From: John Conover <john@email.johncon.com>
Subject: Re: Crash, or long "bear" market?
Date: Sat, 25 Jan 1997 23:42:44 -0800
Gregory Keith Mccomb writes:
>
> Funny, I am no expert on stock markets -- but some initial reading on
> stock prices suggests strong evidence that prices follow a "random walk."
> Theories abound, for example "technical traders" think they can detect
> patterns in the data. But a strong prediction as I have just read seems a
> little more like voodoo than economic science.
>
> Greg McComb,
> U of Manitoba.
Hi Greg. I agree with you, but if stock prices are a true random walk,
(and there is a lot of theoretical and empirical evidence that
supports the argument,) then by definition there is no way to detect
patterns, since none exist. Any such proposal that implies "timing the
market" is ill founded, unless you can forecast the cumulative sum of
a random number generator, which is the definition of a random walk.
John
BTW, there is a good paper on this very subject. It is by the Stanford
economist Brian Arthur. It is "Complexity in Economic and Financial
Markets", and is available at http://www.santafe.edu/arthur. It
references another paper (also authored by him,) "Inductive Reasoning
and Bounded Rationality", also at the same URL. Both are good reading.
The latest "voodoo" I heard is that years that end in the number 7 are
"bear" years for the DJIA, (its not voodoo, it is contemporary
numerology.) I plotted the DJIA, and sure enough, for the entire
century, it is "true." How does such a curcumstance happen? It would
seem like a statistic of astronomical (or economical-the economic
numbers have surpassed astronomical numbers in magnitude,)
proportions.
Actually, if you consider all possibilities of rearranging the DJIA
for the years in this century, (all 97 factorial combinations-a large
but finite number,) and look at the first 9, or so, years of all the
possible combinations, we would expect every 500, or so, in the 97
factorial combinations to have the same characteristics-even if the
DJIA was constructed by a random number generator. (Chances are 1 /
2^9 = 1 / 512 that a 50/50 probability of something will align itself
all in a row, for 9 things, so every 512 of the 97 factorial
combinations, we would expect a "revelation.") The paper by Arthur on
inductive reasoning goes into such things, and there a zillions of
folks banging on keyboards hunting for revelations. Taken
collectively, (aggregately,) the revelations makes the DJIA a random
walk scenario-at least according to Arthur. At any rate, if such
"patterns" really did exist, they would be arbitraged away once
everyone found out about it.
P.S., I didn't post this to sci-econ. Go ahead and do so if you want.
--
John Conover, john@email.johncon.com, http://www.johncon.com/