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

Oscar's Effect -- getting technical

Expand Messages
  • Dean Oliver
    Over in APBR, we ve been discussing the impact Oscar Robertson had on his teammates effectiveness. Ed did a nice little thing on how his teammates did with
    Message 1 of 5 , Oct 3, 2003
    View Source
    • 0 Attachment
      Over in APBR, we've been discussing the impact Oscar Robertson had on
      his teammates' effectiveness. Ed did a nice little thing on how his
      teammates did with and without Oscar. I think it's probably very
      illustrative... "probably" only because I want to rule out one
      possible bias and I want to identify how significant it is. I move
      it over here to try to address that.

      Below are some of Ed's stats. The potential bias is that league
      averages may have been changing in a coincidental fashion to yield
      this result. I think it's unlikely because of the number of players
      over a fairly long period of time, but worthy of checking.

      >
      > Player G_W G_WO Mpg_W Mpg_WO FG%_W
      FG%_WO Improvement
      >
      > Kareem Abdul-jabbar 320 228 42.7 43.1 56.2%
      52.4% -
      > Lucius Allen 293 223 29.2 29.4 48.8%
      44.6% -
      > Jerry Lucas 461 147 43.2 38.6 49.7%
      50.2% +
      > Tom Vanarsdale 148 311 37.9 30.2 44.7%
      43.4% -
      > Bob Dandridge 303 234 36.9 35.2 49.5%
      48.6% -
      > Jack Twyman 460 147 29.8 39.0 47.0%
      42.1% -
      > Arlen Bockhorn 328 146 31.8 29.8 40.8%
      39.0% -
      > Edward Davis 147 120 14.0 12.4 47.1%
      44.5% -
      > Connie Dierking 373 198 27.6 16.8 43.4%
      38.9% -
      > Terry Driscoll 124 120 13.4 13.5 44.4%
      40.2% -
      > Wayne Embry 464 289 31.0 17.3 45.4%
      41.0% -
      > Fred Foster 129 141 20.0 18.5 43.9%
      41.1% -
      > Happy Hairston 286 151 26.2 35.2 47.2%
      48.1% +
      > Tom Hawkins 310 311 24.1 25.8 44.5%
      45.5% +
      > Bob Love 138 131 15.5 28.0 42.7%
      45.5% +
      > Jon Mcglocklin 453 112 24.3 19.5 49.3%
      48.3% -
      > Curtis Perry 223 168 28.3 30.6 43.1%
      48.5% +
      > Hub Reed 235 244 16.9 16.0 44.6%
      42.5% -
      > Larry Staverman 126 106 12.8 10.9 45.3%
      47.0% +
      > Walt Wesley 212 154 14.9 24.9 44.4%
      44.2% -
      > Dick Cunningham 158 139 8.9 11.9 41.5%
      40.8% -
      >
      >
      > Total 5691 3820 27.9 25.8 47.4%
      45.5%

      The shooting percentages being an overall 2% different is
      interesting. That is a huge competitive advantage. What I'd like to
      see are the raw totals of FG and FGA for the individual players (and
      total) to see how many of these splits are statistically significant
      at 90 or 95%. I believe they probably are as having only 6 of 21
      guys improve without Oscar is significantly different from a null
      hypothesis.

      For context, we might want to look at the effect of a couple other
      notable playmakers, like Magic and Stockton and Tiny and Isiah.
      Maybe even look at some that aren't so notable, like Dennis Johnson,
      Derek Harper, Terry Porter, Kevin Porter, etc.

