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

Tendex rebuttal and Copernicus

Expand Messages
  • McKibbin, Stuart
    First, I want to say thanks to everyone for taking the time to comment on anything I have said. You all have great insight and experience here, and I certainly
    Message 1 of 2 , Nov 3, 2001
    • 0 Attachment
      First, I want to say thanks to everyone for taking the time to comment on anything I have said. You all have great insight and experience here, and I certainly don't want to give the impression that I'm barging in here trying to straighten everybody out. You guys know what you're talking about and I'm just happy to be part of the conversation.
       
      Bob Chaikin said: "If you are looking for a "one number" rating for players, which is what Tendex tries to do using readily available statistics, combining offensive and defensive stats is exactly what you want to do, because of the simple fact that the game is half offense and half defense...and don't you want to include as much information as you possibly can in evaluating talent? why is this weird?"
       
      I'm not looking for a one number rating system. So my use of the term Tendex appears to have been mistaken. However, I don't think combining offense and defense into one category with equal weight is a good idea. You play both offense and defense in a game of basketball, but other than that I don't know what can be said. They require different skill sets, offense to score, defense to stop someone from scoring---they're like different tools in a toolbox. Computers and shovels are both tools but I can't conceive of a way to judge their owner's overall usefulness with them using the same criteria or algorithm. Tendex certainly is collecting real numbers into one formula but jumping to the conclusion that the sum of the formula is "real" is something else.
       
      Three point percentage shows how good an outside shooter someone is, assists per 48 show how good a passer, rebounds per 48 show how good a rebounder. Supposedly Tendex tells us who is "really" being more productive---the  dude who gets 25 ppg, 3 rpg and 2 apg or the guy who gets 9ppg, 16 rpg and 3 apg. But as Dean Oliver points out there's no correlation (or at least nobody is saying there is) of the Tendex #  to SOMETHING, like how singles, homeruns, walks, etc. are correlated to runs. For example, in Thursday's game with Utah the Lakers had 105 points, 41 rebs, 22 assists, 8 steals, 8 blocks, 14 turnovers, 33 personal fouls, 10 missed FT and 44 missed field goals. The Tendex sum is 83. Utah had 101 points, 48 rebounds, 24 assists, 9 steals, 9 blocks, 16 turnovers, 30 personal fouls, 11 missed FTs and 45 missed field goals. The Tendex sum is 89. What does that mean, that Utah was more productive than LA? What is being evaluated? I just don't get it.
       
       
      Basketball analysis needs a Copernicus to imagine a new way of looking at the sport. Dean's suggestion of an offense only Tendex that correlates to points scored might be that new way.
    • Mike Goodman
      ... I think it s safe to say we are all happy with any well-thought out opinion. Even a biased or volatile opinion is part of what makes you a fan (=fanatic).
      Message 2 of 2 , Nov 4, 2001
      • 0 Attachment
        --- In APBR_analysis@y..., "McKibbin, Stuart" <smckibbi@c...> wrote:
        >I'm
        > just happy to be part of the conversation.

        I think it's safe to say we are all happy with any well-thought out
        opinion. Even a biased or volatile opinion is part of what makes you
        a fan (=fanatic). So, carry on.

        >You play both
        > offense and defense in a game of basketball, but other than that I
        don't
        > know what can be said. They require different skill sets, offense
        to score,
        > defense to stop someone from scoring---they're like different tools
        in a
        > toolbox. Computers and shovels are both tools but I can't conceive
        of a way
        > to judge their owner's overall usefulness with them using the same
        criteria
        > or algorithm. Tendex certainly is collecting real numbers into one
        formula
        > but jumping to the conclusion that the sum of the formula is "real"
        is
        > something else.

        In baseball, the defensive player wears a glove and a cap; the same
        player on offense holds a bat and wears a helmet. They really are
        separate incarnations of duty.

        In basketball, the manner in which one plays offense quite affects
        the manner in which one plays defense. Fast-breaking on offense
        leads to quick baskets against you. Overexerting on defense prompts
        a more deliberate offensive style.

        Whether this is a coaching strategy, or an individual style, the
        statistics enthusiast can track a player who moves from one style to
        the next. Some players are very consistant, in that a reduction in
        scoring tends to come with an increase in other stats (at least, when
        standardized.) Some players only thrive in an up-tempo game; others,
        only in a controlled game.

        Predicting how a player will do in all situations may not be a
        reasonable goal of statistical analysis. Whether or not a "value to
        team" measurement, "production", or "productivity" measures are
        accurate for one's current situation, predicting the effect of an
        environment change (i.e., trade or coaching change) on a player needs
        some firsthand knowledge of the player.


        > Three point percentage shows how good an outside shooter someone
        is,

        I would call that "accuracy and shot selection". Being "good" means
        you can actually create a shot, or at least get one off.

        >assists
        > per 48 show how good a passer

        Depends on turnover rate.

        , rebounds per 48 show how good a rebounder.

        If you don't foul out in 15 minutes.

        > Supposedly Tendex tells us who is "really" being more productive---
        the dude
        > who gets 25 ppg, 3 rpg and 2 apg or the guy who gets 9ppg, 16 rpg
        and 3 apg.
        > But as Dean Oliver points out there's no correlation (or at least
        nobody is
        > saying there is) of the Tendex # to SOMETHING, like how singles,
        homeruns,
        > walks, etc. are correlated to runs. For example, in Thursday's game
        with
        > Utah the Lakers had 105 points, 41 rebs, 22 assists, 8 steals, 8
        blocks, 14
        > turnovers, 33 personal fouls, 10 missed FT and 44 missed field
        goals. The
        > Tendex sum is 83. Utah had 101 points, 48 rebounds, 24 assists, 9
        steals, 9
        > blocks, 16 turnovers, 30 personal fouls, 11 missed FTs and 45
        missed field
        > goals. The Tendex sum is 89. What does that mean, that Utah was more
        > productive than LA? What is being evaluated? I just don't get it.
        >

        More than one contributor here is referring to Tendex as a single
        method. Even though I don't know what Tendex is, it is clear there
        is more than one use for the term.
        Stuart's Lakers/Jazz summary baldly shows that at least some Tendex
        ratings are meaningless; does that mean every "Tendex-like" system is
        equally bogus?

        What if there were a "perfect" Tendex that still predicted false
        outcomes. Could this be attributable to a coaching victory? Or a
        referee error?

        > Basketball analysis needs a Copernicus to imagine a new way of
        looking at
        > the sport. Dean's suggestion of an offense only Tendex that
        correlates to
        > points scored might be that new way.
      Your message has been successfully submitted and would be delivered to recipients shortly.