The below was posted in the "Things you don't like about Harden" thread. But it wasn't about "things I didn't like about Harden", rather some observations on EWA and EWA/USG (where Harden ranks #5 in the league this year). It also discuses a little bit about Lin vs Dragic/Lowry (just data, no heated arguments please), and looks at EWA/USG for some PF targets discussed on clutchfans. Given that it wasn't about Harden, and certainly nothing to dislike about Harden based on these numbers, I thought I'd post it in a new thread and see if people are interested to discuss it. Not as long as my past thread posts, thankfully. ---------------------------------------------------- The interesting thing is LBJ, Wade and Bosh accounts for 57 wins (their total EWA) for the team. Miami is on track for 64-65 wins. So you can see having superstars that does everything overstate their EWA somewhat, with the role players EWA probably understated. The other thing is the EWA is highly correlated to usage. If we use EWA/USG, for point guards Chris Paul comes out #1 by a long margin at 0.694. Westbrook #2 0.569, then Curry, Calderon, Parker, D Williams, and then Mike Conley at #7. Dragic is #10, Irving is #12, Lillard #14, Lowry #21, Wall #25 and Lin #26 0.254. Before this becomes a Dragic vs Lowry vs lin argument again based on simple numbers/analysis, it's obvious that you have to adjust Lin's ranking due to him playing next to a ball dominant SG in Harden. Steve Nash is ranked #33 0.219 in the list playing next to Kobe, and Mario Chalmers #36 0.144. Basically a ball-dominant SG takes away PER, EWA etc mostly from the PG. There's no way Steve Nash is worse than Jeremy Lin. And the numbers support this. Steve Nash had EWA/USG of 0.423 last season in Phoenix, and this would have ranked him #8 among PGs this year, and he was ranked #4 last year. But instead he has roughly half his EWA/USG this year playing next to Kobe, and is ranked #33. In terms of across all positions, anyone who doubts LBJ and Durant should be fighting it out for MVP this year (and are head and shoulders above the rest of the league) should check out their EWA/USG. 1. LBJ - 1.049 2. Durant - 0.934 3. Tyson Chandler - 0.754 4. Chris Paul - 0.694 5. Harden - 0.668 6. M. Gasol - 0.619 7. Horford - 0.618 8. Kobe - 0.599 9. B Lopez - 0.589 10. Ibaka - 0.586 Note that EWA/USG doesn't rank the best players. It just ranks win share per usage. So it roughly ranks how much a player contributes towards winning per possession used. Note I said 'roughly', I know neither statistic is perfect. But MVP should reflect the amount of wins they contribute to the team, and weighing it for usage is an ok way of doing it. Interesting notes: Miami and OKC each has has 3 players in the top 15 this year. Tyson Chandler was ranked #2 last year, just ahead of Kevin Durant (& #3 this year). And we know Chandler was Morey's target a few years ago, but the trade fell apart. Gortat was #4 last year playing alongside Nash, but fell apart this year (dropped to #88, bickered with coach and teammates etc) with Dragic as his PG. Perhaps one of the reasons why Dragic's stats look fine, but Phoenix has fallen off so much this year vs having Nash as PG. And one of the arguments against Dragic (I won't go into 'for' Lin arguments ... just look at the Linsanity roster last year with Melo and Amare sidelined, and the Knicks record) is that he doesn't really make his team/teammates better. Bynum and Howard were ranked #6/#7 respectively last year. Kevin Love #9. James Harden was #17. Ibaka #19, Westbrook #20, so OKC had 4 players in the top 20!!! If they had kept the team together, with more seasoning, this young group would trump Miami, in my opinion. Note Bynum, Howard, Love all appear(ed) to be Daryl Morey targets. With Harden gone this year, the other players for OKC have picked up the slack, especially Westbrook, which again shows that PGs that play with ball-dominant SGs have their EWA/USG understated. Westbrook's EWA/USG jumped from 0.457 to 0.569 this season. Kevin Martin is #86 on the EWA/USG list this year, which shows that the Rockets probably would've lost a lot more with Martin as opposed to Harden, but Lin would probably have better numbers (not that this is any good for the Rockets and is not what this discussion is intended to lead to). Also interesting is that the #4 SG in the NBA this year based on EWA/USG is JR Smith, and he's ranked #60. So really, only Lin, Nash and Chalmers have their numbers materially adversely impacted by a ball dominant SG. It also means we have by far one of the top 3 SG in the game today, and as Wade and Kobe ages, and Harden (hopefully) improves defensively, we'll have the marquee SG in the game. Larry Sanders is already #23 this year, on rebounds and blocked shots alone. Comparing potential PF targets for Morey LMA is #28 this year and #21 last year. Millsap is #33 this year, and #15 last year. Josh Smith is #25 last year but only #54 this year. Love was #9 last year, I think we can give his numbers this year a miss due to injury. So even with a Love, or LMA/Millsap/Smith, we won't come close to OKC/Miami's 3 top 15 EWA/USG, but what we do have is players on better contracts like Asik (#59 this year), Parsons (#61) and Lin, getting good bang for the buck with Asik and Lin on $8M, and Parsons on a near minimum contract. So I guess for Morey, failing to get top 20 players, he'll try to surround Harden with maybe top 50-60 players but at half of a max contract (like Asik and Lin), and try to get lucky in the draft with players like Parsons. And finish the roster with veterans on cheap contracts like Delfino, players on rookie contracts like Jones, DMo (and Patterson the last couple of years), and overseas great finds like Beverley.
It seems like you're assuming that EWA depends on USG. The formula for EWA is (roughly): EWA = Min * (PER - 11 ) / 2010 Its essentially a VORP ("value over replacement") stat. It assumes that the more a player who has a PER > 11 plays, the more he's helping his team. The formula for USG is something like: USG-Rate = (FGA + FTA*.44 + .33*AST + TOV)*40*LgPace / ( Min*TmPace) You can see that EWA will increase/decrease with more minutes, while USG-Rate is intended to be "minutes adjusted". Therefore, if one was to take the ratio between the two, what does that really signify? Supposing that PER correlates highly to USG-Rate which I think is your main assumption, I would think that it mainly just reflects how many minutes the player got.
Good post, solid statistics. People need to read beyond the numbers more often. You can spin a stat to support or contradict any argument so it's nice to see someone take a comprehensive high level look. Nice job clarifying that you're not "pro/con" specific players. Truly you know your audience :grin:
Actually, not that EWA depends on usage, but that it is quite highly correlated, and that those with a high EWA/USG are more effective/efficient and/or helps the team win more (adjusted roughly for possessions used). I think we can see that the top players in the league all get around 35 to 40 minutes per game (if healthy), and yet they have different PERs, USG, and EWA. Their effectiveness in those minutes played and even with the number of possessions used are quite different. So no ... for someone that plays 38 minutes, you can either have a Rudy Gay PER or a Lebron James PER or EWA, and that is not sufficiently explained by their usage levels. If you pull out the raw USG, PER and EWA data, you'll appreciate that it doesn't just reflect minutes played. LBJ and Durant have 4.5 times the EWA/USG of Rudy Gay, and they certainly don't play 4.5 times more minutes.
I hope my audience are people who want to discuss basketball objectively ... and offer insights that help me understand basketball and our Rockets team more.
