Statistical Analysis. Humor. Knicks.

Monday, July 28, 2014

Knicks Morning News (Friday, Aug 30 2013)

  • [New York Times] Sports Briefing | Pro Basketball: Suns Trade Butler to the Bucks (Fri, 30 Aug 2013 04:56:36 GMT)
    The Phoenix Suns traded small forward Caron Butler to the Milwaukee Bucks for point guard Ish Smith and center Viacheslav Kravstov.    

  • 76 comments on “Knicks Morning News (Friday, Aug 30 2013)

    1. Z-man

      So now that Ivan Johnson is gone, seems more likely that either Haddadi, Barron or Aldrich are the top candidates. I’m thinking that GG has been traumatized by Hibbert’s domination and is looking for a possible anti-Hibbert. Haddadi might be the best match for that role. Barron plays smaller than 7 feet. Aldrich is a foul machine.

    2. SeeWhyDee77

      I have no clue what Hammond is doing out there. Don’t get me wrong, Butler’s a good player and one I would love to have on the Knicks..but I just don’t understand the team he’s putting together. 1st he let Ellis walk. Then at Drew’s behest he swung and missed on Teague. Then end end up moving Jennings for Knight. He also gave Sanders 44 mil too early. To me, Brandon Knight’s an undersized 2. With the collection of guys on that roster, u need a guy capable of running the show. At least Jennings could do that. Say what u will about GG’s latest moves, but at least they make sense whether we agree with them or not. I think they should’ve moved Jennings for a big who can shoot and installed Redick and Ellis as the starting backcourt. And then Butler would have made more sense. And why is Udoh still there when they have Henson and Sanders? Hell, goin after Asik would have made a lot more sense

    3. SeeWhyDee77

      Z-man: So now that Ivan Johnson is gone, seems more likely that either Haddadi, Barron or Aldrich are the top candidates. I’m thinking that GG has been traumatized by Hibbert’s domination and is looking for a possible anti-Hibbert. Haddadi might be the best match for that role. Barron plays smaller than 7 feet. Aldrich is a foul machine.

      Yea I think I would lean towards Haddadi at this point too

    4. Nick C.

      I looked at Haddadi and he rebounds and blocks shots better than Aldrich but has an atrocious TS of .504 and fouls/36 of 7.1 (Aldrich is 5.7). Fortunately the role we envision is 5-10 mpg to keep Tyson and Martin from being ground into paste by March. At any rate having seen pictures of Aldrich and Haddadi I doubt Woodson gives either of them any playing time unless he has no choice so this may be a moot point.

    5. thenamestsam

      SeeWhyDee77:
      I have no clue what Hammond is doing out there. Don’t get me wrong, Butler’s a good player and one I would love to have on the Knicks..but I just don’t understand the team he’s putting together. 1st he let Ellis walk. Then at Drew’sbehest he swung and missed on Teague. Then end end up moving Jennings for Knight. He also gave Sanders 44 mil too early. To me, Brandon Knight’s an undersized 2. With the collection of guys on that roster, u need a guy capable of running the show. At least Jennings could do that. Say what u will about GG’s latest moves, but at least they make sense whether we agree with them or not. I think they should’ve moved Jennings for a big who can shoot and installed Redick and Ellis as the starting backcourt. And then Butler would have made more sense. And why is Udoh still there when they have Henson and Sanders? Hell, goin after Asik would have made a lot more sense

      I think it’s a really hard situation for a GM because honestly that team is a total teardown but supposedly the owner has no interest in rebuilding. You can quibble with some of the moves but the far bigger deal in my mind is that the larger strategy just makes no sense. They should be tanking and looking to rebuild but instead they’re doing everything possible to get to that 8th spot so they can be slaughtered by Miami again. He could be doing a better job maybe but he doesn’t have a ton of good options. His owner wants him to try to get better even at the expense of the future of the team but surely he knows that makes no sense in the current situation.

    6. thenamestsam

      Oh also I saw someone tweet the most up to date WoW projections into my timeline and I haven’t seen them discussed here so I figured I’d throw them up. These are the Eastern Conference projections as of 4 days ago and are presented here without comment. I think they speak for themselves. Numbers are number of wins.

      1. Miami 54.9
      2. Detroit 53.5
      3. Chicago 49.7
      4. Atlanta 48.1
      5. Cleveland 45.8
      6. Philadelphia 43.0
      7. Toronto 41.2
      8. Boston 38.8
      9. Indiana 38.2
      10. Brooklyn 35.7
      11. Washington 34.2
      12. New York 33.6
      13. Milwaukee 33.0
      14. Orlando 26.7
      15. Charlotte 26.5

    7. Garson

      thenamestsam:
      Oh also I saw someone tweet the most up to date WoW projections into my timeline and I haven’t seen them discussed here so I figured I’d throw them up. These are the Eastern Conference projections as of 4 days ago and are presented here without comment. I think they speak for themselves. Numbers are number of wins.

      1. Miami 54.9
      2. Detroit 53.5
      3. Chicago 49.7
      4. Atlanta 48.1
      5. Cleveland 45.8
      6. Philadelphia 43.0
      7. Toronto 41.2
      8. Boston 38.8
      9. Indiana 38.2
      10. Brooklyn 35.7
      11. Washington 34.2
      12. New York 33.6
      13. Milwaukee 33.0
      14. Orlando 26.7
      15. Charlotte 26.5

      Wiggins!

    8. The Honorable Cock Jowles

      Garson: Wiggins!

      2014 first round draft pick to Denver
      New York’s 2014 1st round pick to Denver [Denver-Minnesota-New York, 2/22/2011]; Denver may convey this pick to Orlando (see Denver Debits)

    9. Garson

      The Honorable Cock Jowles: 2014 first round draft pick to Denver
      New York’s 2014 1st round pick to Denver [Denver-Minnesota-New York, 2/22/2011]; Denver may convey this pick to Orlando (see Denver Debits)

      For some reason i remmeber it being top 3 protected… I stand corrected.

    10. The Honorable Cock Jowles

      Garson: For some reason i remmeber it being top 3 protected… I stand corrected.

      If we lose Wiggins for Carmelo Anthony… imaging the swift and sharp apostasy from His Underrated Majesty.

    11. flossy

      Garson: Wiggins!

