From: John Conover <john@email.johncon.com>
Subject: more on stock market
Date: Fri, 8 Oct 1993 02:57:19 -0700 (PDT)
Hi again Joel. Let me offer a further explanation of the SM model we
are using with an historical persptective. (I'm operating on some
machine, someplace in net land that does not have a spell checker, so
my appologies for semantics of this text.)
In the 17th century, tullips were introduced in Holland from someplace
in Africa. They really took off, and were in great demand. They were a
great investment, sometimes doubling in value in just a few days. The
trend continued to the point that whole farms were traded for a single
tullip bulb. The trend was also of a relatively long duration, so that
confidence was established in the investment potential of tullip
bulbs. (This is a true story, BTW.) Naturely, the inevatable happened,
and the market collapsed, leaving a lot of Holland financially
destitute, as tullip bulbs assumed their true market value.
One of the problems with Keynesian economics is that it is not capable
of handling these types of situations (supply and demand is based on
true market equilibrium values-equalibrium is the keyword.) It can
not handle the issue of why there is a demand for mink coats, when
their insullating qualities are inferior to wool, but yet they cost
more. (Although, it does work well for industries like table salt
producers, where equilibrium points are ordained by biological need.)
The hidden paradigm of Keynesian economics is that the markets are
analyzied as if they are in equilibrium. (ie., as opposed to being
dynamic.)
To address these problems, in the early 1940's, Oskar Morgenstern and
John Von Neumann wrote "Theory of Games and Economic Behavior"
(published in 1944.) This is the "bible" of "neo-classical" economics.
Although they did not deny that supply and demand forces in the
economic system existed, they did postulate that the market value of
something was based on the "perceived value" of that something.
Unfortunatly, this scenario can not be analyzed with conventional
mathamatics, but a model (more or less accurate) could be formulated
using the conecpt of mini-max from game theory (which Von Neumann
created in 1937 and published as "Aur Theorie der
Gesellschaftsspiele.") As a side bar, Von Neumann also axiomatized
the quantum mechanics (we still use this axiomatization today in
semiconducor process/transistor/circuit design) and got his self into
numerical computation trouble designing the state equations of
plutonium for nuclear weapons-so he also invented the modern
electronic computer, (ie., the "Von Neumann architecture.") This
economic theory is still widely used in corporate America's operations
research departments (3/4 of the Fortune 500 companies have them.) The
buzz words here are "mathematical programming" and "linear
programming," which are the mathematical techniques used to solve the
game theoretic matrix of an economic situation to find the mini-max
saddle point, which is the optimimum operating point for the company
in a given economic scenario. If you want to play with these, they are
in the Unix systems under the names "strategy," "optimize," or
"lp_solve." The LINPACK sources from the US Government are floating
around also. (I wrote strategy and optimize, and lp_solve came out of
ATT Labs.) Most of these optimization concepts were created by George
Dantzig of Stanford, under contract to the USAF to do optimiztions for
If you analyze what happened to the tullips in Holland in the 17th
century, it is really nothing more than a pyramid scheme. And a
pyramid scheme can not have an equalibrium. (If it did, it would not
be a pyramid scheme.) In these scenarios, timing is everything. You
want to get in early, and get out early-but not so early that you
leave a lot of money on the table-and not so late that you run the
risk of holding an investment in a crashing market. The question is
whether the stock market is a "neo-clasical" long term equilibrium
market, or more like a pyramid scheme. The answer is that it is more
like a pyramid scheme. Why? For one reason the statement "buy low,
sell high" is valid-it would not be if the market is in long term
equalibrium, on the average. If that were true, you would make more
money the longer you waited (in general,) and we know that this is not
necessarily true. So we have to conceed that the stock market is a
dynamic process. How dynamic? Emperically, it has a frequency
distribution of 1/f squared, (ie., Brown noise, after "Brownian
motion") which is, also, the theoretical solution to the iterated
pyramid scheme. (It is not coincidence.)
How wide spread are these dynamic processes in the rest of the
economic environment? No one really knows, and it is a source of great
debate amoung economists. The neo-classical economists say
infrequently, and the modern economists (lead by Brian Author of
Stanford) say everywhere. He has referenced the failure of BetaMax to
replace VHS (since equalibrium economics would predict that a
technologically superior product that can be manufactured cheaper will
eventually surplant an inferior product in the market place.) But
this didn't happen. Why? because VHS was in the market first. Since it
was there first, more movies were available, and people would buy VHS
systems as the alternative. And, since more people had VHS, the movie
makers made more VHS movies available, which sold more VHS systems,
and so on.
The important concept here is that a "positive feedback" situation was
envolved. This is heretical in the dogma equilibrium economics. If
you think about it, (from a circuit standpoint,) something with
postive feedback can not have state equations in equalibrium. The
current trend in economics is to analyze these situations using the
theory of "dynamic systems," (which is also called, in lay terms,
"chaos theory.") And if you use these theories to model things like
the tullip bulbs and VHS/BetaMax scenarios, you can predict that the
dynamics will be largely "unpredictable," but will have a 1/f type of
power distribution.
Note that 1/f solutions are not the only solutions in dynamical
systems that exist. Other solutions envolve "phase transitions," (you
can predict that water will turn into ice at about 32 deg. F, by
considering the properties of large number of water molecules as a
dynamical ergotic entropy system, for example-interestingly, this is
the only know way to "prove" that the phase transition of water
exists.) But do these "phase transitions" exist in the capitilst
domain also? Appearently, the answer is yes. There is some evidence
that the mixture of "anarchy" and "order" in social systems (ie., like
companies) does not exist as a continious "blend," (starting with
total anarchy at one end, and ending with total order at the other)
but as a very sharp phase transition-between such a rigid order that
the company can't get out of its own way, and an anarchy so intense
that the company discentagrates. What is very peculiar about these
dynamic systems is that they are "self organizing." Left to their own
means, they organization's "state" will migrate to the phase
transition point (ie., neither anarchy or order, but just in between.)
Since positive feedback is envolved, pushing it one way or the other
from this point will result in an instantaneous "snap" to total
anarchy, or total ridgid order. The phase transition point also has
the added advantage that it is the most efficient operating point for
the organization. (Information theoretic analysis also reveals that
the maximum information flow through the organization occurs at the
phase transition point, also.) If you think about this, it makes
sense. On the one hand, it is not so anarchist/chaotic that it can't
do anything, and, likewise, it is not so ridgedly ordered that it
can't do anything. The important concept (if it is true, and there is
mounting evidence that it is) is that the anarchy/ordered transition
is very "sharp" and non-linear. At this point, the organization will
operate with maximum efficiency and agility. Unfortunatly, the
"phase transition point" is largely "unpredictable." (Like the market
dynamics, above.)
So if these things are "unpredictable," what is the best game plan?
First admit that the non-linearities are killers if you do not know
that they exist. Second, the best game plan is to always be agile and
quick (this implies that you are operating at a phase transition
point, and got there in a "self organizing system" manner,) and never
close off options.
John
--
John Conover, john@email.johncon.com, http://www.johncon.com/