ║ 0.518 ║ 2.9 ║ 16 ║ 9.5 ║ 34.3 ║ 2.8 ║ 2.3 ║ 15.8 ║ 31.7 ║ 2.8 ║ 1.7 ║ 4.4 ║ 0.258 ║ 5.8 ║ 2.2 ║ 8 ║ 7.7 ║
║ 4 ║ Chris Paul ║ PG ║ 29 ║ LAC ║ 21 ║ 730 ║ 26.8 ║ 0.612 ║ 0.282 ║ 0.26 ║ 1.5 ║ 13.6 ║ 7.7 ║ 47.1 ║ 2.9 ║ 0.7 ║ 10.4 ║ 21.9 ║ 3.5 ║ 1 ║ 4.5 ║ 0.296 ║ 7.3 ║ 0.2 ║ 7.4 ║ 6.8 ║
║ 5 ║ Anthony Davis ║ PF ║ 21 ║ NOP ║ 21 ║ 762 ║ 32.9 ║ 0.618 ║ 0.017 ║ 0.419 ║ 8.7 ║ 24.5 ║ 16.4 ║ 8.5 ║ 2.8 ║ 6.1 ║ 6.2 ║ 26.6 ║ 3.4 ║ 1.3 ║ 4.7 ║ 0.299 ║ 4.3 ║ 2.8 ║ 7.1 ║ 6.9 ║
╚═════╩════════════════════╩══════╩══════╩══════╩════╩═════╩══════╩═══════╩═══════╩═══════╩═══════╩═══════╩═══════╩═══════╩═══════╩═══════╩═══════╩═══════╩══════╩══════╩═════╩════════╩═══════╩═══════╩══════╩══════╝
[/size]
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 2:24 pm
by mystic
Your question might be answered by looking at this:
Code: Select all
ePER=3301.2*(3.498*FGM-1.097*FGA+1.412*3PM+1.861*FTM-0.538*FTA+1.062*ORB+0.422*DRB+0.968*AST+1.559*STL+1.062*BLK-1.529*TOV-0.479*PF)/(MP*TM_PACE)*106.8/LgORTG
The results for each boxscore entry and the overall estimated PER (ePER = sum of the other entries):
Code: Select all
Player FGM FGA 3PM FTM FTA ORB DRB AST STL BLK TOV PF ePER
Anthony Davis 33.8 -18.6 0.0 10.3 -3.8 3.1 3.2 1.6 3.0 3.0 -2.0 -1.0 32.5
Russell Westbrook 36.7 -23.9 1.5 16.0 -5.8 1.8 2.2 8.2 3.2 0.0 -8.1 -1.4 30.4
Stephen Curry 28.8 -18.5 4.3 7.7 -2.4 0.7 2.0 7.6 3.0 0.3 -5.1 -1.2 27.2
Chris Paul 24.1 -14.7 2.1 5.7 -1.9 0.5 1.7 9.9 3.1 0.3 -2.7 -1.1 27.0
James Harden 25.7 -19.2 3.1 15.0 -4.9 1.0 2.1 5.9 3.0 1.1 -6.2 -1.3 25.4
Differences to Anthony Davis:
Russell Westbrook 2.9 -5.3 1.5 5.8 -2.0 -1.2 -1.0 6.6 0.2 -3.0 -6.1 -0.4 -2.1
Stephen Curry -5.0 0.2 4.3 -2.6 1.4 -2.4 -1.3 6.1 0.0 -2.7 -3.1 -0.2 -5.3
Chris Paul -9.8 3.9 2.1 -4.6 2.0 -2.6 -1.5 8.3 0.1 -2.7 -0.6 -0.1 -5.5
James Harden -8.1 -0.5 3.1 4.7 -1.0 -2.1 -1.1 4.3 0.0 -1.9 -4.1 -0.2 -7.1
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 3:33 pm
by Crow
I am to the point of almost completely ignoring per and maybe have made my last criticisms of dws too, but mystic that was a good reply. Westbrook's usage is the difference, particularly the assists and the overly generous credit for them on PER.
