Statistical Analysis. Humor. Knicks.

Sunday, October 26, 2014

Knicks Morning News (2014.08.22)

  • [New York Daily News] Carmelo Anthony eager for Knicks turnaround, believes they will make playoffs (Fri, 22 Aug 2014 04:53:26 GMT)
    He wasn’t suited up to play, but an optimistic Carmelo Anthony was on the hardwood on Thursday talking turnaround for the 2014 season.

  • [New York Times] Source: Wolves Get Young as Part of Love Trade (Fri, 22 Aug 2014 06:08:48 GMT)
    The Minnesota Timberwolves have secured the final piece they were looking for in a blockbuster Kevin Love trade.

  • [New York Times] Sports Briefing | Pro Basketball: Lynx’s Moore Named M.V.P. (Fri, 22 Aug 2014 04:17:42 GMT)
    Maya Moore of the Minnesota Lynx earned the W.N.B.A.’s Most Valuable Player award Thursday.

  • [New York Times] AP Source: Phoenix Has Top Coach, Defensive Player (Fri, 22 Aug 2014 03:47:51 GMT)
    Phoenix’s Sandy Brondello is the WNBA’s coach of the year and Mercury center Brittney Griner is the defensive player of the year, said a person close to the situation.

  • [New York Times] AP Source: Griner’s WNBA Defensive Player of Year (Fri, 22 Aug 2014 03:32:54 GMT)
    A person close to the situation says Phoenix’s Brittney Griner has won the WNBA defensive player of the year award.

  • [New York Times] Cancer-Stricken Cooper Leads Dream Into Playoffs (Fri, 22 Aug 2014 02:20:52 GMT)
    Michael Cooper remembers the day well. The sun was shining bright, yet the notorious Atlanta heat wasn’t too stifling. The sky was a brilliant blue, speckled with only a few puffy white clouds.

  • [New York Times] Lynx’s Moore Wins WNBA MVP Award (Fri, 22 Aug 2014 01:08:50 GMT)
    Maya Moore put up incredible numbers all season for Minnesota. Now she’s got her first WNBA MVP award.

  • 96 comments on “Knicks Morning News (2014.08.22)

    1. GoNyGoNYGo

      My favorite article from last night was this one about JR Smith who wants to be a leader.

      LOL.

      If he leads by example and other players follow, the Knicks are in trouble!

    2. Farfa

      About WP, Dirk and role players:

      I am guilty as anyone on this board of barking against Jowles in the past, when he said (even if it was loaded with evident defiant intention) that Mason Plumlee should have been Rookie of the Year because WP48 says so. It seems kinda dumb to say that a limited offensive player like Mason could win anything like that, especially considering that his scoring (and his general offensive contributions) is so much depending on his teammates.

      But I have to admit that in the last months I’ve put myself squarely in Jowles’ camp. Maybe not about everything he says (and certainly not about the way he says it, even though he mellowed his curmudgeon stance quite a bit), but surely about the notion that scoring ability is overrated. It really is, unless we’re talking about supremely efficient scorers – think Durant, Curry, LeBron, Harden and yes, Dirk. We obviously can’t compare Calderon to Dirk, or Dalembert to Dirk, but that is only because they play differently (it’s not about position. It’s about style/skills). You can obviously compare, for example, Dirk to Melo. And you can conclude that 35 years Dirk is a slightly less useful player than 30 years Melo (but that is only because at this stage Dirk is only a scorer with scarce mobility). You can also compare Dalembert to Chandler, and conclude that Chandler is a better player (we already knew that – but Dalembert is better than, let’s say, Spencer Hawes). You can also compare Calderon to other starting PGs and, without delving into beating a figuratively dead Felton-horse, conclude that his savviness shows up in WP48 and that he is in fact better than Deron Williams (yes, right now it is), last year Rondo, Damian Lillard (I think it’s true), Kyrie (of course Jose is better), and even last year John Wall (what! I don’t necessarily agree with the last one).

      All of this to say that I don’t know why we’re still arguing in about WP48. It is a useful stat, not the Bible, but…

    3. GoNyGoNYGo

      Clearly Pablo Prigioni was more important to the Knicks than Carmelo Anthony last year. And Al-Farouq Aminu, James Johnson, Terrence Jones, Draymond Green, DeBarre Carroll, Trevor Booker and Andre Miller were better.

      Sorry. I just don’t buy that at all. I’ll use the eyeball test.

    4. Brian Cronin

      Late last night I saw that the Wolves were getting Thaddeus Young just for a future first rounder and expiring contracts. I thought for sure that they’d have to trade anthony Bennett to get Young, so just a first rounder is awesome. The Love deal now looks even better than it did before and it already looked good. For a guy they were losing no matter what, they’re getting Wiggins, Bennett and Thaddeus Young. That’s good enough that they should still be able to compete for the playoffs this season (they likely won’t make it, but they’ll be competitive).

    5. Farfa

      Clearly Pablo Prigioni was more important to the Knicks than Carmelo Anthony last year. And Al-Farouq Aminu, James Johnson, Terrence Jones, Draymond Green, DeBarre Carroll, Trevor Booker and Andre Miller were better.

      I didn’t make myself clear, evidently. I just wrote this:

      We obviously can’t compare Calderon to Dirk, or Dalembert to Dirk, but that is only because they play differently (it’s not about position. It’s about style/skills).

      There is nobody on that list who can be positively compared to Melo in terms of skills/style of play. So that doesn’t make sense, as you correctly say. But if you look closely, that list is composed of very capable basketball players who I would welcome here much more happily than Jason Smith, for example.

    6. Farfa

      @2 Harden? …i really just think its Durant and lebron right now

      If we’re talking about likability of scoring efforts, I agree that Harden has become almost unwatchable. But he’s really efficient in his scoring outputs, if only because he gets to the line at will (flopping badly).

    7. Z-man

      But it can be used to compare Olajuwan to Wallace and Chandler. And the comparison fails miserably. And if it fails there, how many qualifying statements do you have to make, like the ones in #2, before it works?

      And that’s NOT how Jowles uses WP. He has consistently equated having a higher WP48 score with “being better at basketball.” He would argue that Chandler and Wallace were actually more productive and therefore better at basketball than Hakeem in the years cited. When faced with an indefensible position (e.g. Ed Pinkney and Popeye Jones were “better at basketball” than Hakeem in 1994) he would either cop out by not responding, or he would make some kind of vague blanket statement about the proper way to use WP in determining a player’s value relative to what they get paid.

      WP is not a relative stat. It’s name definitively identifies it as an absolute stat. It says that players with higher scores produce more wins, which is the objective of basketball, and that in all cases, guys with higher WP are better. Once you start qualifying WP with role and interaction effects, it can no longer be used to definitively determine who is better at basketball any more than PER can.

    8. Brian Cronin

      Yeah, Harden’s entire awesomeness is wrapped up in his scoring ability. That’s the only reason he’s a star. Well, he’s also a surprisingly strong assist guy for someone who seems like such a black hole out there, but really, it’s the scoring that makes him so good.

    9. Farfa

      @4 Brian

      Well, given that Love was walking away, Saunders did a very good job, at least in the “hope for fans” department. I’m not convinced a starting five of Rubio, Martin, Wiggins, Young and Pekovic will yield cool results, but they should be fun enough.

    10. Farfa

      But it can be used to compare Olajuwan to Wallace and Chandler. And the comparison fails miserably.