      DeanO


      --- In APBR@yahoogroups.com, igor eduardo küpfer <igorkupfer@r...>
      wrote:
      >
      > ----- Original Message -----
      > From: Dean Oliver
      > To: APBR@yahoogroups.com
      > Sent: Thursday, October 02, 2003 11:21 AM
      > Subject: [APBR] Re: criteria for assists in the 60s
      >
      >
      >
      > A couple good responses on this already.
      >
      > Oscar has been saying this for more than 20 years now. I've
      never
      > been quite sure why anyone feels such a need to emphasize these
      > things, especially him. He is among the tremendous elite in the
      game
      > without adjusting the numbers one bit. No one will argue that.
      > Giving him a couple extra assists isn't going to change the
      position
      > that he is still great. If some current statistician went back
      and
      > rescored all his assists from back then, would it fundamentally
      > change how people view him? Doubtful. If he actually averaged
      30
      > assists a game, then maybe.
      >
      > Taking away Oscar, how well did his teams shoot relative to the
      > league average by year? If, by the definition, it should help a
      > shooter get a shot, you'd like to see that his teammates actually
      > shot better than league average. (I'm at work and can't do the
      calc
      > now.)
      >
      >
      > It looks to me like players shot significantly worse without Oscar.
      I made a list of 21 players who played a significant number of games
      with OR and also a significant number of games in the 2 seasons
      adjacent to their time with Oscar. It seems that of the 21 players,
      15 of them had a better FG% with OR. (The "_W" means "With Oscar"
      and "_WO" means in the 2 seasons just preceding or following the
      seasons playing with Oscar.)
      >
      > Player G_W G_WO Mpg_W Mpg_WO FG%_W
      FG%_WO Improvement
      >
      > Kareem Abdul-jabbar 320 228 42.7 43.1 56.2%
      52.4% -
      > Lucius Allen 293 223 29.2 29.4 48.8%
      44.6% -
      > Jerry Lucas 461 147 43.2 38.6 49.7%
      50.2% +
      > Tom Vanarsdale 148 311 37.9 30.2 44.7%
      43.4% -
      > Bob Dandridge 303 234 36.9 35.2 49.5%
      48.6% -
      > Jack Twyman 460 147 29.8 39.0 47.0%
      42.1% -
      > Arlen Bockhorn 328 146 31.8 29.8 40.8%
      39.0% -
      > Edward Davis 147 120 14.0 12.4 47.1%
      44.5% -
      > Connie Dierking 373 198 27.6 16.8 43.4%
      38.9% -
      > Terry Driscoll 124 120 13.4 13.5 44.4%
      40.2% -
      > Wayne Embry 464 289 31.0 17.3 45.4%
      41.0% -
      > Fred Foster 129 141 20.0 18.5 43.9%
      41.1% -
      > Happy Hairston 286 151 26.2 35.2 47.2%
      48.1% +
      > Tom Hawkins 310 311 24.1 25.8 44.5%
      45.5% +
      > Bob Love 138 131 15.5 28.0 42.7%
      45.5% +
      > Jon Mcglocklin 453 112 24.3 19.5 49.3%
      48.3% -
      > Curtis Perry 223 168 28.3 30.6 43.1%
      48.5% +
      > Hub Reed 235 244 16.9 16.0 44.6%
      42.5% -
      > Larry Staverman 126 106 12.8 10.9 45.3%
      47.0% +
      > Walt Wesley 212 154 14.9 24.9 44.4%
      44.2% -
      > Dick Cunningham 158 139 8.9 11.9 41.5%
      40.8% -
      >
      >
      > Total 5691 3820 27.9 25.8 47.4%
      45.5%
    • igor eduardo küpfer
      ... From: Dean Oliver To: APBR_analysis@yahoogroups.com Sent: Friday, October 03, 2003 11:45 AM Subject: [APBR_analysis] Oscar s Effect -- getting technical
      Message 2 of 5 , Oct 3, 2003
      View Source
      • 0 Attachment
         
        ----- Original Message -----
        Sent: Friday, October 03, 2003 11:45 AM
        Subject: [APBR_analysis] Oscar's Effect -- getting technical

        Over in APBR, we've been discussing the impact Oscar Robertson had on
        his teammates' effectiveness.  Ed did a nice little thing on how his
        teammates did with and without Oscar.  I think it's probably very
        illustrative... "probably" only because I want to rule out one
        possible bias and I want to identify how significant it is.  I move
        it over here to try to address that.