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 4:10 pm
by Mike G
Regarding Win Shares:
Code: Select all
. player TS% TO%
A Davis .618 6.2
Westbrook .579 16.7
Nothing else seems to matter as much as these 2.
It may be that
everything else doesn't matter as much.
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 6:15 pm
by mystic
Crow wrote:I am to the point of almost completely ignoring per and maybe have made my last criticisms of dws too, but mystic that was a good reply. Westbrook's usage is the difference, particularly the assists and the overly generous credit for them on PER.
Why would you want to ignore PER? Is there a specific reason? Most times people complain about the lack of in-season correlation, but let me give you an example of an "adjusted" PER, where I simply used PER to devide the team performance level among the individual players, and then transformed that value back so that the league average is again 15. That explains wins as good as pythagorean expectation in-season. adjPER explains 94% of the variance in PER, thus the adjustment has a rather small effect on the player ranking. Here are the results for this season (sorted by adjPER while using 100 MP as qualifier):
Code: Select all
Player Tm MP adjPER
Russell Westbrook OKC 231 30.1
Anthony Davis NOP 762 30.0
Stephen Curry GSW 691 29.0
Marreese Speights GSW 330 27.4
Chris Paul LAC 730 27.0
James Harden HOU 825 26.8
DeMarcus Cousins SAC 480 26.3
Kevin Durant OKC 147 26.1
Brandan Wright DAL 455 25.5
LeBron James CLE 760 24.4
Marc Gasol MEM 726 24.4
Kyle Lowry TOR 771 24.1
Tim Duncan SAS 575 24.1
Blake Griffin LAC 721 23.8
Damian Lillard POR 772 23.5
Kyle O'Quinn ORL 160 23.4
Klay Thompson GSW 664 23.0
LaMarcus Aldridge POR 745 22.7
Isaiah Thomas PHO 354 22.4
Mike Conley MEM 677 22.4
Tyson Chandler DAL 713 22.2
Dwight Howard HOU 335 22.2
Jeff Teague ATL 652 22.1
Pau Gasol CHI 646 21.8
Kyrie Irving CLE 795 21.8
Jimmy Butler CHI 753 21.6
Louis Williams TOR 472 21.6
John Wall WAS 750 21.5
Zach Randolph MEM 650 21.5
Jonas Valanciunas TOR 559 21.4
Terrence Jones HOU 117 21.3
Dirk Nowitzki DAL 649 21.3
Jamal Crawford LAC 538 21.1
Chris Kaman POR 418 20.5
Andrew Bogut GSW 503 20.4
Kawhi Leonard SAS 638 20.3
Dennis Schröder ATL 346 20.3
Paul Millsap ATL 714 20.1
Dwyane Wade MIA 485 20.1
Al Horford ATL 610 20.0
Rasual Butler WAS 355 20.0
Rudy Gay SAC 762 20.0
Marcin Gortat WAS 643 19.7
DeAndre Jordan LAC 697 19.7
Chris Bosh MIA 776 19.5
Jose Barea DAL 394 19.2
Derrick Favors UTA 656 19.2
Tony Parker SAS 572 19.1
Monta Ellis DAL 809 19.1