      I fail to see where it fails. It says that Olajuwon has a career 0.295 WP48, Ben Wallace (since I think he’s the one you’re referring to) has a career 0.275 WP48, and Tyson Chandler should be hovering around 0.270 WP48 (I can’t seem to find exact results). I think it reflects exactly the real hierarchy.

      If you’re talking about the fact that peak Ben Wallace had WP48 in excess of .340, well… Maybe you don’t remember how freakishly dominant he was on defense and on the glass. We have a tendency of heralding only scoring guys as “stars”, while guys who fails to score more than 15ppg are glorified role players. Well, they’re not. They are absolute stars, only their contributions rarely put butts on seats. Dennis Rodman was a star. Ben Wallace was a star.

    11. DRed

      But it can be used to compare Olajuwan to Wallace and Chandler. And the comparison fails miserably.

      It’s impossible to compare Olajuwan to Wallace because everyone knows Olajuwan was better than Wallace so therefore anything that suggests that a certain Wallace season was more productive than a certain Olajuwan season is wrong because Olajuwan was better because everyone knows Olajuwan was better.

      Can’t argue with that shit.

    12. DRed

      Yeah, Harden’s entire awesomeness is wrapped up in his scoring ability.

      His beard is awesome too, tho

    13. er

      @9 i agree i just dont think hes elite. Look at the playoffs. When he stops getting calls he looks rather ordinary.

      @10
      Idk…how close is minny to at least breaking even in that trade? Heck they may have even won it. They got 2 solid defenders in Thad Young and Wiggy as well as Wiggy’s potential. And the wildcard in Bennett for K Love and a 1.

      @13 im def jelly of that lol

    14. Farfa

      Idk…how close is minny to at least breaking even in that trade? Heck they may have even won it.

      They surely won it if we consider the specific conditions (Love at possibly the lowest price ever, since he said he wanted to be traded). But could they have done better? I would have tried getting Bledsoe and Markieff Morris from Phoenix.

      All in all, a cool trade for the Wolves anyway.

    15. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      But it can be used to compare Olajuwan to Wallace and Chandler.

      Funny how all three of them were dominant centers who started on outstanding teams that won NBA championships…

    16. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      It’s impossible to compare Olajuwan to Wallace because everyone knows Olajuwan was better than Wallace so therefore anything that suggests that a certain Wallace season was more productive than a certain Olajuwan season is wrong because Olajuwan was better because everyone knows Olajuwan was better.

      Olajuwon scored lots more points and had an unguardable offensive move. The only thing Ben Wallace could do is stay in front of his man and box out. Obviously one of those skillsets is better than the other. Obviously. Does WP48 match our hypothesis? No? It’s wrong.

    17. lavor postell

      Yeah Hakeem really sucked at boxing out his man and staying in front of his man. What a one dimensional scorer that guy was.

      Also didn’t realize WP gave points out for being good at staying in front of your man.

    18. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      i agree i just dont think hes elite. Look at the playoffs. When he stops getting calls he looks rather ordinary.

      Free Throw Rate, career, regular season: 0.543

      Free Throw Rate, career, playoffs: 0.540

      I’m not saying you’re wrong about his value, but when it comes to the frequency with which he gets to the line, you’re, uh, wrong. And before you go on about his poor FTr in 2013-14, stop. It was a 6 game sample.

    19. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      Yeah Hakeem really sucked at boxing out his man and staying in front of his man. What a one dimensional scorer that guy was.

      Also didn’t realize WP gave points out for being good at staying in front of your man.

      Hakeem is an all-timer on both ends of the floor. Arguing that he was obviously more productive than also-rans Ben Wallace and Tyson Chandler is the issue. I love Hakeem, but the reason this conversation is happening is because he scored more points than the other two.

      http://www.boxscoregeeks.com/players/1157-hakeem-olajuwon

      Over his career, he scored as many PPS as the average C. If we can just fit that into the debate somehow…

    20. Farfa

      Jowles, how did you find that page? Because I can’t seem to find data before 2012-13 about WP48 (the numbers on my post were taken from old wow articles, and they are not the ones that show up in that box score geeks page).

    21. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      Go to the team pages and click on year, then individual names. They haven’t really fixed the search function.

    22. Z-man

      Did you read my post in the last thread?

      in 1989-90, Olajuwan led the league in DReb%, Blocks, DRating and Defensive Win Shares. He was only 7th in steals (The only other center in the top 20 was the Admiral at #20) But he was only 19th in OReb% even though he was 6th in overall offensive rebounds.

      Look at the rebounding list from that year, you’ll see names like Ewing, Rodman, Oakley, Malone (both Moses and Karl) Barkley, Horace Grant, Buck Williams, etc.

      And he actually had a better year relative to those peers in 1993-94.

      As to Wallace in 2002-03, he also led the league in DRebs and DReb%, and also ORebs (he had 43 more than Amare) but was only 7th in OReb%. He was 3rd in blocks (behind Theo Ratliff and just ahead of Adonal Foyle.) His main rebounding competition was Shaq, Garnett and Duncan. He wasn’t even in the top 20 in steals. Frankly, I don’t see anything “freakish” about that relative to Hakeem’s 1989-90, and that doesn’t even get into scoring.

    23. DRed

      . Frankly, I don’t see anything “freakish” about that relative to Hakeem’s 1989-90, and that doesn’t even get into scoring.

      Good thing you didn’t get into scoring, because Hakeem wasn’t particularly good at it in 1989-1990. Compare Hakeem to the Admiral that year and come back and tell me that Wins Produced values rebounding over scoring.

    24. GoNyGoNYGo

      @6 Sure, if you look at the list, they are all productive in some way. But when I see that Pablo Prigioni is ranked 27th in the NBA in WP48 last year, it tends to make me discount the entire rating system. Yeah, and Jeff Adrien is ranked 13th. Right.

    25. GoNyGoNYGo

      Using my own formula (PPG+RPG+APG)*(Mpg/48) I get a list that makes so much more sense:

      NAME Val
      Durant, Kevin 44.89
      Love, Kevin 43.06
      James, LeBron 40.45
      Anthony, Carmelo 38.65
      Griffin, Blake 37.50
      Cousins, DeMarcus 37.36
      Aldridge, LaMarcus 36.88
      Curry, Stephen 36.79
      Harden, James 36.14
      Jefferson, Al 34.83
      Westbrook, Russell 34.40
      Paul, Chris 34.14
      Davis, Anthony 32.41
      Howard, Dwight 32.27
      Wall, John 32.18
      George, Paul 32.07
      DeRozan, DeMar 31.03
      Nowitzki, Dirk 30.59
      Irving, Kyrie 30.54
      Gasol, Pau 30.35

    26. Farfa

      Using my own formula (PPG+RPG+APG)*(Mpg/48) I get a list that makes so much more sense:

      It makes more sense to you, and that’s ok. But any list that has Melo 4th, Cousins 6th, DeRozan 17th, Kyrie 19th and this Pau Gasol 20th is something that does not resemble to me any fragment of productive NBA basketball. It does make a nice playground list, though.

      Look, we’re both talking to deaf ears, I guess. I say that there are things that WP48 measures much better than, say, PER or your custom formula. There are also inherent flaws in WP48 that make it overestimate Jeff Adrien and Prigioni. But to my eyes Adrien and Prigioni are capable basketball players (if devoid of particular skills), while DeRozan and Kyrie, as the most glaring examples, are extremely skilled basketball players who aren’t quite good at playing 5-on-5 basketball. If I have two spots remaining to fill my team’s roster I take every single time 13-14 Adrien and Prigioni. They won’t win me many games, but sure as hell they won’t lose me any. They always do know what they can do and they do it without getting in the way of their teams.

