Loading ...
Sorry, an error occurred while loading the content.

Re: Pro Basketball Prospectus

Expand Messages
  • Dean Oliver
    ... Same paper. I just felt I owed him some time to really explain what he had. ... straight ... and ... you re ... Yup. ... Tsk, tsk. ;) ... at ... I also
    Message 1 of 39 , May 26 10:02 PM
      --- In APBR_analysis@yahoogroups.com, "Kevin Pelton" <kpelton08@h...>
      > --- In APBR_analysis@yahoogroups.com, "Dean Oliver" <deano@r...>
      > wrote:
      > > It's another linear weights method, like John's, Dave Heeren's,
      > > Bob Belotti's. It's got a dramatically different set of weights
      > > than everyone else's. It's a different approach to finding the
      > > weights than everyone else has taken, though.
      > To me, this is Berri's biggest flaw. I didn't read all of his new
      > stuff (did read the paper you posted here waaaaaaaay back in the
      > day - it was actually my first post here - but looking at the part
      > where he referenced Rob Neyer, he seems to be using linear weights'
      > success in baseball to justify using them in basketball.

      Same paper. I just felt I owed him some time to really explain what
      he had.

      > In baseball, the fragments really do add together. Get four
      > singles and, barring an intervening event like a caught stealing,
      > you will score a run. You can't say the same about something like a
      > defensive rebound. You could grab 10 straight offensive rebounds
      > still not score a point, for example. I guess this gets at one of
      > DeanO's criticisms of linear weights (I think), which is that
      > mixing possessions and points, which are two very different things.


      > Not that it stops me from using linear weights, but whatever.

      Tsk, tsk. ;)

      > The thing of it is - this seems more like where baseball analysis
      > is going. BaseballProspectus' current method for evaluating hitters
      > uses Marginal Lineup Value, the difference created by adding a
      > hitter to an otherwise-average lineup, replacing an average player.
      > Replacement level is subtracted from this to give Value Over
      > Replacement Player.

      I also really like this concept. The Winston-Sagarin method we
      discussed also goes at this but it too had some funky results -- a
      team with Tim Duncan and 4 average guys wins 73 games, according to
      them. But if you can evaluate a marginal value on an average team,
      you can evaluate a marginal value on a good team or a bad team.
      These things are DEFINITELY not the same for many players.

      > Of course, just because baseball folks are doing it doesn't make it
      > right and I'm not sure how accurately we can ever rate defense
      > without something like DeanO's defensive stats to help us

      Unfortunately, it's tough to maintain the Project Defensive
      Scoresheet work we did last year without a full time coordinator of
      the project. I had relatively more time last year to do this and
      still couldn't keep up. This year, I have much less time. I'm
      hoping we find a way to maintain it, however. It did have
      tremendously valuable info -- info I haven't even tapped into fully.

    • fadeawaymike
      I ve bought your book also. First I ve read the Prospectus of Hollinger and after that Basketball on paper. I v read reviews that said: the prospectus is good
      Message 39 of 39 , Feb 1, 2004
        I've bought your book also. First I've read the Prospectus of
        Hollinger and after that Basketball on paper. I'v read reviews that
        said: the prospectus is good but basketball on paper is better. Well
        I don't agree. The analytic methods are there but the conclusions
        are sometimes not so strong. For example your rebounding part. You
        make a conclusion and than you begin to correct yourself(offensive
        rebounds create free throw attempts,...)
      Your message has been successfully submitted and would be delivered to recipients shortly.