      Gettin’ Wiggy wit it! Nah nah nah, nah nah nah…

      But seriously though, those win projections are retarded. Our team bus could fall into the East River and we would still win more than 34 games. The fact that both Brooklyn and NYK are projected to miss the playoffs but the 76ers, who more or less did drive their team’s bus into a ravine in the name of a high draft pick, will get the 6th seed… tells you just about all you need to know.

    12. thenamestsam

      flossy: Gettin’ Wiggy wit it!Nah nah nah, nah nah nah…

      But seriously though, those win projections are retarded.Our team bus could fall into the East River and we would still win more than 34 games.The fact that both Brooklyn and NYK are projected to miss the playoffs but the 76ers, who more or less did drive their team’s bus into a ravine in the name of a high draft pick, will get the 6th seed… tells you just about all you need to know.

      Don’t forget Indy is on the outside looking in as well. I wonder what kind of Vegas odds you could get on an Atlantic division order of finish of Philly, Toronto, Boston, Brooklyn, New York?

      That’s how you know even those guys don’t take their projections seriously. If they did they’d all be millionaires and their formula would be a preciously guarded secret not bloggers scraping by and publishing it for free.

    13. ruruland

      The Honorable Cock Jowles: If we lose Wiggins for Carmelo Anthony… imaging the swift and sharp apostasy from His Underrated Majesty.

      The Knicks are going to lose 21 MORE GAMES NEXT YEAR!!!!!! because they lost…….Jason Kidd and Ronnie Brewer?????

      So Jowles, how do you explain the 12 win disparity between your “projection” and WOW’s?

      What knowledge and insight are they (or the formula) missing (which just sound impossible) that you have?

      Are there any other EC WOW projections you disagree with?

      Name them and explain, please.

      Why do the Pacers lose 11 more games next season?

      How does Detroit ADD 24 WINS!!! in a single season given what they added in the off-season?

      Sixers + 9 wins?

      Brooklyn – 14 wins?

      If something close to this happens it would amount to one of the biggest one-year conference shake-ups of all-time.

    14. The Honorable Cock Jowles

      flossy: But seriously though, those win projections are retarded. Our team bus could fall into the East River and we would still win more than 34 games. The fact that both Brooklyn and NYK are projected to miss the playoffs but the 76ers, who more or less did drive their team’s bus into a ravine in the name of a high draft pick, will get the 6th seed… tells you just about all you need to know.

      What happens if 34-year-old Metta World Peace’s value is more anecdotal than actual, and if black-hole Andrea Bargnani sees the 20 mpg that he’s probably going to get? And J.R. Smith keeps his moonlight work as an exotic dancer talent scout? And Jason Kidd’s hot start isn’t replicated by any of our SG? You don’t think it’s possible that the Knicks could crash and burn and that Carmelo could be asking for a trade by February?

      God forbid that Chandler goes down and we have to explain how Chandler could possibly be the diffference between a 25-win team and a 45-win team.

    15. nckev

      ruruland: The Knicks are going to lose 21 MORE GAMES NEXT YEAR!!!!!! because they lost…….Jason Kidd and Ronnie Brewer?????

      So Jowles, how do you explain the 12 win disparity between your “projection” and WOW’s?

      What knowledge and insight are they (or the formula) missing (which just sound impossible) that you have?

      Are there any other EC WOW projections you disagree with?

      Name them and explain, please.

      Why do the Pacers lose 11 more games next season?

      How does Detroit ADD 24 WINS!!! in a single season given what they added in the off-season?

      Sixers + 9 wins?

      Brooklyn – 14 wins?

      If something close to this happens it would amount to one of the biggest one-year conference shake-ups of all-time.

      Here’s the actual article: link
      The prediction of wins is completely dependent on minute allocation of a team’s players. The reason the predictions are so low for the Knicks is that they have Bargs getting 23.6 MPG, JR at 11.9 and Shump at 19.0. JR and Shump will get more burn than that, and I really really really hope Bargs doesn’t see the floor that much. Before you criticize the method, take a look at it and try to determine why it’s off rather than casually dismissing it.

    16. thenamestsam

      The Honorable Cock Jowles: What happens if 34-year-old Metta World Peace’s value is more anecdotal than actual, and if black-hole Andrea Bargnani sees the 20 mpg that he’s probably going to get? And J.R. Smith keeps his moonlight work as an exotic dancer talent scout? And Jason Kidd’s hot start isn’t replicated by any of our SG? You don’t think it’s possible that the Knicks could crash and burn and that Carmelo could be asking for a trade by February?

      God forbid that Chandler goes down and we have to explain how Chandler could possibly be the diffference between a 25-win team and a 45-win team.

      Things could go horribly, horribly wrong. That’s true for any team in the league. But that projection doesn’t say the Knicks could win 34 games in a worst case scenario. It says that’s the expected number of wins meaning (if I’m understanding the methodology for generating the projection correctly) that it’s equally likely that the Knicks win more or less than 34 games.

      Obviously you agree that that’s way off since you (much more reasonably) projected 46 wins. 34 wins is a truly horrible projection. It’s going to be a minimum of 10-15 games off the Vegas over/under (haven’t seen numbers for the coming season although they should be out soon) which means it’s either generating insane value in terms of betting or it’s horribly flawed. Which do you think it is?

    17. The Honorable Cock Jowles

      See, I don’t think it’s way off. I think they’re going to give Smith way more burn than that, and hopefully Bargnani will retire before the first game and we won’t have to talk about diminishing returns and stretch-fours and how a PF could possibly be valuable if he rebounds like a point-guard.

      But if they do break .500, it will be a huge success. I think Metta World Peace is, at this point, awful — horrid shooting that necessitates stellar defense to compensate — and these additions are going to look a lot more like subtraction by midseason.

    18. The Honorable Cock Jowles

      If Vegas sets the under/over at 48 wins, I would wager a large sum of money on the under. All I’m saying.

    19. ruruland

      nckev: Here’s the actual article: link
      The prediction of wins is completely dependent on minute allocation of a team’s players. The reason the predictions are so low for the Knicks is that they have Bargs getting 23.6 MPG,JR at 11.9 and Shump at 19.0. JR and Shump will get more burn than that, and I really really really hope Bargs doesn’t see the floor that much. Before you criticize the method, take a look at it and try to determine why it’s off rather than casually dismissing it.

      LOL, yes, because those things amount to 21 more losses.

      Let this be the thread that ends the WoW discussion once for all. I know I’m going to be using it all season long if others persist in arguing for the merits of its methodology.

