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

RE: [APBR_analysis] The most important stat that nobody seems to use! (PER differential)

Expand Messages
  • Michael Tamada
    ... From: mrintp2000 [mailto:shzys@netscape.net] Sent: Monday, May 03, 2004 4:11 PM ... [...] ... This is a good thing to look at. But ... [...] ... [...] ...
    Message 1 of 34 , May 3 5:37 PM
    • 0 Attachment
      -----Original Message-----
      From: mrintp2000 [mailto:shzys@...]
      Sent: Monday, May 03, 2004 4:11 PM

      >I've read a lot of the discussion about the PER and many people seem
      >to correctly regard it as the best way to measure a players
      >production. But even better, is to measure a player production vs. his
      >opposing player. This way, defense is correctly valued by the stats

      [...]

      >I call this the PER differential. The best players in the L are at the
      >top of course-

      This is a good thing to look at. But ...


      [...]

      >What really proves the tremendous value of the PER differential is the
      >correlation between team PER differential and winning percentage. You
      >guys have to forgive me, I have no stats background (just years of
      >watching hoops) so I can't give you a number on the correlation. But
      >check this out, here are the top teams ranked by PER differential
      >during the regular season. (done just before season end so these
      >numbers may have changed slightly-taken from 82games.com)

      [...]

      >I got these by adding up the net differentials at each position. Let
      >me know if it isn't clear how I arrived at these numbers.
      >
      >THESE NUMBERS PROVE THAT THE BEST WAY TO BUILD A WINNING TEAM IS WITH
      >PLAYERS WHO GIVE YOU THE HIGHEST PER DIFF AT THEIR RESPECTIVE
      >POSITIONS! The caveat of course is that they play enough minutes. This

      ... this is getting dangerously to the old strategy advocated by
      World B. Free (or was it George Gervin): "everybody outscore their
      man, and our team will win". A true enough statement (and if one
      uses points rather than PER, one doesn't even need any statistics,
      because a team which has each of its players outscore each of his
      opponents is *guaranteed* to win the game). But a bad recipe for
      team basketball, because there's more to basketball than just going
      out there and trying to light up the scoreboard more than your
      opponent does. Someone like Free (or Kobe) would love to have
      his coach use such a strategy, but as Phil Jackson and Shaq will
      tell you, it's not the way to win a championship.

      PER of course is not as vulnerable to this criticism because it
      is a more broadly based stat which looks at more than scoring. But
      the same general problem arises: basketball is not a set of five
      1-on-1 matchups, with whoever wins the most matchups winning the game.
      It's a team game.

      Why is this important? Two reasons: a negative PER differential
      is not necessarily a sign of failure. If Bruce Bowen has to go up
      against the likes of Kobe and Ray Allen while Tony Parker's matched
      up against an aged Payton (or Derek Fisher) and Brent Barry, Bowen
      might get outscored and out-PERed, but if he holds Kobe and Allen
      below their average or hold their efficiency below average, while
      not missing too many shots or committing too many turnovers of his
      own (and any points he scores are a bonus), he's done his job and
      put the Spurs in a good position to win. I doubt that he out-PERs
      his opponent most nights, because he's usually matched up against
      some hugely productive offensive machine while he himself will not
      score very many points. But (to repeat) that negative differential
      doesn't mean he's a bad player or doing a bad job. At least, not
      on this San Antonio team; he might be an undesirable player on a
      lesser team which needs more scoring from all of its players.

      Meanwhile, Parker would have a much easier time out-PERing his
      opponent, because Payton usually doesn't score that much these
      days are Barry, while an efficient players, doesn't pile up
      many points or assists.

      The other problem with PER differential is the team nature of
      defense. Houston had a nice defensive scheme against Kobe, but
      it was dependent on the whole team giving help; defense is NOT
      a 1-on-1 task, at least not when teams are using good team
      defense. At the end of Game 4, the TV commentators noted that
      for perhaps the first time in the game, the Houston defense
      slipped and left the Houston defender (I forget who it was,
      Mobeley?) 1-on-1 against Kobe; that was enough for Kobe to
      drive in and get one of his few field goals of the game.

      Kobe played 50 minutes that game, and for 49 minutes the
      Houston defender was apparently bottling Kobe up on 6-20
      shooting. For 1 minute (or really, about 3 seconds) the
      Houston team defense was slow to help out, and Kobe went
      1-1 in that time period. Was Kobe's (presumably) mediocre
      PER due to Mobeley's defense? Was his high PER during
      those closing moments due to lack of Mobeley defense?
      Or was Kobe's PER determined by the quality of the
      Houston *team* defense, and not due to Mobelely alone?

      The PER differential is a good thing to look at, and DeanO's
      defensive box scores are good things to look at, but an
      over-reliance on an assumption that one player's defense is
      responsible for the performance of his opponent is not a good
      thing. I suspect that's why Popovich and Larry Brown were
      driven batty by Rick Carlisle's use of defensive stats to
      pump up Ron Artest for Defensive Player of the Year.


      --MKT
    • Daniel Dickey
      ... Isn t zone played some in the NBA? Don t players play a variety of matchups in a game? Isn t it often hard to classify what position a player really plays
      Message 34 of 34 , May 7 5:27 PM
      • 0 Attachment
        >From: "mrintp2000" <shzys@...>
        >Reply-To: APBR_analysis@yahoogroups.com
        >To: APBR_analysis@yahoogroups.com
        >Subject: [APBR_analysis] The most important stat that nobody seems to use!
        >(PER differential)
        >Date: Mon, 03 May 2004 23:11:09 -0000
        >
        >I've read a lot of the discussion about the PER and many people seem
        >to correctly regard it as the best way to measure a players
        >production. But even better, is to measure a player production vs. his
        >opposing player. This way, defense is correctly valued by the stats
        >without an overemphasis on creating defensive stats that don't exist
        >or overvaluing unimportant stats. Allow me to illustrate with an
        >example...

        Isn't zone played some in the NBA? Don't players play a variety of matchups
        in a game?

        Isn't it often hard to classify what position a player really plays (let
        alone the position he defends)?

        Don't get me wrong - the PER differential can give SOME insight. But we
        need to remember the limitations.

        I can promise you that Reggie Miller's low opposing SG PER can be greatly
        attributed to Ron Artest (and maybe a few others). That's the type of thing
        I'm talking about.

        Dan D.

        _________________________________________________________________
        FREE pop-up blocking with the new MSN Toolbar � get it now!
        http://toolbar.msn.com/go/onm00200415ave/direct/01/
      Your message has been successfully submitted and would be delivered to recipients shortly.