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

298Re: [APBR_analysis] Tendex rebuttal

Expand Messages
  • Michael K. Tamada
    Nov 3 6:33 AM
      On Sat, 3 Nov 2001 bchaikin@... wrote:

      [...]

      > williams slightly more power, much better OB%), williams was at best an
      > average outfielder defensively and dimaggio was considered not only the best
      > defensive outfielder of his time but one of the all time best. that tells me
      > he was the better player, without question...

      I'm getting off topic here but I don't think that's a view that most
      people, in particular most baseball analysts, have. Dimaggio had better
      defense, no question. But the Splendid Splinter was the greatest hitter
      who ever lived and who wasn't named Babe Ruth. His offense was way way
      better than Dimaggio's, more than could be made up for by an occasional
      great catch or throw.

      I do agree with your point about the inherent usefulness of combining
      offensive and defensive attributes (assuming that good measures of defense
      can be found, which is rare).

      > . a TO and ST have opposite values, one loses possession of the ball and one

      Basically true. Although I have a suspicion that ST's lead to more
      fastbreaks and more points than ordinary TO's. The stats that Dean O and
      others have been talking about collecting could solve this once and for
      all. But it's clearly a good approximation to say that a ST = -TO.

      > gains it. a missed FG is the same as a TO as you lose possession of the ball.

      This one however is clearly false. A TO means that you've lost posession,
      period. And you're correct that that costs your team about 1 point in
      expected value.

      But a missed FG is not as bad as a TO. Because your team still has a
      chance to retain posession, by getting an OR. This is only about a 1/3
      chance (less than that in recent years), but it's still a chance. The
      other team doesn't have posession yet (unlike the case with a TO). So the
      missed FG has a value of -2/3 (a bit more in recent years), not -1.

      > if you get the ball back with an OFF REB you are gaining possession of the

      Yep, the OR cancels the missed FG, so you're back where you started (*),
      so the OR has a value of about +2/3 (a bit more in recent years).

      (*) Actually, you're probably better off than you were before, because
      your chance of scoring directly after grabbing an OR is probably higher
      than your chance of scoring on an ordinary possession. How much better
      off -- that's where we need the more detailed stats.

      > ball. if you get the ball with a DEF REB that is a new possession. a BS is

      Well the DR is only part of the story. The main part was getting the
      opponent to miss the shot, that was -2/3 for him (or +2/3 for you).
      Grabbing the defensive rebound finishes the defensive play, but therefore
      that's only a +1/3 for you.

      Summarizing: if creating a turnover has a value of +1, which is indeed a
      good approximation, then the combination of causing a missed FGA, and then
      grabbing the DR, must have the same value, because they lead to
      equivalent results: gain of possession. So the causing the missed FGA
      and grabbing the DR, combined, have a value of +1. And most of the credit
      goes to the missed FG.

      > the same as a missed FG. a missed FT is definitely a missed opportunity for

      Treating a BS the same as a missed FG may be a decent approximation, but
      they are not the same event. The most important difference of course is
      that box scores give credit to players who block shots. They do not give
      credit to players who forced their opponent to miss a FG (granted, there
      is the question of how much credit they should get, maybe they opponent
      missed the shot on his own accord, even FT shooters will do this 30% of
      the time).

      But the other important difference is that players who block a lot of
      shots are usually creating intimidation or altering other shots, and these
      will not show up in the box score. So it probably is proper to give the
      shot blocker some extra credit, beyond the credit given for forcing a
      missed shot.

      On the other hand, to give no credit at all for the missed shots might be
      going too far. Again this is where Dean O's defensive box scores may
      improve things.

      [...]

      > prove to you that since 1977-78 the value of a ball possession on average is
      > equal to approximately one point. again this is all inherent in the game of
      > NBA basketball...

      The 1 point approximation is indeed a good one. However it is not
      "inherent", it's due to the approximate 1-point-per-possession scoring
      rate that the NBA has. Back in the 1940s and 1950s, the scoring rate was
      a lot lower, and possessions were worth a lot less than 1 point. So it's
      not inherent, it's just a coincidence that possession happens to be worth
      about a point. Due to teams' FG%, 3pt FG%, FT%, and TO% leading to about
      a point per possession.

      > if the league average FG% was 70% and not 50%, then yes it would be

      [...]

      > with defense. fouls result in - on average - one point being scored for the
      > opposition, and more fouling results in the bonus situation and can cost your
      > team on average MORE than one point per foul on average in a game....
      >
      > thus your statement of "...fouling your opponent on average does not increase
      > his odds of scoring..." is at face value true, but more fouling results in
      > more FTA in the bonus situation, and thus that average of one point per foul
      > increases drastically...