    20. nckev

      The Honorable Cock Jowles:
      See, I don’t think it’s way off. I think they’re going to give Smith way more burn than that, and hopefully Bargnani will retire before the first game and we won’t have to talk about diminishing returns and stretch-fours and how a PF could possibly be valuable if he rebounds like a point-guard.

      But if they do break .500, it will be a huge success. I think Metta World Peace is, at this point, awful — horrid shooting that necessitates stellar defense to compensate — and these additions are going to look a lot more like subtraction by midseason.

      Add Tyson being a year older and coming off 2 seasons where he had career high minutes, Prigs being old as dirt, and Amare being Amare, it could get real ugly. But hey, Amare could be stay healthy, Tyson could play 30+MPG and stay healthy, Bargs could never once play PF/C, Shump could very well improve in his third year and the team could be really good. It makes the season really hard to predict.

    21. nckev

      ruruland: LOL, yes, because those things amount to 21 more losses.

      Let this be the thread that ends the WoW discussion once for all. I know I’m going to be using it all season long if others persist in arguing for the merits of its methodology.

      I don’t understand the issues people have with it’s methodology. You’re correlating captured data to wins. How is that controversial? It’s what you do in statistics. Yes, obviously it can’t capture all variables related to wins, but it uses what it has access to. That’s what you do in statistical analysis. You use available data to build models.

    22. Brian Cronin

      See, I don’t think it’s way off. I think they’re going to give Smith way more burn than that, and hopefully Bargnani will retire before the first game and we won’t have to talk about diminishing returns and stretch-fours and how a PF could possibly be valuable if he rebounds like a point-guard.

      But if they do break .500, it will be a huge success. I think Metta World Peace is, at this point, awful — horrid shooting that necessitates stellar defense to compensate — and these additions are going to look a lot more like subtraction by midseason.

      If Vegas sets the under/over at 48 wins, I would wager a large sum of money on the under. All I’m saying.

      But again, you’re predicting 46 wins, so why would breaking .500 be a huge success if that’s what you think that they will do? Similarly, if you think they’ll win 46, would you really wager a large sum of money on under 48 when your prediction is so close to 48?

    23. thenamestsam

      Brian Cronin:
      But again, you’re predicting 46 wins, so why would breaking .500 be a huge success if that’s what you think that they will do? Similarly, if you think they’ll win 46, would you really wager a large sum of money on under 48 when your prediction is so close to 48?

      Exactly. THCJ, I thought your prediction was 46? Now you’re ready to wager a large sum on under 48 and think .500 would be a huge success? Those two positions are completely contradictory. I assume that means you’re no longer predicting 46 wins. Is it the release of these numbers that has changed your mind? If not what caused the reversal? (Those are honest questions in case that’s unclear)

    24. ruruland

      Robert Silverman:
      FWIW, Wages of Wins predicted the Knicks would win 56 games last season.

      http://wagesofwins.com/2012/10/31/nba-win-predictions-for-2012-13-volume-2-the-hand-crafted-edition/

      Well, if we’re going to pull out and measure, I predicted 54-58 on this board when the doomsayers were in full strut last summer, well before Jowles regurgitated WoW, focusing on what I figured would be dramatically reduced turnovers which would elevate the Knicks offense into the top 7.

    25. ruruland

      nckev: I don’t understand the issues people have with it’s methodology. You’re correlating captured data to wins. How is that controversial? It’s what you do in statistics. Yes, obviously it can’t capture all variables related to wins, but it uses what it has access to. That’s what you do in statistical analysis. You use available data to build models.

      We’ve been over this too many times I’m afraid.

    26. The Honorable Cock Jowles

      thenamestsam: Luckily we don’t have to judge the measure on one datapoint because we have tons of datapoints available which tell us that overall it does a very poor job of prediction.

      WP has no PECOTA-like system of variability. That, to me, makes it easy to rail on. No one can accurately predict injuries. We can try to figure out player similarities, but no one’s going to fault a statistician for Karl Malone’s injury in the ’04 playoffs, nor will he fault that statistician when Andre Drummond puts up all-time great numbers for a teenager and still rides the pine.

      I really don’t care if any of you agree or disagree with WP’s predictions, but dismissing them because they predict a huge come-down from last season is pretty silly.

    27. The Honorable Cock Jowles

      nckev: Add Tyson being a year older and coming off 2 seasons where he had career high minutes, Prigs being old as dirt, and Amare being Amare, it could get real ugly. But hey, Amare could be stay healthy, Tyson could play 30+MPG and stay healthy, Bargs could never once play PF/C, Shump could very well improve in his third year and the team could be really good. It makes the season really hard to predict.

      This is all true. But if Amar’e goes down, you’re going to see Andrea Bargnani as PF, possibly as center when Woodson plays how-fast-can-I-lose-my-job ball (commonly known as “small-ball”). Please, no one complain when the Knicks are fighting for a playoff spot like the Lakers did last season.

      I’ll eat crow if they break 50 wins, but that ain’t happenin’ with Bargnani seeing significant time, which he almost certainly will.

      (P.S. I predicted 45 wins, I believe.)

    28. iserp

      nckev: I don’t understand the issues people have with it’s methodology. You’re correlating captured data to wins. How is that controversial? It’s what you do in statistics. Yes, obviously it can’t capture all variables related to wins, but it uses what it has access to. That’s what you do in statistical analysis. You use available data to build models.

      Because it is difficult to use statistics rigorously. WoW care very little about their methodology, use axioms that are dubious at best, and don’t test their model in a scientific way.

      For me, the most glaring hole of WP48 is that when a player changes role, it varies significantly. And there are lots of examples in players that changed teams and roles and their WP48 raises or goes down. And I believe this is tied to the fact that WoW have the axiom that usage and efficiency are not correlated.

      I have yet to see proof that usage and efficiency are not correlated. WoW proponents usually just point a graph where %USG vs %TS is a blur of points, and they believe that it is a scientific proof!

    29. nckev

      iserp: Because it is difficult to use statistics rigorously. WoW care very little about their methodology, use axioms that are dubious at best, and don’t test their model in a scientific way.

      For me, the most glaring hole of WP48 is that when a player changes role, it varies significantly. And there are lots of examples in players that changed teams and roles and their WP48 raises or goes down. And I believe this is tied to the fact that WoW have the axiom that usage and efficiency are not correlated.

      I have yet to see proof that usage and efficiency are not correlated. WoW proponents usuallyjust point a graph where %USG vs %TS is a blur of points, and they believe that it is a scientific proof!

