From: John Conover <john@email.johncon.com>
Subject: Re: Deming's comment "in God we trust-all others bring data."
Date: Fri, 20 Sep 1996 21:29:22 -0700
On tonight's "Nightly Business Review," on PBS, an equity market
analyst, not to mention any names, made the comment that "7 out of the
last 10 election years had financial bull markets," In the spirit of
Deming's comment, "in God we trust-all others bring data," I decided
to see if election years really were responsible for bull markets, as
implied by the analyst, (in the context that the FED runs low discount
rates, intentionally, so the financial markets will be optimistic.)
So, I ran a fractal analysis on the NYSE's Composite, DJIA, and S&P500
time series, and low and behold, the Shannon probability, (and also
the Hurst exponent,) was 0.68-meaning that the equity market indices
are very definitely fractal, which, in a nut shell, means that the
statistics of the indices, over time, are what is called "self-affine
and self-similar." What that means is that the statistics of the
indices do not vary, irregardless of the time scale used, be it
minutes, hours, days, months, years, decades, or whatever. Also, it
means that given any interval of time-no matter how you define
time-68% of that time will be a bull market, and 32% a bear market. Or
in other words, taking the last 40 years for example, or so, and
randomly picking years, about 70% of the years picked will have bull
markets, the remainder bear markets-and it does not make any
difference if I pick every 4th year, every other year, every third
year, or just years at random. That statistic will hold, and be
invariant.
So, are election years responsible for bull markets? Of course not.
It is always easy to find statistical significance where there isn't
any.
John
BTW, he should have known better-he is a certified financial advisor,
and his boards test knowledge of entropic analysis-what, in the lay
vernacular, are called fractals. FWIW, you can quote any thing you
want-for example, 70% of the time we had a Democratic Congress, it was
a bull market. Never mind that 70% of the time we had a Republican
Congress, we also had a bull market. It could also be stated that 70%
of the time that the tide is out, we have a bull market. Or, also,
bull financial markets have a correlation with the price of eggs in
China, for 70% of the time. All, of course, are blatantly incorrect.
Where the guy made his mistake was assuming that the financial markets
can be divided into a bear and bull markets, approximately evenly.
Although it sounds reasonable, it just isn't so-instead of each market
prevailing for half the time, bull markets actually prevail 68% of the
time-skewing the distribution mean from 0 to the positive side. That
is why the markets increase in value over time-if it had a zero mean,
the DJIA, etc., would still have the same value as in the last
century. Funny thing for a financial advisor that specializes in
equity markets not to know.
--
John Conover, john@email.johncon.com, http://www.johncon.com/