This is our old blog. It hasn't been active since 2011. Please see the link above for our current blog or click the logo above to see all of the great data and content on this site.

Layups: Quantifying the “Ewing Theory”?

Posted by Neil Paine on August 3, 2009

Seems like I've been running lots of Bill Simmons-related links recently for some reason... This one comes courtesy of Gravity and Levity (a great intellectual blog about science in general, and physics specifically, in case you want to check out some of the other posts there), which makes a really fascinating connection between Braess’s Paradox -- "closing a road can actually improve traffic" -- and Simmons' old "Ewing Theory", the observation that sometimes teams who lose their best player biggest star actually do better without him (so named after the '99 Knicks went to the Finals sans Patrick).

(H/T: hoopster101 at APBRmetrics.)

11 Responses to “Layups: Quantifying the “Ewing Theory”?”

  1. Jason J Says:

    I always disliked using 1999 Ewing as the example for that phenomenon because it's not as though the team actually became significantly better w/out Patrick. In the regular season he led the team in PER, Pts/gm, and DWS. He got hurt in the playoffs, and the team managed to advance to the finals thanks to a miracle bounce in last second shot by Houston against the Heat, a suspect miracle 4 pointer by LJ against the Pacers, and the blossoming of Marcus Camby.

  2. gravityandlevity Says:

    Please join (or at least peruse) the conversation at the APBRmetrics forum:

    I would love to get your opinions.

  3. Neil Paine Says:

    Are you talking to me or Jason? Not sure if he's on APBRmetrics (I mean, you gotta be, right? Right?), but I'm known as "davis21wylie2121" over in those parts. In fact, I just realized I didn't give a hat tip for the story to "hoopster101", who actually alerted me to your post. I like your blog, btw, I was getting engrossed reading some of the other entries (the one on Gompertz's Law was especially interesting)... A model for the probability of human death by age -- oh those actuaries!

  4. Jason J Says:

    You know, I love APBRmetrics, but I can't get there while I'm at work because of our network's content blocker (which seems to arbitrarily block some blogs and not others).

  5. Walter Says:

    Hi Neil, I have a quick question... would you be willing to post interesting statistical analysis done by others who have no other means of posting their research? The reason I ask is that I am an actuary (which you so kindly referenced above) and am a die-hard basketball fan who loves the advanced statistics. I have worked on some difference advanced statistical analysis in the past (primarily Lakers based as I am a fan) and thought about building my own website to publish it. Unfortunately I do not have the time to pursue such and endeavor but would love to find a site willing to post the work if they feel it meets their standards.
    Let me know. Thanks.

  6. Neil Paine Says:

    Hi Walter, sorry if I seemed disdainful of actuaries above -- I actually knew a few future actuaries at Georgia Tech and they were all fine, upstanding human beings. :) I'm also very glad to hear of your interest in advanced stats; the more people we can get involved from fields like yours, the better. Having a better understanding of risk and uncertainty with regard to player performance is a huge part of the effort we're trying to make in APBRmetrics. In terms of publishing your own work, I think you should sign up for a (free) account at the APBRmetrics Forum and start posting your studies there if you feel that your own site would be too time-consuming. That's essentially how I started out a few years ago -- you can post your research whenever you want, you'll get read by (and feedback from) many of the big names in NBA analysis, and others can have a chance to build off of your work if they want. It's basically the hub for all things related to statistical analysis in basketball, and I have a feeling that someone with your professional background would fit in very well there.

  7. Ty Says:


    What the hell?! I'm so disappointed in you! Don't tell me you're actually endorsing that ridiculous "Ewing Theory"... are you? That can't be! First of all, its illogical. Second of all, your own Win Share numbers rightfully debunk the notion that Ewing was even the best player on the Knicks in 1999. By Win Rate the best player, even in the regular season, was Marcus Camby. And he stepped his game up two levels in the playoffs. And Camby's subsequent play has shown he was the better center. By Win Score, Win Shares, or Marginal Win Score, Camby as a full-time center has well outproduced Ewing. Ewing was slightly above average, nothing more. Camby was and is an efficient machine.

  8. Jason J Says:

    Ty: I'm guessing you ran your numbers on total careers. If you look at Pat and Marcus through the age of 34 (Camby's age at the end of last year) -

    - You see a different story.

    Camby has the superior defensive numbers, but the Drtgs are only one point apart, and overall Ortg -Drtg completely evens out. Ewing's a more efficient scorer and much more prolific as his usage is more than 10pts higher than Camby's.

    Additionally, the factor that makes the two of them look comparable is that we work in metrics that consider stats per minute or per possession. Patrick had a much healthier career through the age of 34 and gave 6 more minutes of production per game, meaning that overtime you get his superior production not just for the 30 minutes that Camby plays but also for 6 additional minutes when Camby's backup is in the game.

    That led Patrick to have a 42.1 edge in WS at the same age, which is significant, but even if we level it out and look at per minute WS Pat's got a .0033 WS/min while Camby has a .0029 WS/min. Outside of rebound and block rate there's really no argument for Marcus over Pat when you compare them through age 34.

    ALL of that being said - you're right in that Camby may have been the better player in '99 and was fantastic in the playoffs.

  9. Neil Paine Says:

    Sorry Ty, I should have phrased it as "when a team loses the player perceived by the public to be the star of the team, and the team does better". However, you've got your Win Shares wrong that year -- the Knicks' top player during the season was neither Ewing nor Camby: it was actually Larry Johnson (although Camby was great during the playoffs and led the team in WS).

    Anyway, the phenomenon (it's really not a "theory", is it?) is informative in that it identifies players who are overrated by the conventional wisdom and are proven so in their absence. Obviously a team's true best player would hurt the team when they miss time, so it's more about players who aren't really their teams' best, but for some reason are viewed as such by the public.

  10. Ricardo Says:

    Why the hell is this thing called "The Ewing Theory" and not "The Ewing Effect"??

  11. Rashidi Says:

    Well, part of the theory I think is that Ewing was taking away shots from Houston and to a lesser extent Houston. The main critique of Patrick was that he didn't trust his teammates enough, and tried to do too much on offense. He wasn't as efficient a scorer as the other centers of his era which is why they were all considered a level above him.

    This was the media's perception of him though, which was hardly ingrained in fact - Ewing DID defer more as he got older.

    Let's not forget that 99 was an asterisk season, and that the "8th seed" Knicks trailed the "1st seed" by a mere 6 games. In an 82 game season the Knicks likely would have been a 4th seed, because they suffered some injuries early and gelled late... Ewing for instance missed 12 games, which might not seem like much, but that was literally a quarter of the regular season.