      But that example you give is an axiom supported by data. There is proof that usage and TS% doesn’t show correlation. Here’s an example.
      When you plot changes in FGA against changes in TS% you get a big blob of random points, like you said. Since you can’t fit a curve to those points with any kind of accuracy, you can say they are not correlated. Based on that link I provided, you can say that from 2002-2012 changes in TS% has shown no correlation with changes in FGA. That’s what the data says. That’s proof. I don’t understand your argument.

    30. Unreason

      One point of formal theory is to make predictions that are as precise as you need them to be for some practical purpose. E.g. given the equations of Newtonian mechanics, if you know the magnitude and direction of the forces acting on a bullet you can make a very precise prediction of it’s final velocity and position after a certain amount of time.

      Predictions based on correlations among variables that are not related to one another in a formal theory can be very imprecise and the “models” that produce them can mislead rather than clarify causal relationships. But predictions based on atheoretical correlations can be useful if they help you predict something to an extent that has practical value. They can also be useful exploratory steps in theory building.

      Discussions of predictions based on atheoretical models – e.g. WP – would be more productive IMO if they focused on more confidence intervals on point estimates (e.g. wins). Predictive models always give both a point estimate (one specific number) and a confidence interval (a range of numbers the actual thing with fall within 95% of the time). The size of the confidence interval as calculated from actual predictive success is a good indicator of the practical value of the prediction. E.g. if the CI is huge – 20 games, you know it’s not helping you much. There’s no need to argue about its value. If it’s 95% accurate +/- 1.5 games, you know it’s a pretty useful predictor of success, even if it doesn’t clarify why teams succeed.

    31. Z-man

      Robert Silverman:
      FWIW, Wages of Wins predicted the Knicks would win 56 games last season.

      http://wagesofwins.com/2012/10/31/nba-win-predictions-for-2012-13-volume-2-the-hand-crafted-edition/

      The Honorable Cock Jowles: WP has no PECOTA-like system of variability. That, to me, makes it easy to rail on. No one can accurately predict injuries. We can try to figure out player similarities, but no one’s going to fault a statistician for Karl Malone’s injury in the ’04 playoffs, nor will he fault that statistician when Andre Drummond puts up all-time great numbers for a teenager and still rides the pine.

      I really don’t care if any of you agree or disagree with WP’s predictions, but dismissing them because they predict a huge come-down from last season is pretty silly.

      Let’s keep in mind that WoW was factoring in the production of WP stars Camby and Brewer when predicting those wins. Pull them out of the equation and what would WP48 have predicted? Then factor in that Melo had his lowest WP48 in 9 years at a career high usage (interestingly, he had a career high WS48, go figure!) Steve Novak produced double the wins (4) that Anthony did!

      http://www.thenbageek.com/teams/nyk?season=2012

      Shocking that we were 47-20 with Melo and 7-8 without him, considering that every other player with over 600 minutes had a higher WP48 than him.

      Why not presume that the breaks even out? WP stars get get injured, yes, but so do WP losers.

      Hollinger uses PER (the anti-WP48) and the eye test to consistently outperform WP48 predictions. Doesn’t he have to deal with unexpected injuries and dumb…

    32. Z-man

      …coaching decisions?

      thenamestsam:
      Oh also I saw someone tweet the most up to date WoW projections into my timeline and I haven’t seen them discussed here so I figured I’d throw them up. These are the Eastern Conference projections as of 4 days ago and are presented here without comment. I think they speak for themselves. Numbers are number of wins.

      1. Miami 54.9
      2. Detroit 53.5
      3. Chicago 49.7
      4. Atlanta 48.1
      5. Cleveland 45.8
      6. Philadelphia 43.0
      7. Toronto 41.2
      8. Boston 38.8
      9. Indiana 38.2
      10. Brooklyn 35.7
      11. Washington 34.2
      12. New York 33.6
      13. Milwaukee 33.0
      14. Orlando 26.7
      15. Charlotte 26.5

      Any predictor or predicting system can come up with excuses in hindsight when things go awry, and I can’t wait to hear WoW’s post-mortem next summer. Seriously, predicting that the Knicks, Nets and Pacers will all miss the playoffs as they are beaten out by the Sixers, Celtics and Raptors? I would be tempted to bet that the typical 12yo NBA fan who wouldn’t know an advanced stat if it bit him in the ass could do a better job of predicting total wins for next season.

    33. KnickfaninNJ

      I looked carefully at the WoW tables in the link. In the bottom table it shows that their prediction for last year (yes) was 41 wins. I assume this is because the players actual win shares and minutes played combine to produce this number. So part of the reason WoW predicts a horrible record this year is that it concludes last year was a complete fluke. The rest of this years predictions from them is that the Knicks will be seven games worse than this and seems to be because our rookies are going to be horrible but get significant minutes anyway and because Carmelo and Andrea are going to get more minutes than they deserve. Bargnani is an interesting case. He played on a bad team, which means the WoW had to distribute negative wins somewhere. He got a lot of them, probably because he rebounds badly for his nominal position (which Carmelo does too) and WoW values rebounds very highly..

    34. BigBlueAL

      Why are we all just focusing on the Knick prediction?? Pacers, Nets and the Knicks all predicted to miss the playoffs?? Philly to win the Atlantic Division?? Detroit to finish 2nd in the East??

    35. iserp

      nckev: But that example you give is an axiom supported by data. There is proof that usage and TS% doesn’t show correlation. Here’s an example.
      When you plot changes in FGA against changes in TS% you get a big blob of random points, like you said. Since you can’t fit a curve to those points with any kind of accuracy, you can say they are not correlated. Based on that link I provided, you can say that from 2002-2012 changes in TS% has shown no correlation with changes in FGA. That’s what the data says. That’s proof. I don’t understand your argument.

      NO! That is not a proof!

      (claiming “that graph is a proof” is the “he scores a lot, he is good” of graphs and statistics)

      I am sorry, that is not a scientific proof. If you want to extract statistic evidence that there is no correlation between %TS and %USG you have to compare the graph of the actual data with your expectations.

      Don’t forget that “eye-test proponers” say that %TS and %USG are supposed to be correlated at the individual level.

      So, what are your expectations for that graph if %TS and %USG were correlated at the individual level?

      For the sake of the debate, let’s say that %TS are perfectly correlated at the individual level: correlation = 1. For example, for each player %TS = %BASE_TS – %USG/2. Each player has different skills, so %BASE_TS is different for Lebron James than for Jared Jeffries, let us say.