Manu Ginobili SAS 457 19.0
Darren Collison SAC 700 18.9
James Johnson TOR 374 18.9
Wesley Matthews POR 735 18.9
Al Jefferson CHO 693 18.8
Draymond Green GSW 678 18.8
Amir Johnson TOR 502 18.7
Nikola Vucevic ORL 673 18.7
Gerald Green PHO 511 18.7
Kevin Love CLE 754 18.6
Carmelo Anthony NYK 749 18.6
Harrison Barnes GSW 650 18.5
Patrick Beverley HOU 315 18.5
Tyler Zeller BOS 365 18.5
Taj Gibson CHI 398 18.4
Robin Lopez POR 618 18.4
Anderson Varejao CLE 509 18.4
Amar'e Stoudemire NYK 594 18.3
Brandon Knight MIL 738 18.3
Kyle Korver ATL 687 18.2
Eric Bledsoe PHO 758 18.2
DeMar DeRozan TOR 538 18.2
Danny Green SAS 644 18.2
Aron Baynes SAS 374 18.2
Tristan Thompson CLE 548 18.1
Ersan Ilyasova MIL 354 18.1
Cory Joseph SAS 416 18.1
Jrue Holiday NOP 711 17.9
Jared Sullinger BOS 587 17.9
Andre Miller WAS 267 17.9
Derrick Rose CHI 347 17.9
Reggie Jackson OKC 682 17.8
Jeremy Lamb OKC 430 17.8
Paul Pierce WAS 572 17.8
Alexey Shved PHI 243 17.7
Marco Belinelli SAS 259 17.6
Markieff Morris PHO 713 17.6
Ty Lawson DEN 773 17.5
Lavoy Allen IND 456 17.4
A.J. Price IND 193 17.4
Goran Dragic PHO 742 17.4
Kevin Martin MIN 268 17.4
Beno Udrih MEM 357 17.4
Courtney Lee MEM 577 17.3
Serge Ibaka OKC 711 17.3
Tobias Harris ORL 778 17.3
Roy Hibbert IND 498 17.2
Jon Leuer MEM 293 17.1
Bismack Biyombo CHO 193 17.1
Michael Kidd-Gilchrist CHO 159 17.1
Patrick Patterson TOR 573 17.0
Chandler Parsons DAL 806 17.0
Joakim Noah CHI 542 17.0
Rudy Gobert UTA 354 16.9
JaVale McGee DEN 162 16.9
John Henson MIL 171 16.7
Omri Casspi SAC 372 16.6
Deron Williams BRK 726 16.6
Brook Lopez BRK 481 16.6
Tony Allen MEM 482 16.6
Gordon Hayward UTA 777 16.5
Isaiah Canaan HOU 308 16.5
Shabazz Muhammad MIN 374 16.5
Nikola Mirotic CHI 371 16.4
Jerome Jordan BRK 172 16.4
Dewayne Dedmon ORL 186 16.4
Kenneth Faried DEN 508 16.3
Matt Bonner SAS 301 16.3
Greg Monroe DET 609 16.3
Omer Asik NOP 468 16.2
Rodney Stuckey IND 354 16.2
DeMarre Carroll ATL 540 16.1
Mike Scott ATL 269 16.1
Aaron Brooks CHI 400 16.0
Timofey Mozgov DEN 586 16.0
Leandro Barbosa GSW 258 15.8
Luis Scola IND 502 15.8
Kobe Bryant LAL 779 15.7
J.J. Redick LAC 614 15.7
Giannis Antetokounmpo MIL 619 15.7
Shaun Livingston GSW 356 15.7
Devin Harris DAL 551 15.7
Jeff Green BOS 684 15.6
Terrence Ross TOR 612 15.6
Jordan Hill LAL 664 15.6
Brandon Jennings DET 560 15.5
Kris Humphries WAS 425 15.5
Boris Diaw SAS 606 15.5
Kosta Koufos MEM 274 15.4
Enes Kanter UTA 568 15.4
Andre Drummond DET 653 15.4
Alonzo Gee DEN 223 15.3
Luol Deng MIA 619 15.3
Jae Crowder DAL 224 15.3
C.J. Watson IND 165 15.3
