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

Re: NCAA -> NBA translation

Expand Messages
  • alleyoop2
    I ve farted around with this from time to time and never got any results that I thought were even mildly accurate. Here s what I perceive as the main
    Message 1 of 20 , Apr 10, 2002
    • 0 Attachment
      I've farted around with this from time to time and never got any
      results that I thought were even mildly accurate. Here's what I
      perceive as the main obstacles:

      1) Adjusting for game pace

      2) Adjusting for strength of schedule

      3) The fact that the rules are different. Here's one example: Let's
      say there's a guy named "Allen" who can create his own shot whenever
      he wants but only makes 43% of his tries. In college, you have 45
      seconds to get a better shot, so this guy isn't nearly as valuable as
      he will be in the pros.

      4) Adjusting for the closer three-point line. Not sure how you
      differentiate who has NBA range from who doesn't.

      5) One of the first questions NBA guys ask when they look at a player
      is "Who can he guard?" - that's pretty much left out of their
      statistics.

      6) Adjusting for the quality of the players own team. This one kills
      me. It seems to me that role players on top-level teams can have
      similar stats and wildly divergent results in the NBA. For one
      example that I worked with, look at the stats for Andre Hutson and
      Richard Jefferson last year. Both were role players on top-level
      teams. Hutson's numbers are in many ways more eye-popping than
      Jefferson's. Yet Jefferson is a key player on one of the league's
      best teams this year; Hutson bags groceries. Chris Wilcox is going to
      be another one; he was the number three weapon on his team this year
      yet has NBA power forward written all over him. Is there a way to
      capture that type of thing statistically?





      --- In APBR_analysis@y..., "HoopStudies" <deano@r...> wrote:
      > --- In APBR_analysis@y..., Ed Weiland <weiland1029@y...> wrote:
      > > Superstar teammates. Vince Carter played second fiddle
      > > to Antawn Jamison at Carolina. His numbers were OK,
      > > but hardly eye-popping, other than the .591 FG pct.
      > > his junior year. Vince gets to the pros and he's a
      > > superstar.
      >
      > Ed did a good job listing factors. This superstar teammate factor
      is
      > one that I want to understand using those curves I put out many
      > emails ago. I think there are shifts in the curves we can identify
      > based on context. I think there is also a basic shift just going
      up
      > a level. Carter was extremely efficient in college at a fairly
      high,
      > but not overly high, number of possessions per minute. Jordan was
      > the same. Both of these guys were tall for the 2G slot, so they
      > wouldn't then see the kind of decline a 6'3" 2G like David Sanders
      > (Ole Miss) will see, who is similarly efficient.
      >
      > Big men like the 7-footers do not have the height problem since
      they
      > join the league without serious changes in their opposition, except
      > in strength, which we account for with the NCAA-NBA shift, I think.
      >
      > This is a worthy project. Anyone want to add to the measurable or
      > possibly measurable factors we should consider?
      >
      > Dean Oliver
    • harlanzo
      I think the baseball translations work a little better than a basketball one would. The important thing to recognize is that translations in baseball seem to
      Message 2 of 20 , Apr 10, 2002
      • 0 Attachment
        I think the baseball translations work a little better than a
        basketball one would. The important thing to recognize is that
        translations in baseball seem to vary by league. Similarly in
        basketball, a 20 ppg scorer in the ivy league does not equal to one
        in the ACC. So I am not sure how to quantify this difference. The
        only two ways to do this is by comparing the success of players from
        these conferences and run something proportionally. Of course this
        is problematic because so few players from the inferior conferences
        make the NBA that the sample size is too small to make a meaningful
        projection. You might generalize between confernences of similar
        ability (ie super conferences, mid-size conferences, and small
        timers) and thus create a bigger pool for comparison. (You could
        also only look at stats of small time players when they play big
        schools).

        on top of this issue, is the fact that most stats are dependents on
        systems and opportunity in basketball much more so than in baseball
        which could also skew comparisons.



