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

Re: Individual W/L and production

Expand Messages
  • ssims22000
    ... Thanks. I do want to inject some severe caution into using this rating I posted as any sort of true player rating. For one thing, there are several ways
    Message 1 of 6 , Jan 3, 2004
    • 0 Attachment
      --- In APBR_analysis@yahoogroups.com, "Mike G" <msg_53@h...> wrote:
      > --- In APBR_analysis@yahoogroups.com, "ssims22000" <ssims2@i...>
      > wrote:
      > ...> Pro/G is my attempt to gauge value over replacement. ...>
      > > Ratings are through Dec. 24.
      >
      > Simmy, this is excellent. I've tried to find time for a
      > comprehensive reply. But before this info becomes too stale, I'll
      > just blurt out some stuff.
      >

      Thanks. I do want to inject some severe caution into using this
      rating I posted as any sort of true player rating. For one thing,
      there are several ways of calculating individual win-loss records,
      and I'm not yet sure which one is the best. Each one will produce
      sometimes significantly different results, so take all of this with
      a grain of salt.

      > I compared your team rankings to my team rankings, which are based
      > on sums of individual production. I found an average of 2% above
      > and below the norms for the West and East, respectively, in
      > individual production.
      >
      > This 2% adjustment I used in lieu of team strength-of-schedule
      > adjustments. And then I compared my individual rankings to
      yours.
      > Here are our biggest discrepancies:
      >
      > (If we assume yours are "correct", guys I have underrated:)
      >

      Once again, I doubt my ratings are any more "correct" than yours.
      They really measure different things. Think of mine as a crude
      measure of the combination between the efficiency with which a
      player performs his roles on a team and the importance of that role
      to his team's success. My W/L ratings are designed to match up to
      team W/L ratings, and so players on good teams will see some
      inflation of their values, while those on bad teams (McGrady, for
      instance) will see theirs deflate.

      > SS . player . . MG
      > 7 Wallace,Ben 43
      > 24 Finley,Michael 81
      > 3 Miller,Brad 19
      > 41 George,Devean 105
      > 8 Bryant,Kobe 26
      > 28 Mobley,Cuttino 66
      > 11 Ginobili,Emanuel 27
      > 14 Artest,Ron 31
      > 36 Nesterovic,Rados 74
      > 29 Hamilton,Richard 60
      > 6 Stojakovic,Predr 14
      > 43 Camby,Marcus 70
      > 25 Payton,Gary 49
      > 18 Billups,Chauncey 30
      > 15 Malone,Karl 21
      > 4 Kirilenko,Andrei 7
      >
      All of these players (excepting Mobley) are on teams that project
      to 50+ wins with my version of the pythagorean approximation. I've
      recently changed the method I use to calculate the W/L values this
      rating is based on, and this resulted in a significant drop in the
      ratings for Sacramento and San Antonio players. I'll explain later.

      >
      > Here are guys's I seem to be "overrating" by the same method:
      >
      > SS . player . . MG
      > 34 Iverson,Allen 11
      > 19 Randolph,Zach 6
      > 20 Kidd,Jason 9
      > 39 Gasol,Pau 17
      > 23 Pierce,Paul 12
      > 10 Brand,Elton 4
      > 2 Duncan,Tim 2
      > 9 O'Neal,Jermaine 5
      > 5 O'Neal,Shaquille 3
      >

      Some of these guys (Iverson, Randolph, Brand) are on teams that
      project to sub-.500. For the others, I'm guessing the difference is
      due to the greater focus on efficiency in Dean's ratings.


      > Pretty much all of the guys you've ranked much higher than I, play
      > for successful teams. And some of my "overranked" players play
      for
      > below-average teams.
      >

      OK, you noticed the same thing.

      > I should add to the 2nd list names such as McGrady, Vince Carter,
      > Shareef, and Marbury, all of whom land in my Top 25, but fail to
      > make the cut for your top 43.
      >

      Shareef is the only one of these 4 who rates better than a .500
      winning % (which, given how bad his team is playing, is pretty
      impressive).

      > Devean George is a good player; but he is no Tracy McGrady.
      >

      Right, but he is playing very well in his role (59% true shooting
      percentage and a good defensive rating), and he gets a surprisingly
      high % of touches on a team as loaded with superstars as the
      Lakers. A lot of players would do as well in his spot, but many
      more wouldn't.
      This comes back to the classic problem of evaluating role players
      (like George) vs. stars (like McGrady). Who is better - the role
      player performing at high efficiency, but who couldn't successfully
      carry a big load, or the star who is struggling with low efficiency,
      but whose ability to carry the load helps out his teammates (or, in
      McGrady's case, would help out his teammates if any of them could
      actually play). You could make arguments either way.

      >
      >
      > > And, here are the top rookies
      > >
      > > Player ORate PP/G %Pos DRate Stop% W L WL%
      Pro/G
      > > C. Anthony 97.0 18.8 26 99.4 49 2.4 2.3 51
      13.55
      > > C. Bosh 104.9 12.9 18 99.2 52 1.8 1.7 51
      10.66
      > > D. Wade 99.7 19.2 25 101.3 50 1.8 2.6 41
      10.24
      > > J. Howard 101.1 8.8 20 97.9 62 1.4 0.8 64
      10.21
      > > L. James 97.5 21.2 26 101.5 50 2.0 3.3 37
      10.21
      >
      >
      > Does anyone think switching Lebron and Carmello would make Denver
      > less successful or Cleveland better?