      I’ll tell you what i would expect, but i want you to think it first.

    36. Z-man

      iserp: Don’t forget that “eye-test proponers” say that %TS and %USG are supposed to be correlated at the individual level.

      I don’t necessarily agree. Eye-test proponents, or better put, WP skeptics, feel that in general, low-usage/high efficiency players will never be in a position to be high usage/high efficiency players. It is beyond obvious that players like Tyson Chandler will never be put in a position to average 20-25 shots a game, and that if he did, he would experience a significant drop in efficiency.

      However, some players who are destined to be high usage/moderate-high efficiency scorers might have lower usage numbers depending on point in career, coach, teammates, etc. and that player’s efficiency might even rise as usage goes up.

      So, in essence, the usage-efficiency correlation should be examined backwards. start with players who show a strong positive correlation and see if there are commonalities. Then look at players that had an inverse relationship and look for commonalities. Then look at players who experienced a major change in efficiency over significant segments of a career and see whether there is any correlation with usage.

      (Importantly, players whose usage or efficiency don’t change much tell you pretty much nothing about the relationship, yet they are the real basis for the “what if” part of the discussion.)

      Then use this information to make future predictions/assumptions about players based on whether they fit into one group or the other.

    37. Z-man

      I did a sort of players that had a 1000+minute season with a TS% greater than .600 and a USG% less than 18. Lots of players came up, so I sorted them by decreasing usage. In looking at individual careers, the following stood out early on.

      Ty Lawson: usg up, TS down (low-usage rookie year)
      David Lee: usg up, TS down (pre-post Knicks)
      Shawn Bradley: usg down, TS up
      Brent Barry: usg down, TS up (cut back on 2-pt attempts)
      Thabo Sefolosha: usg down, TS up
      Andrew Bynum: usg up, TS unchanged

      Its a long list so I don’t have time to go further right now, but most of the players I went through either stayed at low usage or had only an outlier hi-TS% season. Maybe there is a better way to sort.

    38. Z-man

      Tyson Chandler was higher usage and lower TS early in his career. It raises the question: was he missing more of the same shots then, or taking different shots?

    39. EB

      If I recall correctly, WoW predicted the Knick’s win total accurately, but that included 9 or 10 wins from Marcus Camby. So while the prediction was accurate the reasoning was unsound.

    40. EB

      nckev: But that example you give is an axiom supported by data. There is proof that usage and TS% doesn’t show correlation. Here’s an example.
      When you plot changes in FGA against changes in TS% you get a big blob of random points, like you said. Since you can’t fit a curve to those points with any kind of accuracy, you can say they are not correlated. Based on that link I provided, you can say that from 2002-2012 changes in TS% has shown no correlation with changes in FGA. That’s what the data says. That’s proof. I don’t understand your argument.

      Several flaws in this statement. As the link you provides shows ts% explains 34% of the next year’s ts% that means it is difficult to accurately measure the effect of increased usage because it is difficult to judge a baseline rate. Actual improvement can be masked by a hot year from 3. Second, those who are shooting better will get more playing time so an increase in ts% and fga/g may be due to the rationality of coaches. Third, people can construct 2 versions of the USG and ts% argument. There’s a strong claim that says all players suffer from higher USG and a weak claim that says certain players suffer from higher USG who depend on others to create their shots. So really it’s not that USG and efficiency are correlated so much as it approximates a limit on how many times a defense will let you get an open corner 3 or a wide open dunk off a pnr.Then others will say that to increase USG past a version point you need to run low efficiency Isos.

    41. KnickfaninNJ

      Since Camby got almost no minutes last year that would explain why WoW’s retrospective predictions are much worse than their original predictions. Maybe if we had played Camby more we could have won 64 games ;-)

    42. iserp

      Z-man: Eye-test proponents, or better put, WP skeptics, feel that in general, low-usage/high efficiency players will never be in a position to be high usage/high efficiency players. It is beyond obvious that players like Tyson Chandler will never be put in a position to average 20-25 shots a game, and that if he did, he would experience a significant drop in efficiency.
      However, some players who are destined to be high usage/moderate-high efficiency scorers might have lower usage numbers depending on point in career, coach, teammates, etc. and that player’s efficiency might even rise as usage goes up.

      I agree.

      Saying that “%USG and %TS are correlated level” is still simplifying too much. It is something that still is blind to interaction effects, role in the team, etc… However, i was trying to simplify everything a lot. Why? Because i wanted to show that even a very simple model where %TS vs %USG are straight lines -actually, parallel straight lines depending on the player-, the “%TS vs %USG” of all NBA players would show no correlation. And i am trying to do this to try to convince people that WoW methods are just pseudoscience.

      Showing a graph and claiming something is not science. You have to know what happens if your hypothesis is correct, what happens when your hypothesis is false and compare with reality. And if it happens that -whether your hypothesis is correct or not- the results are the same, then you have to design new tests to check your hypothesis. And WoW proponents doesn’t try to test their model, it is the “truth”.

    43. nckev

      iserp: I agree.

      Saying that “%USG and %TS are correlated level” is still simplifying too much. It is something that still is blind to interaction effects, role in the team, etc… However, i was trying to simplify everything a lot. Why? Because i wanted to show that even a very simple model where %TS vs %USG are straight lines -actually, parallel straight lines depending on the player-, the “%TS vs %USG” of all NBA players would show no correlation. And i am trying to do this to try to convince people that WoW methods are just pseudoscience.

      Showing a graph and claiming something is not science. You have to know what happens if your hypothesis is correct, what happens when your hypothesis is false and compare with reality. And if it happens that -whether your hypothesis is correct or not- the results are the same, then you have to design new tests to check your hypothesis. And WoW proponents doesn’t try to test their model, it is the “truth”.

      The relationship (or lack there of) between usage and scoring efficiency has nothing to do with WOW, they just put it in a graph. The data comes from box scores of real games. I don’t know what to tell you. That’s the data. In general, if you plot the change in FGA versus the change in TS% over the last 10 years, you see no correlation between those two variables. That’s not debatable. You can grab that data and plot it yourself. I really have no idea what you’re arguing about…

    44. nckev

      EB: As the link you provides shows ts% explains 34% of the next year’s ts% that means it is difficult to accurately measure the effect of increased usage because it is difficult to judge a baseline rate. Actual improvement can be masked by a hot year from 3. Second, those who are shooting better will get more playing time so an increase in ts% and fga/g may be due to the rationality of coaches.