        --- In APBR_analysis@y..., "HoopStudies" <deano@r...> wrote:
        > --- In APBR_analysis@y..., Ed Weiland <weiland1029@y...> wrote:
        > > Superstar teammates. Vince Carter played second fiddle
        > > to Antawn Jamison at Carolina. His numbers were OK,
        > > but hardly eye-popping, other than the .591 FG pct.
        > > his junior year. Vince gets to the pros and he's a
        > > superstar.
        >
        > Ed did a good job listing factors. This superstar teammate factor
        is
        > one that I want to understand using those curves I put out many
        > emails ago. I think there are shifts in the curves we can identify
        > based on context. I think there is also a basic shift just going
        up
        > a level. Carter was extremely efficient in college at a fairly
        high,
        > but not overly high, number of possessions per minute. Jordan was
        > the same. Both of these guys were tall for the 2G slot, so they
        > wouldn't then see the kind of decline a 6'3" 2G like David Sanders
        > (Ole Miss) will see, who is similarly efficient.
        >
        > Big men like the 7-footers do not have the height problem since
        they
        > join the league without serious changes in their opposition, except
        > in strength, which we account for with the NCAA-NBA shift, I think.
        >
        > This is a worthy project. Anyone want to add to the measurable or
        > possibly measurable factors we should consider?
        >
        > Dean Oliver
      • thedawgsareout
        ... Yes, the baseball translations are not just by level, but by league. The Pacific Coast League, for example, is AAA, as is the International League, but the
        Message 3 of 20 , Apr 11, 2002
        • 0 Attachment
          --- In APBR_analysis@y..., "harlanzo" <harlanzo@y...> wrote:
          > I think the baseball translations work a little better than a
          > basketball one would. The important thing to recognize is that
          > translations in baseball seem to vary by league. Similarly in
          > basketball, a 20 ppg scorer in the ivy league does not equal to one
          > in the ACC. So I am not sure how to quantify this difference.

          Yes, the baseball translations are not just by level, but by league.
          The Pacific Coast League, for example, is AAA, as is the
          International League, but the IL is considered a higher-quality
          league. The way these translations work is with a MLE -- Minor League
          Equivalency, developed, like everything else, by Bill James.

          The first of many problems we find in trying to do the same for the
          NCAA is that even within the leagues themselves, teams are playing
          widely variant schedules. About 40% of games are going to be non-
          conference ones, and assuredly Oregon State isn't playing the same
          non-conference teams as Arizona.

          Team quality also isn't a problem in baseball thanks to the
          development of team-independent stats. As hard as we may try -- and
          Dean's column about the effect of Jerry Stackhouse on the Pistons was
          quite interesting -- it's still tough to de-context a player's stats,
          and I imagine doubly so for NCAA players.

          I think what is a more reasonable goal for the time being might be to
          try to look at what *types* of players make for good pros. It's been
          mentioned that both Carter and Jordan were high-efficiency guys on
          very good teams. Is this type of player particularly successful in
          making the transition? Basically, I guess this gets back to the four
          types of players (high-efficiency, low productivity . . .) that were
          discussed way back in January. Is any specific group translating to
          the pro game better than the others? Why?
        • mikel_ind
          ... Draw a DNA sample, and look for the work ethic gene. Mike Goodman
          Message 4 of 20 , Apr 11, 2002
          • 0 Attachment
            --- In APBR_analysis@y..., "thedawgsareout" <kpelton08@h...> wrote:
            >.... try to look at what *types* of players make for good pros.


            Draw a DNA sample, and look for the "work ethic" gene.


            Mike Goodman
          • HoopStudies
            ... These are not big hurdles. Game pace is definitely not too bad. Strength of schedule can be accounted for using Sagarin s number or Massey s number or...
            Message 5 of 20 , Apr 11, 2002
            • 0 Attachment
              --- In APBR_analysis@y..., "alleyoop2" <alleyoop2@y...> wrote:
              > I've farted around with this from time to time and never got any
              > results that I thought were even mildly accurate. Here's what I
              > perceive as the main obstacles:
              >
              > 1) Adjusting for game pace
              >
              > 2) Adjusting for strength of schedule
              >

              These are not big hurdles. Game pace is definitely not too bad.
              Strength of schedule can be accounted for using Sagarin's number or
              Massey's number or... I frankly don't perfectly trust strength of
              schedule numbers, but they are a start and there are bigger hurdles
              to leap.

              > 3) The fact that the rules are different. Here's one example: Let's
              > say there's a guy named "Allen" who can create his own shot
              whenever
              > he wants but only makes 43% of his tries. In college, you have 45
              > seconds to get a better shot, so this guy isn't nearly as valuable
              as
              > he will be in the pros.
              >

              Creating your own shot without loss of efficiency is more valuable in
              the NBA. I'm working on this.

              Any other examples of critical rule differences?

              > 4) Adjusting for the closer three-point line. Not sure how you
              > differentiate who has NBA range from who doesn't.
              >

              I'd start off with a straight reduction in percentage. Also have the
              adjustments mentioned before for height.

