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

1922Re: similarity (was Consistency)

Expand Messages
  • Mike G
    Mar 16, 2003
    • 0 Attachment
      --- In APBR_analysis@yahoogroups.com, "Michael Tamada" <tamada@o...>
      wrote:
      >...> I don't recall if MikeG's Euclidean distance measures include
      FGMiss; my impression is that they do not because he doesn't seem to
      include them when he lists players' similar stats.

      Shooting % is pre-included in the 'standardized' pts-per-36-min.
      rate. If Bob Love scores 24 ppg shooting .435, and Artis Gilmore
      scores 16 ppg shooting .605, they might both be rated as 20 pt. guys.

      Actually, I use not a raw FG% but a combined shooting % that
      includes FT.

      Giving credit for greater or lesser scoring based on scoring % is
      obviously questionable on many levels. I do it not so much as an
      attempt to 'punish' or 'reward' players, but to 'project' how they'd
      score in an 'average' situation.

      In other words, on average, high-% guys 'should' shoot more; low-%
      guys should shoot less.


      > If I'm comparing various trips and routes, it simply is not
      adequate to only look at distance travelled and time used; it's also
      important to look at distance/time, i.e. speed. And for basketball,
      it's not enough to look at FGM and FGMiss, we have to look at FG%
      too.

      This is amazing. I just had this conversation yesterday driving
      back from Wisconsin. Once I had a mother-in-law who was insistant
      on driving the route which provided the highest speed. Sometimes
      I'd be tempted to mention that while we were indeed travelling fast,
      we were going in the wrong direction.

      Mike T here raises speed as an issue, but I can't tell which way he
      leans here. Is it better to drive faster and arrive at the same
      time? Is fuel economy also an issue? Safety?


      > I suspect the reason that many people are hesitant or resistant is
      because FG% is measured in units which are on a totally different
      scale from FGM, rebounds, etc....

      As mentioned, shooting % can be used (along with other factors)
      to 'scale' the scoring rate that is used.


      >...multivariate
      statistical methods: principal components analysis, factor
      analysis, cluster analysis, etc.
      .... Discriminant analysis, probit
      or logistic (aka logit) regression, or even regular multivariate
      regression....

      This sounds like voodoo to me, and might be hard to translate to
      the casual fan.


      can be used to easily come up with weights which can be
      used to measure Hall of Fame probability

      Now you're talking. But I think you have to be able to factor in
      weird and (to me) irrelevant things like whether a guy played in New
      York, and other 'popularity' issues.


      > Similarity scores however have a variety of potentially important
      and useful applications, but I do not find the similarity measures
      that I've seen to be very convincing. E.g. MikeG's list showing
      Steve Hawes to be the 4th most similar player to Charles Oakley.
      His list also included the likes of Curtis Perry, Horace Grant, AC
      Green, etc. -- all instantly recognizable as kindred of the PF
      banger Oakley. But Hawes was an spot-up-shooting (79% on FTs -- few
      non-Malone PFs achieve that), mediocre-rebounding center, nothing
      like the Oakleys, Silases, Grants, Greens, etc. on that list.

      Here's part of that list, for reference. Per-36 rates.

      .00 Charles Oakley 11.3 11.5 2.8 1.2 .3
      .28 Paul Silas 10.5 10.6 2.5 .8 .3
      .40 Wes Unseld 10.4 12.5 3.8 1.1 .7
      .44 Steve Hawes 11.5 9.1 2.7 1.1 .6
      .46 Jim Fox 12.6 9.4 2.2 .8 .4
      .51 Anthony Mason 13.0 9.4 3.7 .8 .3
      .52 Curtis Perry 11.4 10.1 2.3 1.3 1.0
      .58 David Greenwood 11.6 10.0 2.2 .9 1.1
      .61 Horace Grant 12.7 9.5 2.4 1.1 1.1
      .61 Grant Long 11.9 8.1 2.1 1.5 .5
      .64 Clifford Ray 10.0 11.1 2.9 1.1 1.4
      .65 Kenny Carr 14.3 10.1 1.7 1.0 .6
      .65 Bill Laimbeer 13.9 11.0 2.1 .8 1.0

      what we have here are PF who do not score a lot, but pass a bit.
      And centers who do not score a lot nor block a lot of shots, but
      pass a bit.

      What I don't attempt to include are 'stylistic' elements of scoring,
      like whether points were gotten by outside or inside moves, FT or 3-
      pters, overhand or underhand, etc. There has to be a point we stop
      comparing; so I stand by my 'points is points' doctrine.

      Maybe Oakley would have been pressed into service as a center, had
      he played with a '70s expansion team. Maybe Hawes would have been
      primarily a PF alongside a dominant center. I don't think their
      stats would be a lot different, and they'd still look 'similar' in
      this list.

      This spreadsheet does equate players of different eras and
      positions. I find it rather interesting that Hawes and Oak are
      similar statistically. Is it a shortcoming of the program? Depends
      what you are looking for.
    • Show all 11 messages in this topic