Donatas Motiejunas HOU 592 15.3
Greivis Vasquez TOR 469 15.2
Tarik Black HOU 348 15.2
Ryan Anderson NOP 584 15.2
Gorgui Dieng MIN 545 15.2
Larry Sanders MIL 470 15.1
Trevor Ariza HOU 834 15.1
Spencer Hawes LAC 368 15.1
Joe Johnson BRK 629 15.0
Alexis Ajinca NOP 160 15.0
Jabari Parker MIL 688 15.0
Marcus Thornton BOS 286 14.8
Nicolas Batum POR 610 14.7
Nene Hilario WAS 353 14.6
Brandon Bass BOS 387 14.6
Al-Farouq Aminu DAL 380 14.6
Victor Oladipo ORL 469 14.6
Jason Terry HOU 555 14.5
Wilson Chandler DEN 667 14.4
J.J. Hickson DEN 268 14.4
Ed Davis LAL 493 14.4
Aaron Gordon ORL 165 14.4
Alex Len PHO 425 14.4
Tayshaun Prince MEM 283 14.3
Carl Landry SAC 434 14.2
Mirza Teletovic BRK 466 14.1
Rajon Rondo BOS 610 14.1
Kelly Olynyk BOS 506 14.1
Mason Plumlee BRK 283 14.0
Darrell Arthur DEN 357 14.0
Mario Chalmers MIA 667 14.0
C.J. McCollum POR 144 13.9
Festus Ezeli GSW 155 13.9
Kevin Seraphin WAS 302 13.8
Robert Covington PHI 213 13.8
Kemba Walker CHO 738 13.7
Jeff Ayres SAS 130 13.7
Otto Porter WAS 420 13.7
Arron Afflalo DEN 684 13.6
Ian Mahinmi IND 366 13.6
Mike Dunleavy CHI 643 13.6
Kevin Garnett BRK 428 13.6
Tyreke Evans NOP 716 13.6
Jonas Jerebko DET 268 13.5
Marcus Morris PHO 575 13.5
Anthony Morrow OKC 402 13.4
Allen Crabbe POR 268 13.4
Steven Adams OKC 534 13.3
Donald Sloan IND 562 13.3
Hedo Turkoglu LAC 158 13.3
Sebastian Telfair OKC 327 13.2
Andre Iguodala GSW 566 13.2
Bradley Beal WAS 384 13.2
Evan Turner BOS 454 13.0
O.J. Mayo MIL 560 13.0
Reggie Evans SAC 315 12.9
Derrick Williams SAC 235 12.9
Kostas Papanikolaou HOU 485 12.9
Josh Smith DET 725 12.9
Nikola Pekovic MIN 240 12.7
Cole Aldrich NYK 147 12.7
Joel Freeland POR 194 12.7
Tony Wroten PHI 489 12.7
Ricky Rubio MIN 144 12.7
David West IND 186 12.6
P.J. Tucker PHO 507 12.6
Ryan Hollins SAC 131 12.6
Tim Hardaway NYK 429 12.6
Cody Zeller CHO 500 12.4
Zaza Pachulia MIL 422 12.4
Drew Gooden WAS 223 12.4
Jarrett Jack BRK 500 12.4
Steve Blake POR 482 12.4
Gary Neal CHO 359 12.4
Garrett Temple WAS 343 12.3
Dion Waiters CLE 470 12.3
Richard Jefferson DAL 228 12.2
Miles Plumlee PHO 517 12.1
Khris Middleton MIL 458 12.1
Carlos Boozer LAL 550 12.1
Wayne Ellington LAL 258 12.1
Danilo Gallinari DEN 373 12.0
Jordan Clarkson LAL 126 12.0
Randy Foye DEN 209 11.9
Pablo Prigioni NYK 421 11.9
Joey Dorsey HOU 192 11.9
Tyler Hansbrough TOR 225 11.8
Glen Davis LAC 165 11.7
Shawne Williams MIA 545 11.7
Jameer Nelson DAL 507 11.7
Nick Young LAL 302 11.7
Evan Fournier ORL 752 11.7
Matt Barnes LAC 516 11.7
Reggie Bullock LAC 120 11.7
Thabo Sefolosha ATL 365 11.6