      Just one more thought.

      This is the roster of 2004 USA national team:

      Iverson, Marbury, Wade, Boozer, Melo, LeBron, Okafor (the Laettner award winner), Marion, Amar’e, Duncan, Odom, R. Jefferson.

      Much talent, right? Sure. Third place, though. But swap Marbury (.142 WP48 in 03-04, not that bad) and Iverson (-.019 WP48 in 03-04. Yikes!) with Prigioni and Calderon and they probably win the Olympics. You need productive players. Not skilled but unproductive ones.

    27. lavor postell

      But to my eyes Adrien and Prigioni are capable basketball players (if devoid of particular skills), while DeRozan and Kyrie, as the most glaring examples, are extremely skilled basketball players who aren’t quite good at playing 5-on-5 basketball. If I have two spots remaining to fill my team’s roster I take every single time 13-14 Adrien and Prigioni. They won’t win me many games, but sure as hell they won’t lose me any. They always do know what they can do and they do it without getting in the way of their teams.

      How many times have you watched Jeff Adrien play?

    28. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      Bro I can’t even remember which pair of pants I wore to work on Monday, and I was there the whole time.

      But what if I had a spreadsheet that said, “Brown, Gap, Standard Cut, Button Fly.” Could I maybe then guess that (1) I wore pants and (2) they were the brown Gap standard cuts with the button fly (a really great-fitting pair of pants for the $25 I bought them for in a Philadelphia airport in late February) and (3) that they were the brown pants I believe them to be?

      I don’t know, man. The data is big, man.

    29. lavor postell

      I’m not arguing about Hakeem Olajuwon in 1989. All I asked Farfa was how many times he’s watched Jeff Adrien play basketball, which apparently is some kind of sin according to Jowles. I guess basketball is a lot more fun if you just watch it on boxscoregeeks or WoW.

    30. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      lavor,

      My points here aren’t even that WP can automatically tell you how amazing a player is. My point is that if you begin with the assumption “Hakeem Olajuwon was more productive than Tyson Chandler” and then you say that WP48 is wrong because it calls both of them a top-production talent (IN THEIR RESPECTIVE ERAS, RELATIVE TO THE STRENGTH OF THEIR LEAGUE AND POSITION) and doesn’t account for the fact (yes, fact) that Hakeem had a much more diverse offensive skillset, then you are being a real silly guy and should stop it.

    31. lavor postell

      Bro I can’t even remember which pair of pants I wore to work on Monday, and I was there the whole time.

      But what if I had a spreadsheet that said, “Brown, Gap, Standard Cut, Button Fly.” Could I maybe then guess that (1) I wore pants and (2) they were the brown Gap standard cuts with the button fly (a really great-fitting pair of pants for the $25 I bought them for in a Philadelphia airport in late February) and (3) that they were the brown pants I believe them to be?

      I don’t know, man. The data is big, man.

      This isn’t even relevant to the conversation.

      The difference between me and you is I can value WP as a stat without blindly accepting it as the end all be all of player evaluation. If WP gives DeAndre Jordan a very high score that tells me he’s good at basketball. If I then take that number and make arguments that he’s the best center in the league and was better and more valuable to the Clippers than Chris Paul when he was shut down by the immortal Kendrick Perkins over the course of 6 games that’s a completely different matter.

    32. Farfa

      How many times have you watched Jeff Adrien play?

      At least 40. I’m a League Pass addict. And I will say this: he’s not starting PF material, that’s for sure. But I never saw him make a boneheaded play. He’s good (and very good at that) for 10 minutes per night in the token 4th big role in a rotation, and he’s the classic guy who never gets noticed on losing teams. Put him on a well coached team (say, the Clippers) and he’s a better version of ’08 Leon Powe (who got everyone falling head over heels for him, don’t you remember?). The MJ Bulls always had a Jeff Adrien on their team. Ditto for the Shaq-Kobe Lakers.

    33. lavor postell

      lavor,

      My points here aren’t even that WP can automatically tell you how amazing a player is. My point is that if you begin with the assumption “Hakeem Olajuwon was more productive than Tyson Chandler” and then you say that WP48 is wrong because it calls both of them a top-production talent (IN THEIR RESPECTIVE ERAS, RELATIVE TO THE STRENGTH OF THEIR LEAGUE AND POSITION) and doesn’t account for the fact (yes, fact) that Hakeem had a much more diverse offensive skillset, then you are being a real silly guy and should stop it.

      You don’t think the diverse offensive skill set is valuable?

      I would think that if Hakeem was strictly an offensive options as the roll man in high PNR he could emulate Chandler’s amazing efficiency. It’s certainly a valuable skill that Chandler brings to the table, but it isn’t enough to take up the numerous possessions throughout the course of a game.

      I have no idea how I could do it, but if you tracked the type of possessions that Chandler uses and compared them to similar possessions by Olajuwon I would bet that he scored at a very similar level. The difference is that Olajuwon could also score in other ways at very good efficiency so he was able to use more possessions which is incredibly valuable over the course of an 82 game season and a playoff run. Olajuwon was also a great rebounder and defensive player like Chandler.

      That doesn’t mean that Chandler is some scrub with limited value. IMO it just means that Olajuwon is a more valuable player and I think it’s absurd to put the two on the same level. We’ll never agree on that probably.

    34. lavor postell

      @Farfa

      Adrien is a good player for 10 minutes a night and that has value, but WP is telling me he’s the 13th best player in the league and that doesn’t make any sense to me based on the fact that a basketball game is 48 minutes long.

      WP rates Adrien as a better basketball player than Dirk Nowitzki last year. Do you think if you replaced Dirk with Adrien the Mavs get to the playoffs in the West and push the Spurs to 7 games?

    35. DRed

      LOL, Chris Paul. You’re really calling the dude who was shut down for 5 games by old man Chauncey Billups a star? Get real.

    36. Farfa

      more valuable to the Clippers than Chris Paul when he was shut down by the immortal Kendrick Perkins over the course of 6 games

      1) Sample size.
      2) No single player can shut down DeAndre Jordan. He doesn’t play a type of offense that can be shut down. You can deny the pick and roll, but that’s a different matter.
      3) Put DeAndre Jordan on the Rockets and they go to the Western Finals.
      4) While non relevant to this argument, Perkins is actually still a great man-to-man defender. I wouldn’t play him more than 5 minutes per game in most of the games, but when paired with a low-post scorer with middling athleticism he’s good.

    37. lavor postell

      Did I ever say that DeAndre Jordan is a bad player or isn’t a valuable player?

      If you want to believe that he’s the best center in the league that’s fine and you can make a reasonable argument for it. Personally I don’t.

    38. DRed

      You don’t think the diverse offensive skill set is valuable?

      Only if it leads to you making buckets. Akeem in 89 wasn’t that efficient. Still a hell of a ballplayer because of the other things he brought to the table, but not because he averaged 24 points on 19 shots a night.

    39. Farfa

      Do you think if you replaced Dirk with Adrien the Mavs get to the playoffs in the West and push the Spurs to 7 games?

      I already answered to that and the answer is no. But WP48 tells you that in his limited minutes Adrien was super productive. It would also probably tell you that Adrien, given more minutes, would become a .190 WP48, which is still much better than, I don’t know, Boozer or McRoberts or this Garnett or Patrick Patterson (all of them saw major minutes in last year’s playoffs).

    40. lavor postell

      LOL, Chris Paul. You’re really calling the dude who was shut down for 5 games by old man Chauncey Billups a star? Get real.

