Statistical Wisdom from CHiP’s

Today I feature a lesson on statistics brought to us courtesy of an episode of CHiP’s. The episode “Bio Rhythms” originally aired February 17, 1979.

This particular conversation took place between officers Frank Poncharello (“Ponch,” played by Erik Estrada) and Sindy Cahill (played by Brianne Leary).

      [Ponch] Hey Sindy. Feedback on the bio rhythms,
      right? Looks pretty good I guess, huh?

      [Cahill] Well it’s much too early to tell, Ponch.
      Ask me in a couple of months.

      [Ponch] Yeah but you must have enough to tell
      if the system works…

      [Cahill] Well, most everybody in chronobiology
      agrees that we all do have rhythms: cycles.
      But we don’t know exactly how it works or how
      the date of birth is involved. It’s going to
      take me a long time to run a large enough
      sample to eliminate coincidence. So for
      the moment, nope… I don’t have enough data.

      [Ponch] Sindy uh, listen… I’ve got a special
      interest. I mean, it does look good. Doesn’t it?

      [Cahill] Ponch, “looks good” is not a term we
      use in statistical study.

Many times I have heard traders speak with overconfidence about strategies that produced one or two winning trades.

I have also seen many casual traders extremely eager to pounce on any promising idea they hear, read, or see.

Both examples are germane to the conversation between Cahill and Ponch. Small sample sizes are susceptible to the possibility of fluke or, as Officer Cahill stated, coincidence. We therefore know nothing until we get a larger sample size. Heuristic thinking like confirmation bias is notorious for driving action during this phase and quite often, people circumvent the hard work altogether by not insisting upon (or being aware of) proper statistical validation.