1. Welcome! Please take a few seconds to create your free account to post threads, make some friends, remove a few ads while surfing and much more. ClutchFans has been bringing fans together to talk Houston Sports since 1996. Join us!

A partial analysis of Birnbaums refutation of Berri's WoW

Discussion in 'NBA Dish' started by jtr, Oct 9, 2013.

Tags:
Page 1 of 2
  1. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    I have decided to take an in depth look at one of my favorite NBA statistical analysis articles:

    http://blog.philbirnbaum.com/2011/01/do-players-steal-rebounding.html

    in order to test drive new statistical analysis software. A couple of days ago I again downloaded the optimized Perl mathematics suite, which I had used a couple of years ago on a couple of huge data mining and statistical analysis projects. But that software package is a cruise missile. Why use a cruise missile when you are trying to open a small box of statistical data? So I went and downloaded trial versions the IBM SPSS and Minitab software packages each fronted by an easy to use UI.

    The article is dated 12/2011. That was a time period in which Birnbaum and others were in a titanic struggle with Berri and his WoW website. The issues were somewhat technical but not overly complicated.

    I decided to recreate his study using 2012-2013 NBA season data. One of the central tenants of his article is:

    “The variance in team rebounds per game is LESS than the variance in rebounds per game by the center alone!”

    Well to sum it up that would be a huge deal in advanced statistical analysis of the NBA. So I plugged the 2012-2013 data into those nifty expensive software applications and that conclusion did not hold for last seasons data.

    Creb Oreb Treb
    VARP 2.04 2.71 2.55
    STDP 1.43 1.65 1.60

    At that point I realized that either my tools were wrong or perhaps Birnbaum had identified a single season statistical anomaly. Or maybe something else. I then looked at the 2008-2009 season, using the exact same data that Birnbaum had used. Interestingly enough the statistics of the data, while close, are not identical.

    VARP 2.54 2.83 1.83
    STDP 1.59 1.68 1.35


    Birnbaum found that

    “-- For centers, the SD was 1.63, and the range from worst to best was 7.4 (10.3 to 17.7).
    -- For teams, the SD was only 1.43. And the range was only 4.0 (38.8 to 43.8). “

    Not identical, but close. Why are they not identical? In Birnbaum's article he had made an error:

    “(UPDATE, 1/23: I realized I accidentally left out one team, and entered one team twice. Corrections have been made below.) “

    But corrections were never made above. Since I do not know the team left out nor the team that was included twice, I have to assume that IBM's SPSS and Minitab are performing up to expectations.
     
    #1 jtr, Oct 9, 2013
    Last edited: Oct 9, 2013
    1 person likes this.
  2. Outlier

    Outlier Member

    Joined:
    Mar 25, 2006
    Messages:
    8,529
    Likes Received:
    1,351
    Wtf is this
     
    #2 Outlier, Oct 9, 2013
  3. Patterned919

    Patterned919 Member

    Joined:
    Jan 19, 2013
    Messages:
    2,332
    Likes Received:
    163
    In on jtr thread
     
    #3 Patterned919, Oct 9, 2013
  4. napalm06

    napalm06 Huge Flopping Fan

    Joined:
    Sep 30, 2008
    Messages:
    26,929
    Likes Received:
    30,543
    It's a beautiful day today. I think I'll go outside.
     
    #4 napalm06, Oct 9, 2013
  5. durvasa

    durvasa Member

    Joined:
    Feb 11, 2006
    Messages:
    38,893
    Likes Received:
    16,449
    I think the central claim of his article is this:

    That is: if a player gets a lot of rebounds, even adjusted for what position he plays, a large part of his total is rebounds that other players would have gotten to regardless. ​

    Based on your review of the data, would you agree or disagree with that?
     
    #5 durvasa, Oct 9, 2013
  6. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    Actually the argument is much more nuanced than that. It was a struggle between the two major advanced stats camps 22 months ago. Berri had made a serious statistical case that awarding the player who corralled the rebound full credit. Baunbaum and most of the other stats gurus took an opposing position.
     
    #6 jtr, Oct 9, 2013
  7. durvasa

    durvasa Member

    Joined:
    Feb 11, 2006
    Messages:
    38,893
    Likes Received:
    16,449
    Yeah, I'm somewhat familiar with it (though by now I've probably forgotten a lot of the details). I just recall not being at all convinced by Berri's arguments. In effect, the way he defines "possessions employed" (PE) leads him to treat a missed field goal as equivalent to a turnover, and a steal as equivalent to an offensive rebound. But there was no convincing rationale presented for defining PE in that way.
     
    #7 durvasa, Oct 9, 2013
  8. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    Realistically it was a trial run of new software. But I focused in on a debate that I find fascinating. I find it interesting that Birnbaum's major bombshell is probably just a statistical anomaly. Since he did not pursue the topic I assume that it was discredited. I just had never understood why.
     