      Chauncey Billups was really good at basketball in 08-09. Perkins was not good at basketball last season.

      1) Sample size.
      2) No single player can shut down DeAndre Jordan. He doesn’t play a type of offense that can be shut down. You can deny the pick and roll, but that’s a different matter.
      3) Put DeAndre Jordan on the Rockets and they go to the Western Finals.
      4) While non relevant to this argument, Perkins is actually still a great man-to-man defender. I wouldn’t play him more than 5 minutes per game in most of the games, but when paired with a low-post scorer with middling athleticism he’s good.

      1. Fair enough
      2. And this is why I don’t think he is the most valuable center in the league. I doubt this is an argument we’ll agree on because I think Jordan is an overrated defensive player, although he has improved a lot on that end.
      3. This isn’t an argument it’s just a statement that you’re not even bothering to support and one that I disagree with. Howard isn’t what he once was, but he’s still very good.
      4. Perkins might be a good man-to-man defender, but if you’re supposedly the best center in the league alongside Andre Drummond shouldn’t that matchup be a massive win for your team over the course of 6 games instead of a push?

    41. yoda4554

      I’m not sure why this discussions is proceeding as it is. WP is a formula–but then, royal astronomers in the 16th century had really good formulas for a geocentric system. That doesn’t make geocentrism valid. In particular, it was declared invalid because it was bad at predicting all these little consequent errors in planetary motion. As has been conceded, WP is not especially predictive, so we don’t need to pay much attention to it.

      Incidentally, here are two major weaknesses of WP. First, yes, it really does overvalue rebounds. Here’s why: its base formula values rebounds equally to steals. That means it considers a defensive rebound equal to single-handedly ending your opponent’s possession. By watching basketball, though, we know that’s not how rebounds work–the stop is a collaboration between whoever shuts down the shooter and whoever collected the rebound. It would be great if we had a stat called “stops,” in which players were assigned credit for causing shooters to miss, in which case advanced stats could split defensive credit between them and the rebounders. We don’t, though, and WP just gives full credit to the rebounder. Thus, WP overrates all players whose rebounding significantly supersedes their defense (e.g., DeAndre Jordan and Kevin Love).

      Second, its positional system misunderstands turnovers. Whenever any team is on offense, it needs a player to handle it and not turn it over. Ideally, we’d track individual player possession and evaluate turnovers by time-handling-ball. As we don’t track time-handling stats, though, WP simply assumes that guards handle the ball more than forwards and centers and assumes they’ll have more turnovers. Consequently, a team whose PF handles the ball more than its SG (e.g., the Knicks playing Carmelo Anthony and Ronnie Brewer) will find its SG getting undue credit for low turnovers and its PF getting penalized too much for them.

    42. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      I’m certain that defensive rebounds have been adjusted to reflect the likelihood that a teammate would collect the rebound in a player’s place.

    43. yoda4554

      “I’m certain that defensive rebounds have been adjusted to reflect the likelihood that a teammate would collect the rebound in a player’s place.”

      As the formula page on Wages of Wins says, the correction has minimal impact: the correlation between the old and new system is 0.98. And again, the point is not that defensive rebounding is somewhat fungible, it’s that it’s only a fraction of what the defense does in any given instance of regaining the ball from the opponents, not the entire act.

    44. Alecto

      Isn’t WP48 designed to be a descriptive, not predictive statistical model? Also, what are we comparing WP48s predictive power to? Because most other metrics fare little to no better while being far less explanatory. If all you wanted was predictive power just do a weighted average of several models together or apply the prior year of point differentials and it’ll predict pretty well, but you’re going to learn nothing from that.

      I’m not going to argue on the other points, but acting as if low predictive power disqualifies a statistical model from consideration is quite silly. What is even more silly is to imply that the only thing that matters in a model (scientific, statistical, etc.) is predictive power.

      Also: the invalidation of geocentrism as a scientific theory was actually because of the descriptive power of heliocentrism rather than the predictive power–Galileo and others observed phenomena that flew in the face of geocentrism (the moon being pock-marked, Venus in all phases) and concluded that that the evidence did not match the Ptolemaic description of the cosmos. They then came up came up with new explanations that they then turned into predictive models. The Copernican system actually had no better predictions that the Ptolemaic system, until after Galileo and others did the requisite astronomical work and then came up with laws and models to explain their new descriptive discoveries. You’ve got the causal chain backwards on that one.

    45. yoda4554

      The fact that the predictive power of the Copernican model could only be verified until after Galileo is precisely why geocentrism held as the dominant model until well after Copernicus’s death. That’s the weakness of purely “descriptive” models–you can “describe” anything in any way you’d like retrospectively. Geocentric astronomers, for instance, found perfectly coherent ways to describe the kinds of phenomenon you mention within their paradigm. (This is one of the major points in Kuhn’s The Structure of Scientific Revolutions.) We find these explanations far-fetched and convoluted today, but there’s nothing inherently wrong with them so long as you offer them as mere descriptive explanations.

      Any scientific model that is not predictive has serious difficulties being falsifiable, which means that it verges on pseudoscience. (If something is not predictive, after all, it admits that it can not be experimentally verified.) After all, a model which looks purely at PPG is a perfectly good “descriptive” model of a basketball player’s quality. After all, the teams that score the most points over their opponents win the most! The only way to confirm that this is a bad model is to show that teams over-valuing PPG will win fewer games over time than teams using some other model of player valuation.

    46. iserp

      WP48 is not even a good descriptive metric. Take a look at Chris Bosh WP48 in Miami or in Toronto. Or David Lee WP48 in NY or in GS. WP48 numbers are as much a number of the player as a number of the system they play in. OTOH, PER is quite stable even when players changes teams.

      You can argue that PER is not a good statistic to gauge a player’s value. But it is a reliable statistic. If you say a player has 22 PER, it is likely to be true the next year, or in a different team. It tells you something about the player. But if you say a player has .200 WP48, you are compounding the inherent value of the player with the role he plays. If he changes teams, his WP48 is likely to change, too.

    47. Robtachi

      Aren’t Eric Bledsoe’s contract demands of the Suns fairly ridiculous? He wants max money for… what, exactly? He hasn’t even played one full season of starter’s minutes at a high level and he wants the max? Sure, he could be worth that eventually, but he’s utterly out of his gourd if he thinks someone is giving him $17 million/year for 5 years based on half a season of .578 TS%, fairly mediocre defense and the potential that he develops into a MUCH more complete player.

    48. yellowboy90

      Maybe Eric Bledsoe wants out of Phx and is trying to negotiate his way out by asking for the max. It does seem silly for him to ask for the max given that he only really played one good year.

    49. Totes McGoats

      I read on a thread here within the past few days or so that someone would be happy if the Knicks shelled out a max deal to Bledsoe if he took the QO and was a UFA next offseason. I respectfully disagree with that notion. Not that I don’t like Bledsoe..I think he should mature into a really good 2 way PG. And he did have a nice season for PHX. But I don’t think he’s worth more than Lowry money. He hasn’t been healthy lately..but he’s been great when he played. He’s still not much of a shooter, though he’s improving. We still haven’t seen him handle a team without another PG alongside him and still put up 17 ppg and 5.5 apg. Why give him the max if he’s largely unproven? Would I love to see him as a Knick? Absolutely. But I fear that folks may be overrating him a tad because of his age and that dreaded word “potential”. Now..say PHX moves Dragic or finds a taker for Bledsoe and he plays at least 75 games and gets 18 and 6 or 7 while shooting 45% from the field and 36% from three, all while playing to his standard on D. I would be a lot less apprehensive on giving him more than Lowry. But man..just imagine a truly healthy Shump and Bledsoe in the same backcourt. That would be fun to watch.