      Yes, exactly! TS% obviously changes from year to year. What the graphs show is that, in general, you are unable to come to a conclusion on what is going to happen to a player’s efficiency based on only their change in usage. You would be wrong if you said that a low usage high efficiency player would have lower efficiency if he took more shots. You would also be wrong if you said that players efficiency would increase or stay the same with increased, decreased, or equal usage. In general there is no clear relationship between a change in FGA and a change in TS%. Sorry to keep repeating myself…

    45. nckev

      Z-man:
      I did a sort of players that had a 1000+minute season with a TS% greater than .600 and a USG% less than 18. Lots of players came up, so I sorted them by decreasing usage. In looking at individual careers, the following stood out early on.

      Ty Lawson: usg up, TS down (low-usage rookie year)
      David Lee: usg up, TS down (pre-post Knicks)
      Shawn Bradley: usg down, TS up
      Brent Barry: usg down, TS up (cut back on 2-pt attempts)
      Thabo Sefolosha: usg down, TS up
      Andrew Bynum: usg up, TS unchanged

      Its a long list so I don’t have time to go further right now, but most of the players I went through either stayed at low usage or had only an outlier hi-TS% season. Maybe there is a better way to sort.

      In that link I provided, they did look at just high efficiency players, and how there TS% changed with changing FGA (third graph). Again, no correlation. I do find this surprising, too. You would think there would at least be weak correlation, but in a general sense there is not. Now I’m not saying this shouldn’t be dug into more. It would be interesting to see it broken down further.

      I’m really interested to see what does correspond to increased efficiency. I’ve seen plots that suggest age is correlated to efficiency, in general, efficiency increases in a player’s first few seasons, levels out and then decreases once they pass their prime. There are a lot of other variables to look at, it’s just a matter of gathering all the data. This would make a good article.

    46. nckev

      Z-man: I don’t necessarily agree. Eye-test proponents, or better put, WP skeptics, feel that in general, low-usage/high efficiency players will never be in a position to be high usage/high efficiency players.It is beyond obvious that players like Tyson Chandler will never be put in a position to average 20-25 shots a game, and that if he did, he would experience a significant drop in efficiency.

      Keep this in mind and make sure you check this site on Tuesday. I have to go now folks, but I really do enjoy this conversation. I’ll be back around on Tuesday. Thanks!

    47. iserp

      nckev: The relationship (or lack there of) between usage and scoring efficiency has nothing to do with WOW, they just put it in a graph. The data comes from box scores of real games. I don’t know what to tell you. That’s the data. In general, if you plot the change in FGA versus the change in TS% over the last 10 years, you see no correlation between those two variables. That’s not debatable. You can grab that data and plot it yourself. I really have no idea what you’re arguing about…

      I’ll try to explain myself a bit better. When people say that %USG and %TS are correlated, it means that if a player with 20% USG and 56%TS, gets a higher usage, then he will get a lower %TS. Let us say that if during the same season, he had a 22% USG, then he would have 55 %TS. I think until here it is clear.

      But in the graph, that player is only a point, and a collection of players, are only a collection of points. Why should the shape of that collection of points show the same correlation? It is not an experiment where we have repeated the measurement over and over with the same characteristics. It is 300 totally different experiments, each one with its own and different correlation between %USG and %TS. Under some assumptions, the correlation would show (e.g.: the players are similar enough, or perhaps the variance along usage is high, much higher than the variance in the skills of the players)

      But in more normal assumptions, the correlation would not show in the graphs. Normally “skill” is very widely distributed, probably following a gaussian if we could ever make that a number. Coaches take out players (points) with really bad %TS, and also make the players with very very little usage… In the end, everything conspires to make it a blur of points. But that doesn’t mean that “%USG and %TS aren’t correlated at the individual level”

    48. Z-man

      The most prevalent argument that in my view needs to be debunked (or proven beyond a reasonable doubt) is that a low-usage player like Chandler or Novak could maintain close to their efficiency at a higher usage. It is used to assert stuff like: if only Melo would defer more (or all) of his attempts to these two guys, the overall effect would be a rise in efficiency and by extension more wins. The counterargument of course is that Chandler and Novak are maximizing theor efficiency at their current usage levels, and that any significant increase in usage would have them taking shots that are less efficient than the ones that they are now taking. Or that in order to create more high-efficiency type shots (say, dunks for Chandler, uncontested 3′s for Novak) would require a change in the game landscape, via either coaching (say, going to an up-tempo style of play a la D’Antoni) or personnel (say, bringing in Chris Paul or LeBron to open up the floor for others.) In short, it’s not as simple as just changing shot distribution.

      Put another way, if Chandler and Novak take 10 shots away from Melo, what would be the overall change in efficiency for the team? Would efficiency actually go down because Novak and Chandler are now taking lower efficiency shots with those extra attempts? Is their a theoretical usage cap on the TS% of players with limited offensive repertoires in a given lineup or offensive scheme, or vs. a given team or defensive scheme? I would contend that there absolutely is one.

    49. Z-man

      Look at these 2012-13 shooting splits:

      Attempts—Shot type—-eFG%

      Melo
      44 Dunks: .932
      388 Lay-ups: .477
      36 Tips: .611
      1326 Jump Shots: .475

      Chandler
      169 Dunks: .941
      165 Lay-ups: .467
      63 Tips: .397
      44 Jump shots: .341

      So, this data suggests that if you want to make the Knicks scoring efficiency higher by transferring shot attempts from Melo to Chandler, you could only do so by creating dunks for Chandler. Any other shot type for Chandler is less efficient than any shot type by Melo.

      Right now, Chandler averages less than 3 dunks per game.

    50. nicos

      nckev: The relationship (or lack there of) between usage and scoring efficiency has nothing to do with WOW, they just put it in a graph. The data comes from box scores of real games. I don’t know what to tell you. That’s the data. In general, if you plot the change in FGA versus the change in TS% over the last 10 years, you see no correlation between those two variables. That’s not debatable. You can grab that data and plot it yourself. I really have no idea what you’re arguing about…

      Since argument regarding TS vs. Usage usually centers around guys with established skill sets (guys like Chandler and Novak on one end or Melo on the other) it’d make sense to limit that graph to mid-career players and then see if there’s any correlation. I’d say there’s a good chance you’re getting a lot of noise from young players whose TS% and usage are rising together as they improve and older guys whose TS% and usage both fall off late in their careers.