Jose Calderon NYK 312 11.6
Jordan Farmar LAC 281 11.6
James Ennis MIA 307 11.6
Kendall Marshall MIL 173 11.5
Jerryd Bayless MIL 432 11.5
Ben Gordon ORL 335 11.5
Solomon Hill IND 712 11.4
Corey Brewer MIN 581 11.4
Marcus Smart BOS 160 11.4
Matthew Dellavedova CLE 130 11.3
Henry Sims PHI 514 11.3
Perry Jones OKC 192 11.3
Shelvin Mack ATL 285 11.3
Trevor Booker UTA 445 11.2
Kirk Hinrich CHI 531 11.2
Andre Roberson OKC 319 11.1
D.J. Augustin DET 501 11.1
K.J. McDaniels PHI 538 11.0
Jared Dudley MIL 436 10.9
Anthony Bennett MIN 347 10.9
Brian Roberts CHO 382 10.8
Michael Carter-Williams PHI 473 10.8
Ben McLemore SAC 754 10.7
Iman Shumpert NYK 608 10.7
Shawn Marion CLE 510 10.7
Kendrick Perkins OKC 449 10.5
Nick Collison OKC 382 10.5
Samuel Dalembert NYK 418 10.4
Jeremy Lin LAL 635 10.4
Luke Ridnour ORL 265 10.4
Pero Antic ATL 320 10.4
Elfrid Payton ORL 598 10.3
Mo Williams MIN 429 10.2
Chuck Hayes TOR 129 10.2
J.R. Smith NYK 528 10.2
Kyle Singler DET 525 10.2
Jason Smith NYK 389 10.1
Alec Burks UTA 666 10.1
Ramon Sessions SAC 376 10.0
Avery Bradley BOS 600 10.0
Vince Carter MEM 287 9.9
Norris Cole MIA 442 9.9
Thaddeus Young MIN 516 9.9
Nate Robinson DEN 249 9.9
Lance Stephenson CHO 714 9.7
Justin Hamilton MIA 134 9.7
Robert Sacre LAL 276 9.6
Trey Burke UTA 712 9.6
Anthony Tolliver PHO 248 9.5
Gerald Henderson CHO 448 9.5
Chris Copeland IND 585 9.5
Austin Rivers NOP 454 9.4
Quincy Acy NYK 464 9.2
Quincy Pondexter MEM 324 9.0
P.J. Hairston CHO 202 9.0
Kentavious Caldwell-Pope DET 736 8.9
Luc Mbah a Moute PHI 542 8.9
Nick Johnson HOU 121 8.8
Josh McRoberts MIA 296 8.8
Brandon Davies PHI 379 8.8
Lance Thomas OKC 445 8.8
Damjan Rudez IND 364 8.8
Tony Snell CHI 171 8.7
Jason Thompson SAC 633 8.7
Channing Frye ORL 741 8.7
Bojan Bogdanovic BRK 593 8.6
Shane Larkin NYK 490 8.5
Nerlens Noel PHI 557 8.5
Marvin Williams CHO 464 8.4
Eric Gordon NOP 372 8.3
Kyle Anderson SAS 142 8.2
Will Barton POR 119 8.2
Shabazz Napier MIA 482 8.0
Caron Butler DET 542 7.9
Maurice Harkless ORL 232 7.9
Phil Pressey BOS 149 7.8
C.J. Miles IND 346 7.5
Francisco Garcia HOU 200 7.5
Ray McCallum SAC 102 7.4
Wesley Johnson LAL 684 7.3
Luke Babbitt NOP 252 7.3
Kent Bazemore ATL 148 7.2
John Salmons NOP 172 7.2
Udonis Haslem MIA 192 6.9
Andrew Wiggins MIN 633 6.8
Travis Wear NYK 182 6.8
Austin Daye SAS 173 6.6
Zach LaVine MIN 438 6.6
Hollis Thompson PHI 585 6.5
Ronnie Price LAL 439 6.4
Robbie Hummel MIN 218 6.3
Doug McDermott CHI 198 6.1
Nik Stauskas SAC 322 6.1
Joe Harris CLE 200 6.0
Gary Harris DEN 140 5.9
Dante Cunningham NOP 116 5.8