    50. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      WP48 is not even a good descriptive metric. Take a look at Chris Bosh WP48 in Miami or in Toronto. Or David Lee WP48 in NY or in GS. WP48 numbers are as much a number of the player as a number of the system they play in. OTOH, PER is quite stable even when players changes teams.

      You say that it’s not good at description, and then talk about how it failed to predict (or account for?) the differences between Bosh’s value in Toronto and Miami, Lee in NY and GS, whatever.

      The metric is designed to tell you how productive (overall) a player is, and, broadly, how that production is achieved.

      It is not designed to explain whether Carmelo Anthony’s missed 17-foot jumper is better or worse for a team than Iman Shumpert’s missed layup.

    51. ptmilo

      Wins produced is flawed. Being flawed isn’t great, but being flawed with misleading hype is worse. WP is promoted as explaining player value, hoisting its massive r-squared in tow as ballast. The r-squared does not in fact do any of the “interesting” work that contributes to WP’s sometimes odd results. Rather, it shows the stunning result that teams that score more points per possession than they allow win reliably Bam. Book deal. What it does not show is how individual players contribute to that effort. That requires a translation from team efficiency to player contribution. That translation is not empirical and importantly flawed. It all comes down to the use of “possessions employed.”

      The underlying WP team-level regression uses a formula for possessions that equate the following as equally valuable because they use or gain possessions: A missed FG, a rebound, a turnover, a steal. From a player perspective, this is plainly silly. Is it possible that securing one defensive rebound (ignoring for now the minor future amendment related to the diminishing marginal value of rebounds which is not relevant to this point) is as valuable as a turnover is costly? Not even close. A defensive rebound turns a situation in which a team has ~65% of a chance of securing a possession into 100%. A turnover entirely eliminates 100% of a possession. But from a team level, this does not present an important problem and thus does not hurt a basic point-per-possession regression that defines possessions as FGA – oREB + .45FTA – TO. This point is slightly subtle.

      The team formula implicitly charges the shooter for the team’s failure to secure the offensive rebound. Similarly, the entire team on defense is charged for the failure to grab a dReb. But this doesn’t hurt a team-level regression, because someone on the team is still being charged, even if it’s the wrong person. In fact, we could get a similarly great r-squared (cont…)

    52. ptmilo

      leaving out rebounds altogether! Every rebound results in either a FGA, FTA or a turnover. You can add those up to get possessions and they will incorporate all of the team-level information contained in rebounds. Points per possession would drop roughly symmetrically for offense and defense. You could suddenly “explain” over 90% of NBA wins without mentioning rebounds.

      But this does not tell you that rebounds have no value and allow you to assign it zero value. Similarly, you cannot simply assign a rebound value of “one possession” as WP did before its very minor future adjustment. This critical mistake is at the root of WP’s biggest flaw. WP actually mitigates this flaw somewhat, by making several “adjustments” before coming up with WP48. These adjustments include blocks, assists, the aforementioned rebound change, and a feckless team defense adjustment, but the big one is the positional adjustment. The adjustments are not themselves arbitrary, but bring the underlying model even further from the original (and only) regression that connects team statistics to wins. In that sense they are highly arbitrary.

      These adjustments have the effect of covering up the original WP flaws. Specifically, the major mistake of excessively valuing rebounds, under-penalizing turnovers and over-penalizing shot attempts happens to be reduced with the position adjustment. This is not surprising. Almost any formula with a nonrandom bias will improve with a normalization factor like position adjustment. There is nothing special about wins produced versus other advanced statistical efforts (ignoring PER) except (1) its one big mistake (2) its promoters’ ridiculous assertions about the amount of information it contains about NBA player value and the evidence supporting it. Yes, efficiency matters and points are overrated. But WP adds zero beyond that. And subtracts quite a bit.

    53. Brian Cronin

      I love the Suns trying to trade Bledsoe for Love NOW. First of all, it won’t work, but more importantly, how in the hell do you try to make this move now?! It’s like a day away from the trade becoming official and now is when they’re going to try to make a move? Huh?!

    54. yellowboy90

      So does Minny think Young is going to opt in next summer? Did they just give up a 1st rounder to rent Young? I guess his bird rights transfers which give them leverage to overpay to keep him.

    55. JK47

      If you put a 500-minute minimum on it, the following players rank as top 20 players in WP48:
      2. Andre Drummond
      4. DeAndre Jordan
      7. Andrew Bogut
      8. Brandan Wright
      10. Chris Andersen
      13. Jeff Adrien
      19. Mason Plumlee
      20. Bismack Biyombo

      Those are 8 of the top 20 players in the NBA? Some of those guys are damn good, but it is pretty odd that almost half of the top 20 players in the league according to WP48 are low-usage centers/forwards that grab a lot of rebounds, and furthermore that only two or three of them managed to play major minutes. Either low-usage good-rebounding centers are the most undervalued asset in all of sports or WP48 is inherently biased to this type of player.

    56. iserp

      You say that it’s not good at description, and then talk about how it failed to predict (or account for?) the differences between Bosh’s value in Toronto and Miami, Lee in NY and GS, whatever.

      The metric is designed to tell you how productive (overall) a player is, and, broadly, how that production is achieved.

      It is not designed to explain whether Carmelo Anthony’s missed 17-foot jumper is better or worse for a team than Iman Shumpert’s missed layup.

      Kind of a circular logic, as Wages of Wins has defined “production” as what WP48 measures. (like intelligence and the iq test)

      Since WP48 doesnt explain the difference of Anthony’s missed 17-foot jumpers and Shumpert missed layups, when a coach decides that Anthony should only shoot layups, and Shumpert should be the one shooting 17-foot jumpers, then, their WP48 is going to change.

      So… WP48 measures talent as well as role.

      But since PER takes into account shooting volume (as a positive thing), it negates (at some extent) the role a player has. PER measures “talent” much better than WP48. Feel free to argue that “talent” does not mean more wins, but PER is a reliable measurement of a player.

      So, if you trade for player that has 22 PER, you know what you get. If you trade for a player that has .200WP48 you have to ask what role did he play. If you put DeAndre Jordan to play the Olajuwon role, his WP48 is gonna change (although his PER is not)

    57. Z-man

      WP’s biggest flaw for me is that it rewards guys for being extremely limited offensive players and penalizes guys for being expected to score by their coach, their teammates, their fans, and all students of the game aside from WP zealots.

      For example, I agree that Carmelo Anthony shoots some inefficient shots where there is clearly a better option.

      But to conclude that Tyson Chandler is a better offensive player because he shoots at a higher efficiency is rewarding someone for lacking skill, and therefore expectations based on having that skill.
      WP does not consider role in terms of “expected scoring output.” That’s how you wind up with the gross distortion that says Ed Pinkney produces more wins in 1989-90 than Hakeem Olajuwan.

      I have no problem with WP until hear the sickening general phrase “he’s better at basketball” based on WP score. It’s like saying that a dermatologist is a better physician than a brain surgeon on the basis of having a lower mortality rate.

      I would contend that if Hakeem Olajuwan was unable to shoot anything but dunks and layups, and had to play Tyson Chandler’s or Ben Wallace’s game, he might have set an all-time record for WP. But WP bizarrely penalizes him for being good, and rewards Chandler for being terrible.