      Some of the reasoning here is sound -- PFs are indeed negatives, not
      neutral. And for pretty much the reason that you say: they either lead
      directly to FTs, with their high scoring opportunities (except for Shaq
      and Polynice), or they put the opponent closer to reaching those bonus
      opportunities.

      But even if you're already in the penalty, fouling a player and sending
      him to the line for two FTAs is not costing your team a point (or two
      points), not relative to where your team was already.

      Remember, possession is worth about a point. Because (to state it
      oversimplistically) teams which have the ball usually end up shooting a
      shot with about a 50% chance of going in.

      If you foul a 70% shooter and send him to the line, he'll probably come
      away with 1.4 points.

      Definitely not a good thing for your team. But hey, the other team on
      average was going to score 1 point ANYWAY. If you hadn't fouled the guy,
      he or one of his teammates would likely have put up a shot with a 50%
      chance of going in.

      So your foul has only cost your team .4 points. Yes, you truly gave up
      1.4 points, but your team was, on average, going to give up 1 point
      anyway.

      And that .4 figure is only for teams in the bonus. Most fouls occur in
      non-penalty situations ... of course they do hurt your team by putting it
      closer to the penalty situation. And of course there's three-point plays
      and four-point plays. And offensive fouls are TOs, with their associated
      1 point cost. But combining them all, the average foul is worth well
      under 1 point, under .4 points for that matter.


      > robinson's regular season Tendex over 845 games is an amazing .721 per
      > minute, his playoff figures 96 games and an excellent .608 tendex. for
      > comparison:
      >
      > G Tendex Ply G Tendex
      > robinson 845 .721 96 .608
      > jordan 930 .728 179 .668
      > bird 897 .709 164 .634
      > magic 906 .732 190 .696
      > duncan 288 .603 39 .574
      > o'neal 608 .710 105 .644
      >
      > now considering that jordan, magic, and bird are considered by most the 3
      > greatest players in the NBA game over the past 20-25 years (if not all time),
      > and robinson's numbers are right up there with them, not quite as good but
      > very very very close, i think that says something for the admiral. he has one

      Well the other aspect of these numbers is that the Admiral is right up
      there with BAM (Bird-Air-Magic) in the regular season, but appreciably
      behind them in the playoffs. None of the players that you list show such
      a big dropoff in the playoffs. (We expect some dropoff, as Mike G has
      repeatedly shown, but not such a big dropoff.)

      I'm still not quite sure what to make of the Admiral. I do agree with you
      that he's Top 50 all time, easy. But whereas his regular season Tendex
      says he's right up there with BAM, I and I think most observers would
      agree that his true rating is a step below BAM. So that regular season
      Tendex is leaving something out ... like perhaps his playoff stats.

      > >finally, i think some stuff is double counted. tendex wants to give shaq
      > credit for off >rebs and points scored as if they were separate events, but
      > in reality shaq rewards >himself by taking those off rebs and jamming them in
      > the hoop or getting fouled...

      Double counting can be good, if we're aware when we're doing it. Your
      remarks about counting all the components which lead to scoring (or which
      prevent the opponent from scoring) are good. Another analogy is if we're
      looking at baseball, if we start counting runs scored by individual
      players, then it does make sense to count their RBIs as well, even though
      we know that the resulting sum will be almost twice as large as the actual
      number of runs scored.

      One area where Tendex's double counting is bad (but unavoidable with
      current statistics) is that it gives 2 points for a FG, and then also
      gives a point to the assist guy, if any. So an unassisted FG only
      generates 2 points in Tendex, whereas an assisted generates 3 points.

      All FGs are worth 2 points (i.e. +1 compared to the value of a possession)
      and it is silly to make some of them worth 3 points (in Tendex) and others
      worth only 2.

      This is an example where double-counting does lead to problems. The
      unassisted FG should be worth 2 points (+1 compared to possession). The
      assisted FG should also be worth +1 therefore. To be split between the
      assist guy and the scoring guy. How much credit the assist guy should get
      is a very interesting question, one which I think rates with Dean O's list
      of conceptual questions. And one which could be answered if we had truly
      good box scores which recorded assisted FGs, unassisted FGs, and ALSO FG
      misses which would've generated an assist, and FG misses which were
      unassisted. Then we could compare the relative FG%s, and see how much the
      presence of the assist guy raises teams' FG%.

      > as for allen iverson, i agree that he was not the league MVP last season -
      > not even close. he gave a defensive minded team exactly what it needed -

      Yup, this is why I give little credence to people who try to claim that
      Player M or Player D must have been great because they were chosen on the
      All-time Top 50 list. These lists contain mistakes, and the MVP list
      contains mistakes too, with Iverson in 2001 being merely one of the most
      obvious ones.



      --MKT
    • Show all 9 messages in this topic