Alan Anderson BRK 386 5.7
Rodney Hood UTA 205 5.5
Chris Andersen MIA 142 5.4
Joe Ingles UTA 388 5.3
Dante Exum UTA 391 5.1
Chase Budinger MIN 229 5.0
Willie Green ORL 317 4.7
Chris Johnson PHI 187 4.6
Mike Miller CLE 158 4.5
Andrew Nicholson ORL 109 4.4
Jason Maxiell CHO 186 4.1
Brandon Rush GSW 102 3.1
Jimmer Fredette NOP 113 2.1
JaKarr Sampson PHI 141 1.2
Hassan Whiteside MIA 9 34.9
Dwight Powell BOS 7 30.5
C.J. Wilcox LAC 9 29.3
Jarnell Stokes MEM 34 28.0
Alex Kirk CLE 6 26.3
Tiago Splitter SAS 18 26.1
Shayne Whittington IND 41 25.0
David Lee GSW 7 24.5
Mike Muscala ATL 75 22.0
Jeff Withey NOP 62 21.4
James Young BOS 18 21.0
Bruno Caboclo TOR 12 20.4
Meyers Leonard POR 55 18.8
Erick Green DEN 49 17.9
James Jones CLE 67 16.8
Ryan Kelly LAL 25 15.7
Greg Stiemsma TOR 41 15.1
Charlie Villanueva DAL 56 14.8
Elton Brand ATL 56 14.4
Jeff Adrien MIN 68 14.0
Landry Fields TOR 45 13.6
Noah Vonleh CHO 24 13.4
Dorell Wright POR 54 13.1
Malcolm Thomas PHI 71 12.5
Greg Smith DAL 63 12.3
Archie Goodwin PHO 63 12.2
John Jenkins ATL 24 12.2
Jerami Grant PHI 61 12.0
Jannero Pargo CHO 54 11.9
Thomas Robinson POR 67 11.7
Cory Jefferson BRK 61 11.7
Ish Smith OKC 44 11.5
Luigi Datome DET 12 11.3
E'Twaun Moore CHI 35 11.3
Sergey Karasev BRK 66 11.3
Glenn Robinson MIN 41 11.0
Jorge Gutierrez BRK 44 10.8
Steve Novak UTA 54 10.7
Nick Calathes MEM 31 10.3
T.J. Warren PHO 84 10.3
Jordan Adams MEM 41 10.1
Devyn Marble ORL 5 9.8
Jusuf Nurkic DEN 82 9.5
Ian Clark UTA 49 8.9
Brendan Haywood CLE 25 8.5
Shavlik Randolph PHO 58 8.4
Spencer Dinwiddie DET 78 8.2
Cleanthony Early NYK 54 8.2
Jeremy Evans UTA 15 8.0
Nazr Mohammed CHI 40 7.9
A.J. Price CLE 5 7.8
Will Cherry CLE 69 7.7
Joel Anthony DET 59 7.0
Tyler Ennis PHO 41 6.5
Louis Amundson CLE 59 6.3
Markel Brown BRK 30 5.9
Jared Cunningham LAC 72 5.7
Ekpe Udoh LAC 31 5.2
Troy Daniels HOU 90 5.0
Nate Wolters MIL 98 4.7
Shannon Brown MIA 89 3.5
Andrei Kirilenko BRK 36 3.4
DeJuan Blair WAS 34 3.1
Ognjen Kuzmic GSW 32 3.0
Kalin Lucas MEM 6 2.9
Xavier Henry LAL 86 2.8
Lucas Nogueira TOR 9 2.7
Drew Gordon PHI 71 2.5
Ronny Turiaf MIN 19 2.3
Chris Douglas-Roberts LAC 45 2.0
Justin Holiday GSW 47 1.8
Cartier Martin DET 89 1.7
Danny Granger MIA 84 1.3
Gal Mekel NOP 43 1.1
Gerald Wallace BOS 62 0.7
Cameron Bairstow CHI 39 0.1
Eric Moreland SAC 2 -0.1
Glen Rice WAS 43 -0.5
Andre Dawkins MIA 13 -1.7
Grant Jerrett OKC 9 -3.7
Darius Miller NOP 43 -3.9
Julius Randle LAL 14 -8.3
Zoran Dragic PHO 2 -15.5
Malcolm Lee PHI 2 -18.2
Clint Capela HOU 12 -19.0
Russ Smith NOP 11 -30.0
Does that make PER any better?