    58. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      Chandler, Andersen and Jordan have one thing in common: they’re tall and they jump high for lobs. That’s a valuable commodity in the NBA, whether or not you think it’s harder than taking a 20-foot jumper.

    59. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      But WP bizarrely penalizes him for being good, and rewards Chandler for being terrible.

      This sentence makes no sense whatsoever.

    60. Z-man

      I was referring to Chandler’s limited responsibility as a scorer.

      And you’re missing my point. I agree that Chandler, Andersen and Jordan are valuable players, and are the 3 best players in their niche. But their niche only requires physical talents and a limited, easy to acquire skill set, and an acute awareness of one’s own limitations. They can’t be compared to guys who are primary scoring options.

      A good example of this is Wilt Chamberlain. He demonstrated that he could set scoring efficiency records at will just by shooting less and cherry picking. All of the top 10 centers could have done this. It’s just not that hard for the physically capable compared to volume scoring.

    61. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      Being 7’1″ is not an easy-to-acquire skillset. Neither is jumping high. Those three guys are mobile seven-footers who rebound at will and simply need someone to run the PnR with them.

      If being the roll man were that easy, centers around the league would be capitalizing the way those three do. Do you really think that every big man is capable of doing what they do? If so, why don’t they put up an career .644 eFG (!!!) in the playoffs like Andersen, or a .679 eFG% over the course of 2000 MP like Chandler?

      Every ISO attempt by any player is a missed opportunity to run the pick and roll. And that’s not “rewarding” Chandler.

    62. JK47

      Somebody needs to develop a stat for hoops that is similar to “leverage” in baseball. The concept of leverage is that not every out recorded is of equal value– getting a guy out in the 7th inning with a three-run lead and nobody on base has different value than getting a guy out in the 9th inning with two guys on and a one-run lead.

      In basketball, a contested 17-footer with 15 seconds remaining on the shot clock is a terrible shot, an unnecessary shot in a low-leverage situation. The same contested 17-footer with :01 remaining on the shot clock, however, is actually a highly valuable shot, even if it misses, because if that shot it not taken, the result will be a turnover and 100% certainty of the loss of the possession. Those two 17-foot jumpers really should not be treated the same.

      And this is not the kind of thing that evens out over a big sample size; players like Carmelo Anthony are routinely asked to take way more of these “bail-out” type shots than players like Tyson Chandler or DeAndre Jordan. In fact, for most teams, the default play when the offense has failed to generate a quality shot in the first 20 seconds of the shot clock is to just pass the ball to the best one-on-one player on the team and hope he can make something happen.

      How many times does this happen over the course of a season– how many FGA by a volume scorer like Melo are actually “bail-out” type shots? I don’t know, but I’m guessing it’s not an insignificant number. This is why adjusting for usage is important; high-usage players are often forced to take low-percentage shots like this. Start dumping the ball to Tyson Chandler every time there’s 3 seconds left on the shot clock and all of a sudden he will start to look like not such a great WP48 player.

    63. lavor postell

      Every ISO attempt by any player is a missed opportunity to run the pick and roll. And that’s not “rewarding” Chandler.

      This is bullshit. Every single isolation isn’t of equal value and often times will occur because unless Chandler gets the ball in a position to immediately finish he will kick it back out and force the offense to reset with less time on the clock.

      If being the roll man were that easy, centers around the league would be capitalizing the way those three do. Do you really think that every big man is capable of doing what they do? If so, why don’t they put up an career .644 eFG (!!!) in the playoffs like Andersen, or a .679 eFG% over the course of 2000 MP like Chandler?

      Yes, those guys are very valuable NBA players. I don’t think anybody will question that, but how many possessions can those guys use in a game with that limited offensive skill set.

      Chandler’s career high usage is 18.1, Andersen’s is 19.9 and Jordan’s is 15.9. The shots they take are very efficient and they are also very good at finishing off of the high PNR. Unfortunately those shots are extremely hard to generate more than 7-8 times a game and since they can’t do anything else it is virtually impossible for them to up their usage even when they get have a huge mismatch with a smaller guard.

      Somebody else has to take those shots and if the argument is we could end up with a better shot opportunity than isolation for Melo you’re right, but if the argument is Chandler, Jordan and Andersen are better at scoring the ball than Melo because their TS% is insanely high I can’t agree with that.

    64. lavor postell

      I’d love to see the WOW people’s take on @26 (nicos’s post) from the previous thread.

      The fact that nobody responded to it should give you an answer.

      Here is the quote:

      Okay I looked up players who had a TS% above 57%, a usage of over 30%, and played over 50 games (lots of Jordan and Karl Malone)- it’s happened 50 times and all but 2 of those players teams were above .500 and made the playoffs (and one of the teams that didn’t was the 84 Knicks when BK only played 55 games). Then I checked teams with guys who had an offensive rebound% over 15% and played over 25 minutes a night for more than 50 games. It happened 37 times (lots of Moses and Rodman) with 10 of those teams finishing below .500, nine of those missing the playoffs. Less than a third of those teams won 50 or more games (just 10) while about 70% of teams with a high volume efficient scorer did. If offensive rebounding is so much more important than volume scoring shouldn’t those number be reversed? I mean what about all of those extra possessions??

    65. Eyal

      I was a sports writer in the 90’s and when WoW came out with the old blog (early 2000’s?) it was great to have someone talk about the value of efficiency and rebounding and turnovers and not just point points points all the time.

      And I LOVED Berri’s audacity. But it may be time to start looking at WoW in a historical context: great for starting the conversation, and still valuable for what it does, but far from being the unifying theory of physics.

    66. Alecto

      That’s the weakness of purely “descriptive” models–you can “describe” anything in any way you’d like retrospectively. Geocentric astronomers, for instance, found perfectly coherent ways to describe the kinds of phenomenon you mention within their paradigm. We find these explanations far-fetched and convoluted today, but there’s nothing inherently wrong with them so long as you offer them as mere descriptive explanations.

      Any scientific model that is not predictive has serious difficulties being falsifiable, which means that it verges on pseudoscience.

      You may have to tell me the geocentrists explanations for the pock-marks on the moon and Venus being in all phases because I’m pretty sure there was no theoretically consistent way to describe them in the geocentric model, as the moon was supposed to be made of aether and perfectly spherical and Venus was located in such relation to the sun that there were only two phases it could enter. What the Ptolemaics did wasn’t come up with explanations for these things to fit their paradigm–they abandoned these two theses entirely.

      There’s nothing inherently wrong with each descriptive model except that one is more correct than the other even when the balance of predictive power is neutral because one is a more accurate reflection of reality. So it may or may not go for WP vis a vis other metrics. I wasn’t implying that WP focused exclusively on description, just that the predictions arise from the descriptive qualities of the model qua explanans rather than being engineered specifically for the purpose of prediction (a model with a focus on prediction over description may look very different–it’s contingent upon your goals in bball analysis)–the predictions may be flawed, but that doesn’t thereby necessitate that the underlying model must be incorrect/not useful, especially when most other stats offer less explanation with equally bad.

    67. Alecto

      Perhaps WP is flat-out wrong like the geocentric model–but it also could be on the right track in explaining player production and just needs more work and theory behind it to really get the kinks out. What this requires is a willingness to take WP48 claims seriously and compare them to other statistical methods and maybe, (depending on your views of it), common sense claims. It also requires WP48 adherents to further refine their stats and variables as well as the explanations justifying their inclusion.