    51. EB

      nicos: So, this data suggests that if you want to make the Knicks scoring efficiency higher by transferring shot attempts from Melo to Chandler, you could only do so by creating dunks for Chandler. Any other shot type for Chandler is less efficient than any shot type by Melo.

      Of course we should also not forget that the demographic we are generally talking about (the Novaks and Chandlers) will at no point see a rise in usage without them seriously augmenting their games. Novak learning to drive or Chandler using some post moves. Because of the sharp fall in efficiency that they would experience no coach will ever let them take those shots and they themselves will not want to take those shots either since they know they aren’t good at taking them. Therefore we can never gather empirical evidence that higher USG causes lower TS%. The players awarded a higher USG are given it because they have started to shoot better, and players lose USG because their shooting has begun to fall off.

    52. Z-man

      Bynum is instructive in this case. In 2011-12, he shot at a higher percentage than Chandler on dunks, lay-ups and tips. However, he shot worse on jump shots and hooks but took many more jump shots and hooks than Chandler did. So if Bynum decreased his usage by eliminating his jumpers (he made hooks at a decent .486 clip) he could easily have better TS% numbers than Chandler. But what is better for the team?

    53. Unreason

      EB:Because of the sharp fall in efficiency that they would experience no coach will ever let them take those shots and they themselves will not want to take those shots either since they know they aren’t good at taking them. Therefore we can never gather empirical evidence that higher USG causes lower TS%. The players awarded a higher USG are given it because they have started to shoot better, and players lose USG because their shooting has begun to fall off.

      Love this both for the clear insight into the danger of inferring from correlation and the latent insight on selection pressure. Natural limits on USG caused by versatility confound the association between USG and efficiency. As you imply, a manipulation would be required to see what would happen if low versatility/high efficiency players increased their usage beyond those natural limits. Without the ability to manipulate in controlled experiments, evolution could provide an alternative paradigm for examining the question. Such a view would predict this lack of an association because efficiency and the forces affecting it act like a fitness landscape on player behavior. Maybe the green light to shoot over the course of a career is shaped in the same way that phenotypes are selected for by the environmental factors that determine reproductive success in organisms on an evolutionary time scale. Evidence for this could come from analysis of data on changes in shot-specific USG and efficiency in players who accommodate or fail to accommodate to decreased athleticism by driving less, developing a three, or a post game, etc. Looking for the USG efficiency link within players but across shot types, in other words, instead across players.

    54. The Honorable Cock Jowles

      Unreason: Maybe the green light to shoot over the course of a career is shaped in the same way that phenotypes are selected for by the environmental factors that determine reproductive success in organisms on an evolutionary time scale.

      I don’t understand this at all. Are you saying that players who have the ability to shoot at will are the players who should be shooting at will? Nick Young would like to have a word with you.

    55. EB

      Unreason: Love this both for the clear insight into the danger of inferring from correlation and the latent insight on selection pressure. Natural limits on USG caused by versatility confound the association between USG and efficiency. As you imply, a manipulation would be required to see what would happen if low versatility/high efficiency players increased their usage beyond those natural limits. Without the ability to manipulate in controlled experiments, evolution could provide an alternative paradigm for examining the question. Such a view would predict this lack of an association because efficiency and the forces affecting it act like a fitness landscape on player behavior. Maybe the green light to shoot over the course of a career is shaped in the same way that phenotypes are selected for by the environmental factors that determine reproductive success in organisms on an evolutionary time scale. Evidence for this could come from analysis of data on changes in shot-specific USG and efficiency in players who accommodate or fail to accommodate to decreased athleticism by driving less, developing a three, or a post game, etc. Looking for the USG efficiency link within players but across shot types, in other words, instead across players.

      I appreciate the praise!

    56. Z-man

      The Honorable Cock Jowles:
      http://www.hickory-high.com/?page_id=6155

      Chandler is not a player simply converts high-quality shots at a high rate. According to xPPS, he not only takes high-quality shots at one of the highest rates in the league, he achieves a higher PPS than would be expected of his shot selection.

      Interesting website and methodology. My take is that the flaw in this stat is that it lumps all shots in the restricted area into one category.

      Z-man: Attempts—Shot type—-eFG%

      Melo
      44 Dunks: .932
      388 Lay-ups: .477
      36 Tips: .611
      1326 Jump Shots: .475

      Chandler
      169 Dunks: .941
      165 Lay-ups: .467
      63 Tips: .397
      44 Jump shots: .341

      If you lump dunks, tips and layups together, Chandler’s eFG% on these shots is higher than Melo’s, but that is only because he has a higher proportion of dunks in the restricted area. So my qualm with Hickory’s methodology is that it doesn’t account for the expected PPS at a micro enough level. If they segregated out dunks vs. non-dunks in the restricted area, Melo’s xPPS stats would likely improve and Chandler’s stats would likely decline. I guess you could argue that Chandler dunks balls that Melo would lay up given the same circumstances, but that would only account for part of the differential.

      The point I made in @57 remains the same: that Chandler’s uberefficiency is mainly due to the high ratio of dunks to total shots, not to close-in shots to total shots. His eFG on non-dunks in the restricted area is rather low. So, the only way to keep him at that level of efficiency at a higher usage would be to add dunks, which is probably the hardest shot type to generate.

    57. nicos

      The Honorable Cock Jowles:
      http://www.hickory-high.com/?page_id=6155

      Chandler is not a player simply converts high-quality shots at a high rate. According to xPPS, he not only takes high-quality shots at one of the highest rates in the league, he achieves a higher PPS than would be expected of his shot selection.

      Nobody would deny he’s a terrific pnr player, offensive rebounder, etc… The question is does he cost other players high quality looks because his man is able to drop into the lane due to Chandler’s complete unwillingness to take even a ten foot jumper. It’d be interesting to see how many shots at the rim Chandler’s teams are getting when he’s on/off the floor- is he really generating more shots at the rim for the team? There has to be an asterisk next to Chandler’s efficiency the question is how big- I’d say he’s still a top 30-40 offensive player but not the top 10-20 guy you seem to think he is.

    58. nicos

      I also think turnovers have to be factored in- for his career Chandler has been very turnover prone (though he was above average for a center last year) and- EYE TEST ALERT- it seems to me the Knicks guards turned the ball over a lot trying to force the ball to him with ill-advised lobs etc… I think you can ask if he’s giving back some of the extra shots he’s generating (if he is generating them) by way of turnovers.