        Below are some of Ed's stats.  The potential bias is that league
        averages may have been changing in a coincidental fashion to yield
        this result.  I think it's unlikely because of the number of players
        over a fairly long period of time, but worthy of checking. 
         
        <snip>

        The shooting percentages being an overall 2% different is
        interesting.  That is a huge competitive advantage.  What I'd like to
        see are the raw totals of FG and FGA for the individual players (and
        total) to see how many of these splits are statistically significant
        at 90 or 95%.  I believe they probably are as having only 6 of 21
        guys improve without Oscar is significantly different from a null
        hypothesis.
         
        Unfortunately I don't have time now for commentary, but I did manage to generate significance levels. For your perusal:
         
        Player                FGM_w    FGA_w    FGM_wo   FGA_wo   Improvement    
        Kareem Abdul-jabbar   4152     7393     2698     5153      3.8%**   p = .0000
        Lucius Allen          1692     1772     1278     1588      4.3%**   p = .0008
        Jerry Lucas           3632     3672     1028     1021     -0.4%     p = .7264
        Tom Vanarsdale        1167     1442     1857     2422      1.3%     p = .2820
        Bob Dandridge         2445     2497     1775     1876      0.9%     p = .4449
        Jack Twyman           3326     3755     1580     2174      4.9%**   p = .0000
        Arlen Bockhorn        1628     2363      617      966      1.8%     p = .2146
        Edward Davis           321      361      229      286      2.6%     p = .3801
        Connie Dierking       2036     2654      534      840      4.5%*    p = .0029
        Terry Driscoll         228      286      175      260      4.1%     p = .2108
        Wayne Embry           2696     3243      915     1319      4.4%**   p = .0003
        Fred Foster            535      684      495      710      2.8%     p = .1627
        Happy Hairston        1471     1643     1013     1091     -0.9%     p = .5342
        Tom Hawkins           1048     1309     1263     1511     -1.1%     p = .4473
        Bob Love               366      492      748      897     -2.8%*    p = .0228
        Jon Mcglocklin        1998     2056      386      413      1.0%     p = .6149
        Curtis Perry           771     1019      844      897     -5.4%**   p = .0013
        Hub Reed               558      693      523      707      2.1%     p = .3114
        Larry Staverman        209      252      171      193     -1.6%     p = .6731
        Walt Wesley            564      707      835     1054      0.2%     p = .9418
        Dick Cunningham        148      209      193      280      0.7%     p = .8867
        * Significant at .05      ** Significant at .01
         
        ed
         
      • John Hollinger
        The 2% difference is interesting, but I wonder how much of it was because of the fact that Oscar was taking all the shots. So the other guys were taking fewer
        Message 3 of 5 , Oct 3, 2003
        View Source
        • 0 Attachment
          The 2% difference is interesting, but I wonder how much of it was
          because of the fact that Oscar was taking all the shots. So the other
          guys were taking fewer shots, but because they had to create fewer
          for themselves, they ended up shooting a higher percentage. Kind of
          like what will happen to Gary Payton and Karl Malone this year. If
          that's the case, you should see the same effect with a lot of high-
          scoring players.

          Let's call this the "Ainge Effect" (side note: this theory is still
          in beta, which is one reason I'm testing it out on this audience).
          Danny Ainge averaged 15.7 points a game for title-contending Boston
          while shooting 49 percent. He gets traded to crappy Sacramento and
          suddenly he has to creat a lot more shots for himself. As a result,
          his average goes up to 17.5 and 17.9, but his percentage drops to
          45.7 and 43.8. Then he goes to title-contending Portland and can be a
          complementary player again; his average drops back down to 11.0, but
          he shoots 47.2 percent.

          Thoughts?
        • Mike G
          ... other ... Using a slightly different sample of players, I found a significantly smaller increase in shooting %, with/without Oscar. I used the combined
          Message 4 of 5 , Oct 4, 2003
          View Source
          • 0 Attachment
            --- In APBR_analysis@yahoogroups.com, "John Hollinger"
            <alleyoop2@y...> wrote:
            > The 2% difference is interesting, but I wonder how much of it was
            > because of the fact that Oscar was taking all the shots. So the
            other
            > guys were taking fewer shots, but because they had to create fewer
            > for themselves, they ended up shooting a higher percentage...