    68. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      I’m not sure that Kuhn (which I have myself read) is appropriate here. The question isn’t whether linear regression or modeling itself can lead to reliable information; it’s whether the data points used are accurate or precise in any reasonable explanation. In that sense, Berri is very much operating within the current scientific paradigm, in which we can reasonably posit that a valid conclusion can be gotten from regression techniques, given the assumption that the datapoints are valid.

      I am willing to argue that a low-volume, high-efficiency scorer is highly productive on a team because of a common sense assumption that his not-taken shot opportunities are not necessarily worse shot opportunities by a teammate. Common sense claims to the contrary say that because Chandler does not take a shot, it results in a poor shot attempt by a teammate (which the strange “end-of-shotclock” theory seems to argue is a commonplace thing).

      I have never said that wins produced is perfect. There is no way that the box score can “tell the story” of a basketball game (or team, or season) aside from the crude “how many shots did they take” information that can be had.

      Again, it’s an assumption that I draw from WP and WS, which each say that ORB% is a good thing to have a lot of, that the players who achieve high offensive rebounding rates are exceptionally valuable to their team. Just like players like LeBron and Durant, who score the ball with astounding efficiency while using a large number of shot attempts relative to their peers. Carmelo Anthony does neither of these things, so I believe (note the word choice) that his value is accurately assessed by WP48. (Plus, the Knicks sucked last year, so–.)

    69. johnno

      It’s a real shame that Chandler and Jordan have spent their entire careers playing with crappy point guards and for crappy coaches because the unsophisticated basketball fan will never realize what great offensive players they are. If Chandler ever played with a point guard like Kidd and for a smart coach like Carlisle, they’d run a high pnr every time down the court and he’d score like 40 points a game. And could you imagine what an unstoppable force Jordan would be if he ever got to play with someone like Chris Paul or for a coach like Doc Rivers? He might break Wilt’s 100! Instead they play with and for idiots and average about 10ppg and idiot basketball fans who think that great scorers should actually score sometimes don’t realize their greatness. What a disgrace…

    70. ruruland

      This is how these arguments almost always end, with Jowles disappearing when it’s brought up that high efficiency/low usage player can’t increase their usage w/out decreasing efficiency.

      Numerous times the last four years I’ve pointed out to Jowles (with fga by distance data) that the biggest reason Chandler, Andersen and Jordan have higher overall efficiency than the average big isn’t that they finish at abnormally high rates at the rim (there are many bigs with lower overall efficiency who are more efficient at the basket than Chandler)…

      No, what makes those three players exceptional is that they only shoot high efficiency shots.

      Because we can’t yet determine the value of skills and choices, there’s no way to know how much those three players are hurting their respective teams because of their lack of skill or because they choose not to take optimal shots in certain situations.

      Everyone not in WP denial acknowledges that even the worst NBA defense are not going to consistently allow layups to one-dimensional scorers throughout a game.

      But it seems we’re still a long way away from figuring out methods of quantifying the finite limit of choices and opportunities present withing each possession.

    71. DRed

      The fact that nobody responded to it should give you an answer

      I did respond to that one, I think. The idea that having high usage high efficiency players helps your team win lots of games is about as uncontroversial a statement as you can make. Except for maybe a Rodman season or two, those guys are generally going to be more valuable than a great rebounder.

    72. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      And that’s only because Rodman was so much better than his peers that it was like having an extra rebounder on the floor when he was on it. One year he had something like 530 ORB and had 200 more than the #3 player. The idea that he was anything but exceptionally productive is foolish to me.

    73. DRed

      By watching basketball, though, we know that’s not how rebounds work–the stop is a collaboration between whoever shuts down the shooter and whoever collected the rebound

      If Melo draws a double team and kicks out the ball to a wide open Jose Calderon, who bricks a 3 that is rebounded by Kevin Love, who should get the credit for that stop?

      Also, if rebounds were so defense dependent, wouldn’t we expect to see significant fluctuations in year to year rebounding based on team defense? To pick a player and a team completely at random, Tyson Chandler was a member of the 2011-2012 Knicks, who had one of the better defenses in the NBA. Last year, on an atrociously bad defensive team, and while exerting no effort whatsoever, he rebounded pretty much just as well as he did in 2011-2012. (actually he was a bit better on the glass last year, probably because he was just loafing around the rim) So who gets the credit for the extra rebounds Tyson got last year?

    74. Eyal

      The idea that having high usage high efficiency players helps your team win lots of games is about as uncontroversial a statement as you can make.

      DRed, the point is that it seems harder for these guys to get into the PW48 .300’s than it is for the rebounder / dunker type.

      None of this says that Melo is god and WOW sucks, as Ruru may suggest. Just for fun I imagine last year’s Knicks getting Ben Wallace instead of Bargs. The Pistons 2 year run to the finals Ben Wallace. I think this team, with a better coach, has a shot at a championship — Wallace was that dominant. So if that happens who gets finals MVP? Melo, of course, just like Billups won in 2004. Did he deserve it? naaaa… Big Ben did, but he was not the big time scorer so he didn’t.

    75. yoda4554

      @Jowles–Your conflation of WP and WS is telling. WS ranks Jordan as the 10th-best player in the league in ’14–but only the 2nd-best center and 3rd-best Clipper. It also ranks Nowitzki and Anthony as top-15 players. These are clear deviations from WP, and yet WS is also calculated by a formula using the same core data and partly determined by regression. Part of the reason they are different is because–for reasons I mentioned–WP misinterprets its data in ways that invalidate it.

      @DRed–On your first point: that’s like asking whether, if Bradley Beal dribbles through the Knick defense for an uncontested layup, he should really be credited with points. Basketball is a zero-sum game, and since we always give someone on offense credit when a team scores, even if they didn’t do much to deserve it, we might as well do so for the defense, too.

      On your second: if you get a stop, one of two things happened: a) you forced a turnover, or b) the opponents shot and missed, and your team recovered the ball (usually through a rebound). If teams have a high FG%, there will be fewer rebounds available, but that’s not the only way a defense can fail. In the case of the ’12 vs. ’14 Knick defenses, the major differences did not involve more missed shots. For instance, the ’12 team’s defense was based on forcing a ton of turnovers; the ’14 team did not do that. The ’14 team was atrocious at defending threes, while the ’12 team did that reasonably well. The ’14 team also fouled much more than the ’12 team. All these worsen a team’s defense without reducing its rebound total. That still means, though, that on the thirty instances a game in which the Knicks got a defensive rebound, some player or players on the team (usually other than the rebounder) needed to help coax missed shots from the other team. There is no stat by which they are credited for that, which is why WP gives (almost) all the credit to the rebounder.

    76. iserp

      I did respond to that one, I think. The idea that having high usage high efficiency players helps your team win lots of games is about as uncontroversial a statement as you can make.

      No, you are answering a totally different question.

      Jowles and you usually say that Chandler is exceptionally efficient, that there are not many roll man as efficient as Chandler. Ruru says that Chandler’s numbers at the P&R are not unusual, but since he only takes shots in that situation, his efficiency numbers don’t go down. So he requires an explanation about why a player that does exactly the same things and Chandler and a few others, is a worse player than Chandler.

      You could argue that doing less is more; but someone has to take shots, and Chandler’s (and Jordan’s…) usage is not going to cut it. And not for lack of trying.

      P&R may be better than anything else. But sometimes you start the play with Chandler setting a screen, Felton using it, being well defended and starting all over again with less time on the clock. It would be nice if someone tracked how many times a P&R attempt finishes with a shot attempt (and to add more detail, when it is a shot attempt by the roll man or by the ballhandler). I am just guessing now, but what number could it be, less than 30%?