              > 5) One of the first questions NBA guys ask when they look at a
              player
              > is "Who can he guard?" - that's pretty much left out of their
              > statistics.
              >

              Yeah, this is a big one. This one was really important in the
              late '80's and into the '90's when "athletes" were seen as more
              valuable than basketball players. That is changing a bit with the
              success of the nonathletic European basketball players (thank
              goodness). But it is still important. We have no fundamental
              measure of defensive quickness or good hands or defensive desire.
              Defense in general is tough and it is a big factor in determining
              playing time in the NBA.

              > 6) Adjusting for the quality of the players own team. This one
              kills
              > me. It seems to me that role players on top-level teams can have
              > similar stats and wildly divergent results in the NBA. For one
              > example that I worked with, look at the stats for Andre Hutson and
              > Richard Jefferson last year. Both were role players on top-level
              > teams. Hutson's numbers are in many ways more eye-popping than
              > Jefferson's. Yet Jefferson is a key player on one of the league's
              > best teams this year; Hutson bags groceries. Chris Wilcox is going
              to
              > be another one; he was the number three weapon on his team this
              year
              > yet has NBA power forward written all over him. Is there a way to
              > capture that type of thing statistically?

              The fact that Hutson really isn't getting a chance makes this
              comparison difficult, I think. Scouts look FIRST at whether they
              have an NBA body and NBA athleticism, then they look at their
              skills. It's mainly because that is what they see first and what
              they are trained to see first. You are supposed to go down to the
              floor and get a sense of size, strength, jumping ability. For some
              reason, there is this belief that you can teach the basketball skills.

              Still, assuming the scouts were right and that Hutson doesn't have
              NBA skills, it appears to be due to him being undersized (maybe not
              quick enough) for the skills he exhibited. There is some mismatch
              between height (measurable), strength (possibly measurable),
              quickness (uhhh), and style of game (uh-oh). Jefferson had the
              height, quickness, and style of game to go to the next level. Hutson
              didn't meet these requirements (apparently), even if he had the
              stats.

              I do think that boxscores are going to be critical to doing better in
              all of this. That's not going to be a fun task to work on...almost
              as little fun as collecting the DNA that MikeG wants. Although with
              all the paternity suits floating around, maybe getting DNA won't be
              so bad.

              So, does anyone have good college stats we can all look at? They are
              pretty hard to get, actually, in any consistent format.


              DeanO
            • Ed Weiland
              ... Hutson was listed at 6 8 240. Hardly tiny, but not exactly what you want in your PF either. Especially since Hutson is the unathletic grinder type. He
              Message 6 of 20 , Apr 12, 2002
              • 0 Attachment
                --- HoopStudies <deano@...> wrote:
                > >
                > Still, assuming the scouts were right and that
                > Hutson doesn't have
                > NBA skills, it appears to be due to him being
                > undersized (maybe not
                > quick enough) for the skills he exhibited. There is
                > some mismatch
                > between height (measurable), strength (possibly
                > measurable),
                > quickness (uhhh), and style of game (uh-oh).
                > Jefferson had the
                > height, quickness, and style of game to go to the
                > next level. Hutson
                > didn't meet these requirements (apparently), even if
                > he had the
                > stats.

                Hutson was listed at 6'8 240. Hardly tiny, but not
                exactly what you want in your PF either. Especially
                since Hutson is the unathletic grinder type. He played
                in Europe this past season and I think Milwaukee (the
                team that drafted Hutson in the 2nd round) has plans
                for him. We may get to see him yet.
                >
                > I do think that boxscores are going to be critical
                > to doing better in
                > all of this. That's not going to be a fun task to
                > work on...almost
                > as little fun as collecting the DNA that MikeG
                > wants. Although with
                > all the paternity suits floating around, maybe
                > getting DNA won't be
                > so bad.
                >
                > So, does anyone have good college stats we can all
                > look at? They are
                > pretty hard to get, actually, in any consistent
                > format.

                The Usenet draft page :

                http://www.ibiblio.org/craig/draft/usenet.html

                This page lists stats on prospects going back to the
                '94 draft. They have the complete stats, including
                turnovers. That might be a start. I'm not sure they
                have all the players listed though.

                Ed Weiland




                __________________________________________________
                Do You Yahoo!?
                Yahoo! Tax Center - online filing with TurboTax
                http://taxes.yahoo.com/
              Your message has been successfully submitted and would be delivered to recipients shortly.