    59. The Honorable Cock Jowles

      No one’s going to argue that Chandler could duplicate Anthony’s numbers (or vice-versa) if you switched their roles.

      The argument is simply that if you take away four of Carmelo’s low-efficiency shot attempts (ISOs, for one) per game, and distribute them through similar plays (Chandler’s PnR, or a kick-out to a player in the corner), there won’t be a significant drop-off in efficiency in those players.

      It’s not so hard to understand that the ISO is a decision, and not always a consequence, and should be eliminated as much as possible from the team’s shot distribution.

    60. Z-man

      I agree that a 36+% usage% is too high and not healthy for the team. It is especially troubling that it occurred in a season that started off with Melo musing that scoring lots of points didn’t mean anything to him anymore.

      That said, shooting at that usage% to a .560 TS% is pretty remarkable. The guy is just an incredible offensive player. His basketball IQ isn’t the greatest, but he is not the problem on this team.

      I think that swapping out Beno for JKidd, Bargnani for Novak, MWP for Cope, Tyler for Camby and TH2 for White will make this team stronger, and partly for the reason you cite…less reason for Melo to iso.

    61. Unreason

      The Honorable Cock Jowles: I don’t understand this at all. Are you saying that players who have the ability to shoot at will are the players who should be shooting at will? Nick Young would like to have a word with you.

      Well if you get the idea that rewards shape behavior but don’t determine it absolutely you get the gist. Inefficiency for certain types of shots has consequences -getting pulled, yelled at, thinking “I suck at this”, etc. – that shape behavior. If a long sequence of attempts suggests that you’re to bad at a shot, it has a lower chance of surviving to be reproduced. The consequences don’t guarantee optimality. They just shape shot attempt behavior in a way that is analogous to fitness shaping reproductive success. Those who are only good at one thing but take every opportunity to do stuff they suck at in most cases will experience consequences that limit future opportunities.

      The potential value of thinking in theses terms would be if borrowed concepts and analytic approaches used in evolutionary theory could be usefully applied to hoops. Not sure whether it will or not.

      Re Nick Young, the metric for defining such examples as extreme just shows there’s an assumption that inefficiency should have consequences that affect opportunities to shoot. The analogous case in evolution is where conditions allow for reproductive success despite some diminished capacity: islands where generations of birds gradually become flightless because food is plentiful on the ground and there are no predators.

      Martin Nowak’s “Evolutionary Dynamics” presents and applies powerful analytic strategies from evolutionary game theory to many topics. That’s the kind of thing I was thinking of. I’m not sure how appropriate those are to hoops but on the surface they seem apt.

    62. Z-man

      I don’t know how much running additional P&Rs w/ Chandler will increase efficiency. P&R plays have undetermined outcomes, and turnovers are part of the picture, as are missed layups for the ballhandler or other breakdowns. The success of the roll man is very dependent on the skill of the ball handler.

      In other words, the argument that Melo should iso less is not necessarily consistent with the argument that Chandler should shoot more. More accurately, one could argue that Melo should look carefully at his iso stats and figure out what shots to replace by other shots, no matter who takes them (including him!).

    63. The Honorable Cock Jowles

      Unreason: Inefficiency for certain types of shots has consequences -getting pulled, yelled at, thinking “I suck at this”, etc. – that shape behavior. If a long sequence of attempts suggests that you’re to bad at a shot, it has a lower chance of surviving to be reproduced. The consequences don’t guarantee optimality. They just shape shot attempt behavior in a way that is analogous to fitness shaping reproductive success. Those who are only good at one thing but take every opportunity to do stuff they suck at in most cases will experience consequences that limit future opportunities.

      I understand what you’re saying. Just as potential mates often “choose” the most fit based on demonstration and subjective analyses, so do NBA coaches (and other decision makers) toward the players.

      When a difference in two or three shots per hundred is the difference between winning and losing, I think it’s much harder to assess value, including “fitness.”

    64. nicos

      Z-man:
      I don’t know how much running additional P&Rs w/ Chandler will increase efficiency. P&R plays have undetermined outcomes, and turnovers are part of the picture, as are missed layups for the ballhandler or other breakdowns. The success of the roll man is very dependent on the skill of the ball handler.

      In other words, the argument that Melo should iso less is not necessarily consistent with the argument that Chandler should shoot more. More accurately, one could argue that Melo should look carefully at his iso stats and figure out what shots to replace by other shots, no matter who takes them (including him!).

      Chandler’s usage (aside from his first two years when he DID take jumpers) topped out at 14.5 when he was playing with a top 5 all-time point guard in CP3. He’s been at 13 the last two years so I’m not sure if you can find him many more shots unless he’s willing to pick and pop every once in a while- the question then is a pick and pop from Chandler (and the possibility that taking a few of those might open up a few more opportunities at the rim) better than a Melo iso. I’m with THCJ in terms of being skeptical that Melo’s going to start passing up iso shots that he’s been taking his entire career. That said, the Knicks run a ton of pnr with Chandler and unless you’re playing Amar’e alongside Melo and Chandler I’m not sure what sets the Knicks should go to outside of iso-melo once the initial sets break down. Certainly Melo should be looking to reset early in the clock more often than he does now. As good as he’s been on both ends of the pnr the last couple of years it’d be nice to have some built-in side pnr options for him off of the catch on the right wing if nothing else.

    65. Brian Cronin

      I agree that a 36+% usage% is too high and not healthy for the team. It is especially troubling that it occurred in a season that started off with Melo musing that scoring lots of points didn’t mean anything to him anymore.

      That, for me, was one of the oddest little twists of fate from last season.

    66. Z-man

      Some of the super-high usage (even for Melo) was defensible:

      Chandler (and later, KMart) took few and only certain types of shots
      Prigioni refused to shoot until compelled to by Woodson
      Felton shot poorly once he injured his hands
      Amare got hurt and rarely played with Melo
      JR took the same shots as Melo at a lower efficiency
      Novak had lots of trouble getting open
      Kidd would only shoot wide-open 3′s and started missing them
      Shump didn’t get it going until just before the playoffs
      Cope was underutilized by Woodson
      White coudn’t shoot
      Sheed, Thomas and Camby broke down

      Switch Durant with Melo last year and I would expect that Durant’s usage% would have been well above his career norm.

    Comments are closed.