            Using a slightly different sample of players, I found a
            significantly smaller increase in shooting %, with/without Oscar.

            I used the combined shooting %, rather than raw FG%. Maybe one
            difference is that with Oscar, guys were spotting up and shooting
            (thus a higher FG%); and without Oscar, guys were 'creating',
            driving, getting to the foul line (thus a better 'scoring
            efficiency').

            In either case, I found no evidence that Oscar cut into anyone's
            stats, and seems to have upped the scoring production. I haven't
            broken it down into his Royals career (dominator), and his Bucks
            career (distributor).


            > Kind of
            > like what will happen to Gary Payton and Karl Malone this year. If
            > that's the case, you should see the same effect with a lot of high-
            > scoring players.

            Are you predicting less scoring but higher efficiencies from these
            guys? It will be interesting, to say the least.

            I wonder about the 'Effect' that a lot of dominating players have on
            various teammates. Not just point guards (though that's essentially
            their job), but other guys who demand the ball a lot.

            One could further break it down by position. Does Karl Malone's
            presence enhance the stats of other forwards, while eating into the
            FGA of shooting guards? etc.

            >
            > Let's call this the "Ainge Effect" (side note: this theory is
            still
            > in beta, which is one reason I'm testing it out on this audience).
            > Danny Ainge averaged 15.7 points a game for title-contending
            Boston
            > while shooting 49 percent. He gets traded to crappy Sacramento and
            > suddenly he has to creat a lot more shots for himself. As a
            result,
            > his average goes up to 17.5 and 17.9, but his percentage drops to
            > 45.7 and 43.8. Then he goes to title-contending Portland and can
            be a
            > complementary player again; his average drops back down to 11.0,
            but
            > he shoots 47.2 percent.

            There's abundant evidence that this happens. There might even be a
            formula that tends to equalize these factors. Multiplying shooting
            % by points-per-minute tends to even out these changes-in-scoring-
            load produced by changes-in-roles.
            >
            > Thoughts?
          • Dean Oliver
            ... other ... of ... still ... Boston ... result, ... be a ... but ... There is no doubt this happens. Those charts that I put out a year or more ago -- those
            Message 5 of 5 , Oct 5, 2003
            View Source
            • 0 Attachment
              --- In APBR_analysis@yahoogroups.com, "John Hollinger"
              <alleyoop2@y...> wrote:
              > The 2% difference is interesting, but I wonder how much of it was
              > because of the fact that Oscar was taking all the shots. So the
              other
              > guys were taking fewer shots, but because they had to create fewer
              > for themselves, they ended up shooting a higher percentage. Kind
              of
              > like what will happen to Gary Payton and Karl Malone this year. If
              > that's the case, you should see the same effect with a lot of high-
              > scoring players.
              >
              > Let's call this the "Ainge Effect" (side note: this theory is
              still
              > in beta, which is one reason I'm testing it out on this audience).
              > Danny Ainge averaged 15.7 points a game for title-contending
              Boston
              > while shooting 49 percent. He gets traded to crappy Sacramento and
              > suddenly he has to creat a lot more shots for himself. As a
              result,
              > his average goes up to 17.5 and 17.9, but his percentage drops to
              > 45.7 and 43.8. Then he goes to title-contending Portland and can
              be a
              > complementary player again; his average drops back down to 11.0,
              but
              > he shoots 47.2 percent.
              >
              > Thoughts?

              There is no doubt this happens. Those charts that I put out a year
              or more ago -- those were designed to look exactly at this in a
              theoretical way because I had seen it so often empirically. Iverson
              improves his teammates pretty significantly even though he isn't
              efficient, just because he's using 30+% of his team's possessions.

              So, with the Lakers, if Phil can get them all cutting down their
              possession use to 23-27% (rather than 27-32%), they should be a
              dynamite offense.

              DeanO
            Your message has been successfully submitted and would be delivered to recipients shortly.