    77. DRed

      It would be great if we had a stat called “stops,” in which players were assigned credit for causing shooters to miss, in which case advanced stats could split defensive credit between them and the rebounders. We don’t, though, and WP just gives full credit to the rebounder. Thus, WP overrates all players whose rebounding significantly supersedes their defense (e.g., DeAndre Jordan and Kevin Love).

      Basketball is a zero-sum game, and since we always give someone on offense credit when a team scores, even if they didn’t do much to deserve it, we might as well do so for the defense, too.

      Well, you said it. . .

    78. DRed

      No, you are answering a totally different question.

      Here was nicos question: If offensive rebounding is so much more important than volume scoring shouldn’t those number be reversed? I mean what about all of those extra possessions??

      Here’s what he looked at: I looked up players who had a TS% above 57%, a usage of over 30%, and played over 50 games and Then I checked teams with guys who had an offensive rebound% over 15% and played over 25 minutes a night for more than 50 games.

      Nicos found that the teams with the scoring guys on them did better than the teams with the rebounding guys. Setting aside the sample size problems, and that the test ignores all the other players stats, and that nicos is looking at team results to decide individual player value, nicos wasn’t looking at the right kind of scorer. A guy derided for being a ‘volume scorer’ is one that only scores a lot of points because he takes a lot of shots, because his efficiency is around or below average. Think of Allen Iverson leading the league in USG with a sub .500 TS%. Wins Produced says high efficiency scoring is the single best attribute for a player to have. So pointing out that teams with high usage, high efficiency scorers do really well doesn’t show that the model is flawed.

    79. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      It would be great if we had a stat called “stops,” in which players were assigned credit for causing shooters to miss, in which case advanced stats could split defensive credit between them and the rebounders. We don’t, though, and WP just gives full credit to the rebounder. Thus, WP overrates all players whose rebounding significantly supersedes their defense (e.g., DeAndre Jordan and Kevin Love).

      This is an unhelpful “solution” to the WP’s problem of “overrated” rebounders. “Stops” are totally arbitrary (as opposed to blocks, which have their own limited-information problem) and have no ability to describe the wide spectrum of difference between the player’s ability to make a shot unguarded and to make a shot with a defensive player basically hacking and not being called for the foul, particularly since that spectrum is different for each individual player in each individual place in the floor. I do think that xPPS would be a great statistic to incorporate into any comprehensive statistical model, but since WS and WP already account for opposing team eFG%, it seems that there’s already an approximation built in.

      Defensive Win Shares are no better when it comes to allocating team defense to individual players, too. The developer of Win Shares himself says it’s a major flaw of the stat.

    80. DRed

      Win Shares is a stupid stat. It says DeAndre Jordan is a top 10 player and all he can do is rebound, block shots, and dunk.

    81. Cole Aldrich's Second Cousin's Best Friend's Boyfriend

      And if we’re going to cry about these stats not penalizing players (I still don’t understand why you’d want to do that, or how you’d be able to, particularly for description) for not taking shots at all (e.g. Chandler) we must also realize that these stats (all of them, even PER at a certain efficiency threshold) penalize players like Josh Smith for taking so many bad ones.

      LeBron shoots well even though Chris Andersen can’t do anything except for putbacks and dunks. Carmelo (until last year, and even then, not nearly as well) really didn’t. So we should penalize Carmelo, not the guy who’s “forcing” him to take bad shots. Let’s penalize him for not having the skillset to take more short-range shots instead of settling for the mid-range shots that pepper his highlight reels (and the FG missed totals on b-r.com).

      Each side of this debate works perfectly well given their own sets of assumptions. Unfortunately for ruruland & co., the assumptions made by the WP side have held up during the last three years of Knicks basketball.

    82. nicos

      Nicos found that the teams with the scoring guys on them did better than the teams with the rebounding guys. Setting aside the sample size problems, and that the test ignores all the other players stats, and that nicos is looking at team results to decide individual player value, nicos wasn’t looking at the right kind of scorer. A guy derided for being a ‘volume scorer’ is one that only scores a lot of points because he takes a lot of shots, because his efficiency is around or below average. Think of Allen Iverson leading the league in USG with a sub .500 TS%. Wins Produced says high efficiency scoring is the single best attribute for a player to have. So pointing out that teams with high usage, high efficiency scorers do really well doesn’t show that the model is flawed.

      Wait are you suggesting that the thing I spent 10 minutes on at 1 in the morning wouldn’t get through peer review?? I will say WP certainly seems to prefer freak rebounders over high volume efficient scorers- Rodman rates out solidly better than Jordan, Larry Smith who showed up multiple times on the rebounding list (the 80s version of Reggie Evans for those who don’t remember him) rated out better than Karl Malone on a per minute basis. Reggie Evans has been a much more productive player over his career per minute than Dirk etc… Anyway, just for fun I redid the experiment this time using guys whose TS% was between 56-58 rather than just 57+ because I think most WP guys would say they would much rather have a great offensive rebounder than you know, Carmelo Anthony (who shows up four times on the new list). And you still get a greater percentage of 50 win teams and far more playoff teams than you do with the list of creme de la creme offensive rebounders. Only 4 teams (out of 47) with a guy with a TS% between 56-58, a usage of over 30, and playing at least 50 games have missed the playoffs- can you guess the last time it happened?

    83. nicos

      Oops, I meant out of 42 teams not 47. And of course the same caveats (small sample size, completely ignoring the rest of the teams rosters, etc…) still apply but it is heartening to see that Phil Jackson seemed to have pretty good success running the triangle through high-volume efficient (but not elite level efficiency) scorers- Kobe and Jordan both show up 4 times each.

    84. lavor postell

      Each side of this debate works perfectly well given their own sets of assumptions. Unfortunately for ruruland & co., the assumptions made by the WP side have held up during the last three years of Knicks basketball.

      So since WP predicted the Knicks correctly the last 3 years it’s argument has held up? Sure buddy.

      Here are some awesome predictions from WoW last year:
      Detroit: Predicted – 46.4 wins Actual – 29 wins
      Utah: Predicted – 42.5 wins Actual – 25 wins
      Indiana: Predicted – 39.9 wins Actual – 56 wins
      Boston: Predicted – 39.9 wins Actual – 25 wins
      Phoenix: Predicted – 27.1 wins Actual – 48 wins
      Portland: Predicted – 28.6 wins Actual – 54 wins
      Charlotte: Predicted – 29.3 wins Actual – 43 wins
      Golden State: Predicted – 38.3 wins Actual – 51 wins
      Toronto: Predicted – 37.8 wins Actual – 48 wins
      Philadelphia: Predicted – 35.1 wins Actual – 19 wins
      Cleveland: Predicted – 42 wins Actual – 33 wins

      It’s awesome that by some miraculous set of circumstances WP has nailed the Knicks record the past 3 years, but it certainly isn’t because their formula is prescient or far ahead of any other methods being used.

    85. Mike Kurylo Post author

      Using my own formula (PPG+RPG+APG)*(Mpg/48) I get a list that makes so much more sense:

      The problem with this, is that it’s primarily ranking players by minutes played. So it will miss all the great players that are being underutilized. For example take this player:

      PRA
      2005: 3.6
      2008 12.7

      PER
      2005: 15.4
      2008: 19.0

      By PRA standards, he made a huge leap in skill/production in those three years. But using PER, something that accounts for per minute variations, the player was already very useful in 2005, just his coach didn’t give him the playing time.

      So unless you want to toss aside the David Lee’s of the world as worthless, this stat is obviously worthless.

    Comments are closed.