forwarded message from John Conover

From: John Conover <>
Subject: forwarded message from John Conover
Date: Thu, 20 Mar 1997 01:12:48 -0800

Well, after the tech sell off today, (yesterday, Cisco announced that
Internet stuff was slowing down, affecting profitability, followed by
a similar announcement by 3Com,) I got about a zillion emails on
whether to bail out of the tech stocks.

I don't give financial advice, (it requires predicting the future, and
that presents epistemological issues for me,) but I will offer a few
data points to add to your intuitional tool box in such affairs.

If you consider any industrial market, operating in a free economy, it
will have fractal characteristics. (No kidding. Look at the
graphs. You can tell by looking, if you know what you are looking
for.)  This is because such things are a self referential system,
which, bottom line, means that scientific method is not
applicable. (Godelian issues are the reason.)  However, the aggregate
operations of many agents, as they struggle to make sense out of what
to do, with the data they have, and make operational decisions, makes
the characteristics of industrial markets a fractal, (ie., a
stochastic system, sometimes loosing, sometimes winning, in a fashion
that has a Gaussian, or normal, distribution. There are other
distributions that fractals can have, but they are not the discussion

With that said, we would expect that, since such things are fractal,
to see these large swings where the market pro forma, (if you want to
call it that,) swings above "average" for extended periods of time,
and then swings below "average" for extended periods of time. (I did
*_NOT_* say such things were cyclic, nor did I say they were
periodic-I did say that they were stochastic. For the record.)

What needs to be discussed is what "extended periods" means.

Well, here is the way it works. If you look at the graph of a market,
when it goes through zero, the probability that it will stay on that
side of zero for n many time units, (say years, for the sake of
discussion-which is commonly the way we talk about such things,) is
proportional to the reciprocal of n to the three half's power, ie., P
= 1 / (t^(1.5)). If we are in an "extended period," say for n many
years, then the probability that we will stay for one more time unit
in the "extended period" is proportional to the reciprocal of the
square root of n, ie., P = 1 / sqrt (1 + n). These are the definitions
of a Brownian motion fractal. The reasons for such a scenario is quite
complex, but comes from statistical mechanics.

There is one other characteristic of a fractal that needs to be
discussed. And that is the concept of "self similarity." What self
similarity means is that there is no time scale to a fractal. In point
of fact, if you take your data by the day, by the week, by the month,
or by the year, the scaling of the statistical data is a constant,
(which in our case is the square root of time, ie., n = 1 / sqrt
(scale), where scale is days, months, etc. As a side bar, this has
significant implications to the mathematical analysis of such things,
since the mean value theorem breaks, and we can not take derivatives
of such things-the derivative of such things is a mathematical "dust"
at plus and minus infinity. Fractals are everywhere continuous, and
everywhere non-differentiable. Bottom line, is that Newtonian/Leibnitz
calculus, ie., the mathematics of the transcendentals like logarithms,
exponents, sines, etc., is not applicable in such things. However,
early in this century, the stochastic calculus was invented to handle
just such affairs.) But the statistics are constant, ie., they are
time scale invariant.

With that said, what is the interpretation of these things?

Well, we would expect, that once a crossing of "average" in a market
pro forma is made, we would expect a probability of it staying there
of 70.7 percent. (So much for actuary tables, which are based on
statistical independence of time interval, ie., 50%. And, so much for
MBO, and revenue variances, which are based on the same thing.)  And
what about it staying yet another?  About 60%, (57.73%.) And a fourth?
50%. And this holds true, irregardless of whether the zero crossing
was to the "above" or "below" "average." And, it also holds true
whether we are looking at things in days, months, or years.

And so, should you sell out of the high tech markets. How many years
has it been hot? You can figure the probabilities for yourself, and
make your own decision.


BTW, want an intuitive example to support this hypothesis? Lets
suppose you were an operations person for a company in an industrial
market, with a lot of tenure-say over many decades. One thing you
know, from your experience, is that markets go up and down. Note that
on the average, over many decades, a market will be up for 3 years,
and down for 3 years. Walla! What every marketing person knows,
business cycles are between 5 and 6 years. (Actually, the word
"cycles" is quite incorrect, since it also holds for months and days,
etc., also, ie., it can be up on months, but down on years, at the
same time, etc.)

Want another one? What is the probability of an industrial market
staying stationary for 5 years, (the question I am asking is, suppose
we have a really great idea for a company, and that the market really,
really, wants, and our information about such things is 100% accurate,
and we want to form a company to exploit this market-what is the
chance of a VC getting a payoff from investing in us? That depends on
how long the market will remain up, before changing its mind.) Well,
how about a probability of 1 / 5^(1.5) = 0.089, or about 1 in
11. Which, not coincidentally, is very close to what VC's run, about 1
in 12.