Re: Front Porch (for casual chat)
Posted: Fri Dec 12, 2014 6:51 pm
by Crow
Some, but it still over rewards usage imo and completely lacks shot defense. Never got anywhere with J Hollinger about update to pbp era data and got no reply from Ben Alamar now with ESPN either.
Re: Front Porch (for casual chat)
Posted: Sat Dec 13, 2014 12:25 pm
by mystic
Crow wrote:Some, but it still over rewards usage imo and completely lacks shot defense. Never got anywhere with J Hollinger about update to pbp era data and got no reply from Ben Alamar now with ESPN either.
Why would that be better? I didn't change the underlying part which is supposed to measure "player efficiency", just added a team adjustment, which may or may not attribute value correctly to the players on a specific team. What would be your comparison in order to justify the idea of "some better"?
Also, you say that it "over rewards usage". Why do you say that? Is there an objective measurement available to justify such statement? Maybe other metrics undervalue usage? If I remember correctly, EvanZ came up with something similar when evaluating players via pbp data (or was that just for the respective break-even points in terms of shooting efficiency? not quite sure on that point).
And the last point is obsolete to me, because you basically critize PER for something which it isn't. No boxscore-based metric can have a sufficient shot defense included, because the boxscore does not contain that knowledge. That they never came back to you regarding a possible "update to pbp era" is also not particular helpful. It just seems as if you critize PER based on something you want that to be, not on what it is supposed to be. PER can split up players pretty well based on their offensive contribution in a respective role. In order to use PER in a useful fashion, it is necessary to understand the player's role first. You may use PER to compare 1st, 2nd or 3rd options, respectively, and then get a pretty good grasp on their respective offensive impact in that specific role. If you compare two 2nd option players, the one with the higher PER should be the one with the higher impact. It gets tricky for point guards, because compared with RAPM, it undervalues successful passes, while it gives more weight to scoring. PER is not good at evaluating defensive impact, at all. But other boxscore metrics have a similar issue, simply just based on the fact that defense is not represented well in individual boxscore data.
I'm also pretty sure that an adjPER version does better in terms of predicting out of sample, because part of that predictive value of other metrics is simply contained within the team data and more team-related than player-related. How do I know this? Well, the underlying method of my old SPM did also not contain an explicite team adjustment (just a small correction based on the fact that the used equation for player individual possession did not sum up to the team-level amount of possessions), and I saw how the predictive power increased after a team-level adjustment was made.
Maybe it would help, if people would first try to beat PER's in-season correlation to win% without an explicite team-adjustment (whether that is included via starting at the team-level and then subsequently devide the value among the team's players (like statman, win shares or Berri is essentially doing it) or via explicite team-adjustment afterwards (like BPM). Maybe people would realize that PER isn't that bad at the particular thing it wants to do: evaluate individual players based only on boxscore entries.
Re: Front Porch (for casual chat)
Posted: Sat Dec 13, 2014 12:53 pm
by permaximum
I think PER and WP has no usage at all. DWS or DRTG comes handy to predict at team level. When I get yearly roster turnover into account DWS still prevails. However, the roster turnover rate at 2-year-span is more than it can take. So, I would only use it for the next year.
Re: Front Porch (for casual chat)
Posted: Sat Dec 13, 2014 1:59 pm
by mystic
permaximum wrote:I think PER and WP has no usage at all.