      Not me. Carmelo and Lebron shoulder a comparable burden on
      offense, and Lebron is actually slightly more efficient (though both
      are below their teams' average efficiency). Anthony has better
      teammates, especially on defense (which makes his defensive rating
      better, despite a lower stop%). I did note that Lebron's win% was
      almost exactly the same as Cleveland's projected win% - as Lebron
      goes, so do the Cavs. Anthony's win% is lower than his team's - he
      isn't as important to their success. Anthony Miller and Marcus
      Camby come across as the Nugget MVP's, so far. For whatever it's
      worth, I like Lebron better as a player. These ratings don't really
      change that, when taken in context.
    • Mike G
      ... records, ... Can we see what kind of rankings you get if you just ignore the individual losses? I tried some correlations of your win # s , but I don t
      Message 2 of 6 , Jan 4, 2004
      • 0 Attachment
        --- In APBR_analysis@yahoogroups.com, "ssims22000" <ssims2@i...>
        wrote:

        >... there are several ways of calculating individual win-loss
        records, ...

        Can we see what kind of rankings you get if you just ignore the
        individual losses? I tried some correlations of your win #'s , but
        I don't have the comprehensive list.

        For example, if McGrady is 4-7, his net of -3 makes him look like a
        weaker player than Devean George. But if McGrady played alongside 4
        hall-of-famers?

        I look for stats to be less context-dependent. I'm pretty sure we
        could build a team of guys NOT in your list that would whip 5 guys
        who are ON the list.


        > Think of mine as a crude
        > measure of the combination between the efficiency with which a
        > player performs his roles on a team and the importance of that
        role
        > to his team's success. My W/L ratings are designed to match up to
        > team W/L ratings, and so players on good teams will see some
        > inflation of their values, while those on bad teams (McGrady, for
        > instance) will see theirs deflate.

        I actually am all for this connection to team success. A team that
        wins 60 games splits those 60 wins among the players. Players on a
        20-win team have a lot less "spoils" to go around.

        But these damn Losses are what just makes no sense to me. A loss is
        zero wins. That should be bad enough.

        My bet is that by ignoring losses from your formula, you'll get a
        better ranking : one that is more consistent with minutes played,
        general perception, and common sense.


        > > Devean George is a good player; but he is no Tracy McGrady.
        > >
        >
        > Right, but he is playing very well in his role (59% true
        shooting
        > percentage and a good defensive rating), and he gets a
        surprisingly
        > high % of touches on a team as loaded with superstars as the
        > Lakers. A lot of players would do as well in his spot, but many
        > more wouldn't.

        All fair statements. But isn't a lot of his improvement this year
        attributable to his status as a 5th-option player? I mean, he's got
        to be open a Lot.

        Fisher is also shooting a lot this year; he just can't hit his shots.

        One thing I like about your win-relative rankings is that the
        Lakers' Big 4 all appear to be greatly reduced this year in my
        rankings; and much more similar to last year in yours. Consistency
        in spite of drastic context change usually is a sign of a good
        ranking system.




        > This comes back to the classic problem of evaluating role
        players
        > (like George) vs. stars (like McGrady). Who is better - the role
        > player performing at high efficiency, but who couldn't
        successfully
        > carry a big load, or the star who is struggling with low
        efficiency,
        > but whose ability to carry the load helps out his teammates (or,
        in
        > McGrady's case, would help out his teammates if any of them could
        > actually play). You could make arguments either way.
        >
        > >
        > >
        > > > And, here are the top rookies
        > > >
        > > > Player ORate PP/G %Pos DRate Stop% W L WL%
        > Pro/G
        > > > C. Anthony 97.0 18.8 26 99.4 49 2.4 2.3 51
        > 13.55
        > > > C. Bosh 104.9 12.9 18 99.2 52 1.8 1.7 51
        > 10.66
        > > > D. Wade 99.7 19.2 25 101.3 50 1.8 2.6 41
        > 10.24
        > > > J. Howard 101.1 8.8 20 97.9 62 1.4 0.8 64
        > 10.21
        > > > L. James 97.5 21.2 26 101.5 50 2.0 3.3 37
        > 10.21
        > >
        > >
        > > Does anyone think switching Lebron and Carmello would make
        Denver
        > > less successful or Cleveland better?
        >
        > Not me. Carmelo and Lebron shoulder a comparable burden on
        > offense, and Lebron is actually slightly more efficient (though
        both
        > are below their teams' average efficiency). Anthony has better
        > teammates, especially on defense (which makes his defensive rating
        > better, despite a lower stop%). I did note that Lebron's win% was
        > almost exactly the same as Cleveland's projected win% - as Lebron
        > goes, so do the Cavs. Anthony's win% is lower than his team's -
        he
        > isn't as important to their success. Anthony Miller and Marcus
        > Camby come across as the Nugget MVP's, so far. For whatever it's
        > worth, I like Lebron better as a player. These ratings don't
        really
        > change that, when taken in context.
      Your message has been successfully submitted and would be delivered to recipients shortly.