And, just for another one, but at a different time scale, what is the
chance of a market success of an IC design that takes 6 months to
design? (Fortunately, I have an abundance of data on such matters,
since I have been curious about such things for a long time. A polite
way of saying I'm old.) Let me see now, 1 / 6^(1.5) = 6.8%, (which
happens to coincide to the third decimal place with published data for
the ASIC industry. This is an important concept, and is why John's
maxim has been "time is the enemy," and "he that does the most the
quickest, wins," since a 1 / t^n'th type of function falls off very
quickly. Simple concepts, really, and also why programmed traders run
by the minute on the ticker. Faster is better in things fractal.)

------- start of forwarded message (RFC 934 encapsulation) -------
Message-ID: <"0Tnr61.0.YF5.WcCCp"@netcom20>
From: John Conover <>
To: John Conover <>

         NEW YORK (Reuter) - Stocks cut their losses Wednesday after
suffering another bruising as investors turned negative on
technology shares and long-term interest rates jumped briefly
above the 7 percent level.
         The Dow Jones industrial average closed 18.88 points lower
at 6,877.68 following a 70-point plunge. In the broader market,
declining issues beat advances 1,532 to 931 on active volume of
535 million shares on the New York Stock Exchange.
         In the bond market, the 30-year Treasury bond was off 10/32,
and its yield ended at 6.99 percent following a brief rally to 7
percent from Tuesday's close of 6.96 percent.
         ``We're nearly at seven percent on the long bond and we're
certainly not going to make much progress while that's the
case,'' said Harry Laubscher, an analyst at Tucker Anthony.
         ``For now, people seem very concerned about next week's
(Federal Reserve) meeting,'' Laubscher said.
         Contributing to the worries about interest rates was a 0.3
percent rise in the Consumer Price Index for February, up from
0.1 percent in January. Economists had forecast a 0.2 percent
         While the slightly higher reading was not greeted with
alarm, economists said the data failed to show sufficient
weakness to change the unusually uncertain outlook for the March
25 meeting of the central bank's Federal Open Market Committee,
whose job is to determine the level of interest rates.
         ``The February CPI report won't change any minds about what
the Fed will do,'' Bruce Steinberg, manager of Macroeconomics at
Merrill Lynch said in a research note. ``We expect the Fed to
leave policy unchanged, but we admit it is almost a 50-50 call
at this point,'' he said.
         While stock traders remain extremely divided on the outlook
for interest rates, their attitude appears far clearer regarding
technology stocks, which have endured an extended thumping since
peaking in late January.
         That trend continued in force, sending the Nasdaq Composite
index to its lowest since early November, 1996. The Nasdaq lost
20.05 points, or 1.58 percent, to 1,249.29.
         The selling momentum appeared to gain momentum, hitting a
wide variety of tech stocks. Networking, semiconductor and
computer makers all finished with heavy losses.
         Intel Corp. fell 3 3/8 to 133 3/8, Ascend Communications Inc. fell
2 3/8 to 45 3/4, International Business Machines Corp. lost 1 1/2 to 137 7/8
and Microsoft Corp. sank 2 7/8 to 96 3/4.
         ``Everyone who wanted to sell is sold,'' said Gary Kaltbaum,
director of technical research at J.W. Charles Securities Inc.
''I think we're bottoming for a bounce, but that said, I don't
know how well we're going to bounce.''
         Also hit hard were transportation stocks, which backtracked
sharply from their recent rally. Among the losers, Burlington
Northern Santa Fe Corp. fell 3 7/8 to 78 1/2 after the railroad
company warned of weak first-quarter results.
         Airlines also finished lower, hit by profit-taking and
rising fuel costs. Delta Air Lines Inc. fell 3 1/4 to 84 1/2 and
United Airlines' parent UAL Corp. shed 2 1/4 to 68 1/4.
         Among other individual issues, Adobe Systems Inc. jumped 4 1/8
to 39, bucking the trend in tech stocks. The company reported
first-quarter earnings that were stronger than analysts had
         Lexmark International Group Inc. fell 3 7/8 to 23 7/8. The maker
of printers and related products said its first-quarter earnings
would meet analysts' estimates but revenues would fall short of
         The Standard & Poor's composite index of 500 stocks fell
3.89 points to 785.77. The American Stock Exchange index lost
3.65 to 591.80.
         The NYSE Composite index of all listed common stocks fell
1.81 to 413.59. The average share was down 18 cents.
         The Wilshire Associates Equity Index -- the market value of
NYSE, American and Nasdaq issues -- was 7,468.572,down 49.322,
or 0.66 percent.

------- end -------

John Conover,,

Copyright © 1997 John Conover, All Rights Reserved.
Last modified: Fri Mar 26 18:54:52 PST 1999 $Id: 970320011400.20403.html,v 1.0 2001/11/17 23:05:50 conover Exp $
Valid HTML 4.0!