Did you test PER with a team-level adjustment? Also, keep in mind that elements of PER may be covered by OBPM and OWS. Additionally, you only tested it for y1 performances. How about y2, y3 or y4? As Neil's result showed, PER becomes better at predicting further into the future than WS for example, despite being behind in y1 or y2. The question is: What do you pick up in the y1 correlation? The player's "goodness" or just the "player's role"?
Re: Front Porch (for casual chat)
Posted: Sat Dec 13, 2014 3:12 pm
by permaximum
Well, actually I tested Y-1, Y-2, Y-3, Y-4, Y-5 and all combinations of those seasons with all combinations of metrics and even advanced stats. PER and WP gets relatively better (especially PER) going back but still it's not good enough to catch WS or BPM.
Re: Front Porch (for casual chat)
Posted: Sat Dec 13, 2014 4:24 pm
by mystic
Great. Would you mind making your results publically available. That would likely save Neil some trouble and time. Also, did you account for possible in-sample seasons (like 2001 to 2014 for BPM)?
Regarding PER: Did you try to use a team-adjusted version, in the sense to let the team average PER sum up to the overall performance? You can convert PER into a +/- like value by subtracting the league average and deviding it by the leaguewide standard deviation (z-score). Then use minute-weighted z-scores and subtract the overall team performance level (like Ortg-Drtg= Net). The difference can be added into the z-score, then you can multiple the value by the standard deviation again while then adding 15 (as the average value). Rather simple solution, I would say.
Re: Front Porch (for casual chat)
Posted: Sat Dec 13, 2014 7:05 pm
by Crow
The PER adjustment is useful if the underlying formula is accurate / fair. So yeah you right, it probably isn't right to say that the formula based adjustment helps with certainty but it probably does on average.
I think Evan's comparison was just break even point and PER's was still higher than his.
I want PER or any roll-up metric to try to measure total player impact. Not odd to want that. Ignoring shot defense besides blocks is a major flaw to me when there are ways to do something, probably better than doing nothing. But most accept, so use it as is and I won't as much.
PER's strength is measuring usage based offense. Strip out defense, the partial defense, and then the 40-50% of the defense that is missing isn't an issue in a strictly offensive measure. The usage issue and over reward of assists remain. I believe it has been shown that no major metric correlates more with usage than PER and it is probably top 1-3 on assists. Unlikely that these are the right choices when metrics that correlate less with usage and give less for assists have higher overall correlations with team results. But... in a metric blend, you can hedge on this uncertainty.
Re: Front Porch (for casual chat)
Posted: Sun Dec 14, 2014 10:36 am
by permaximum
mystic wrote:Great. Would you mind making your results publically available. That would likely save Neil some trouble and time. Also, did you account for possible in-sample seasons (like 2001 to 2014 for BPM)?
Regarding PER: Did you try to use a team-adjusted version, in the sense to let the team average PER sum up to the overall performance? You can convert PER into a +/- like value by subtracting the league average and deviding it by the leaguewide standard deviation (z-score). Then use minute-weighted z-scores and subtract the overall team performance level (like Ortg-Drtg= Net). The difference can be added into the z-score, then you can multiple the value by the standard deviation again while then adding 15 (as the average value). Rather simple solution, I would say.
Well, my work was a mess since I tried to come with a blend very quickly. After deciding on metrics and weights and I scrapped much of it. However, I saved at a crucial point. Still it would take a day for me to come up with the results again. If people give me their data along with b-r codes, I can do a quick look at metrics.
I can't lie I don't know what's z-score or stuff like that, I learn by trying or looking for "something in my mind" online. However I get what you mean. I do something like that since my purpose was to come up with a blend. I adjust all metrics' average and total sum up to the league overall and average which is 0. I used these adjusted versions of metrics in the blend. The blend doesn't get adjustment after. And I tested every metric's adjusted and unadjusted version. Adjustment help all metrics especially PER.