    #8 jtr, Oct 9, 2013
  9. heypartner

    heypartner Member

    Joined:
    Oct 27, 1999
    Messages:
    63,510
    Likes Received:
    59,002
    But what does that tenant even mean? I mean; coaches actively release players in transition when they have a stellar rebounder. So, yes, people are taught to give up the rebound to the rebounder.

    Also, sometimes that rebounder is the best outlet passer,,,so we can say, you want your best low post passer to get the rebound.

    There are many reasons that this is a planned strategy. Let Player A get all the rebounds for Reasons X, Y and Z.
     
    #9 heypartner, Oct 9, 2013
  10. durvasa

    durvasa Member

    Joined:
    Feb 11, 2006
    Messages:
    38,893
    Likes Received:
    16,449
    What is VARP and STDP? Can you clarify how your results differ from his? I can't tell from the way you presented the numbers in the first post.
     
    #10 durvasa, Oct 9, 2013
  11. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    Absolutely true. You support Taunbaum's position. As do I.
     
    #11 jtr, Oct 9, 2013
  12. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    VARP is simply Variance when the sample set includes all the members of the population in the statistical sample.

    STDP likewise is just he standard deviation over a sample set that includes the entire population.

    If you need further clarification (as I have needed over the last couple of days) please just ask.
     
    #12 jtr, Oct 9, 2013
  13. durvasa

    durvasa Member

    Joined:
    Feb 11, 2006
    Messages:
    38,893
    Likes Received:
    16,449
    This is the sort of question it is addressing: if you replace all the minutes of a player who averages 9 rpg with a player who averages 12 rpg, will his team get 3 extra rebounds per game? More generally, how must impact will an individual player's rebounding ability have on his team's rebounding performance?
     
    #13 durvasa, Oct 9, 2013
  14. iconoclastic

    iconoclastic Member

    Joined:
    Oct 10, 2007
    Messages:
    6,100
    Likes Received:
    422
    Why don't you just watch the games and record how many rebounds a player gets that his teammates could possibly have gotten to instead of drawing imprecise conclusions from simple data?
     
    #14 iconoclastic, Oct 9, 2013
  15. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    Because with 1200+ games in a NBA season (each game includes 2 teams which cuts the available NBA games by 50%) I could not possibly do it. I do not think it is humanly possible.
     
    #15 jtr, Oct 9, 2013
  16. durvasa

    durvasa Member

    Joined:
    Feb 11, 2006
    Messages:
    38,893
    Likes Received:
    16,449
    OK, so you got 1.59 SD for center rebounding by team, and 1.35 for total team rebounds. Birnbaum got 1.63 to 1.43. Did you classify players the same way the did? How did you decide who qualifies as a center? That could be another reason for the discrepancy.

    The thing is, I don't know if the "anomaly" in this case is the season Birnbaum originally looked at, or the 2012-2013 season. Maybe a better way to validate it would be to run the analysis over several seasons.
     
    #16 durvasa, Oct 9, 2013
  17. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    I honestly have not done that analysis. And I know of no references to articles having claimed to have done so. I might be tempted to address the issue, but the complexity of the analysis leaves me clueless. Maybe in the future.
     
    #17 jtr, Oct 9, 2013
  18. jtr

    jtr Member

    Joined:
    Dec 4, 2011
    Messages:
    7,470
    Likes Received:
    275
    I judiciously followed his instructions.

    "So, here's what I did. For every team in the 2008-2009 season, I ran a regression comparing the total rebounds by the team's centers (combined) to the total for the rest of the team (combined). I typed in the data manually from some team pages at the 82games.com website, because that's where I happened to find it broken down by position. (Strangely, basketball-reference.com doesn't do that.)"

    That is the only clue as to his methodology.
     
    #18 jtr, Oct 9, 2013
  19. durvasa

    durvasa Member

    Joined:
    Feb 11, 2006
    Messages:
    38,893
    Likes Received:
    16,449

    http://www.countthebasket.com/blog/...he-value-of-offensive-and-defensive-rebounds/

    The author, Eli Whitus, is an analyst employed by the Rockets for the last several years by the way.
     
    #19 durvasa, Oct 9, 2013
  20. iconoclastic

    iconoclastic Member

    Joined:
    Oct 10, 2007
    Messages:
    6,100
    Likes Received:
    422
    But teams like the Rockets probably do it already with a bunch of analysts all splitting up the work, with the results being proprietary. My point is, a better approach to the solution already exists, so what's the point in doing an inferior analysis?
     
    #20 iconoclastic, Oct 9, 2013
(You must log in or sign up to post here.)
Page 1 of 2

Share This Page

Style
ClutchFans Dark

  • About ClutchFans

    Since 1996, ClutchFans has been loud and proud covering the Houston Rockets, helping set an industry standard for team fan sites. The forums have been a home for Houston sports fans as well as basketball fanatics around the globe.

  • Rockets Content

  • Support ClutchFans!

    If you find that ClutchFans is a valuable resource for you, please consider becoming a Supporting Member. Supporting Members can upload photos and attachments directly to their posts, customize their user title and more. Gold Supporters see zero ads!


    Upgrade Now