From: John Conover <john@email.johncon.com>
Subject: Re: Truth in prognostication
Date: 10 Jan 1999 07:45:11 -0000
Well, it is January 10, 1999. As you will recall, just to refresh your
memory, I made some prognostications-on January 10, 1998-using only
simple fractal analysis that could be done in one's head concerning
whether various time intervals in 1998 would have bull or bear equity
markets.
I kept up with the prognostications throughout the year, as you will
recall, and sent out updates-since most folks don't use search engines
on their email, and I do, they are dutifully attached. Remember, it
was in the spirit of "Truth in Prognostications" that I did this-I was
tired of all of the market druids not being held accountable for their
prognostications. (The original message had the "Subject: forwarded
message from Spirit Of Truth Page", which contained a rather
pessimistic prognostication for 1998, which just flat disagreed with
anything rational-the original message was removed from the attached.)
If you will recall, the prognostications that I made involved whether,
starting on January 10, 1998, the next day, week, month, and year,
would be bear or bull. I made the prognostications based on going to
http://www.stockmaster.com/, getting the DJIA graph, and counting back
by the day, week, month, and year, to find the last day, week, month,
or year, that was opposite to the state of affairs in the markets on
January 10, took the square root of that number, (in my head,) and
said that that was the reciprocal of probability that the state would
continue, or change. (The actual market has about a 10% exponential
increase per year, but I ignored that, since compound interest is hard
to work out without a calculator-and it had to be done in one's
head. Also, it is not a Gaussian or normal distribution on the
increments-instead of being the square root function, n^0.5, it should
be n^0.6, but I ignored that, too, since I couldn't do that in my
head, either. Likewise, I also ignored the error function-it is really
the erf (1 / sqrt (n))-since I couldn't do that in my head. I then
said, since there is correlation between the market indices, that most
markets would be similar.) But what I did was reasonable, with a
calibrated eye, and the ability to take the square root of small
numbers in ones head. As you will recall, (from the second attached
message, dated "Sat, 31 Jan 1998 09:48:40 -0800",) it did quite well,
at least through the month.
And, how about the year?
There was a 62% chance of a "bear" year. The DJIA ended up 10.8%, its
worst showing since 1994, and below it average for the century,
including the depression years of the 1930's, call it neutral-maybe
slightly "bull". Both the S&P and NYSE ended up a little over 20%
gain, call it "bull" and the broader indices, the Value Line, the S&P
2000, and anything that included midcap stocks ended up in the red,
call it "bear". So, it is kind of a judgment call, I would call it
slightly "bull", and give it a C+ at the end of the year for the
prognostication. (I went out on a limb with the 62% number-what that
number means is that I would expect to be correct 62% of the time, and
wrong, 38% of the time.)
All in all, not that bad of a set of prognostications, for something
so simple it can be figured out in one's head, by looking at a graph.
John
BTW, if I would have said that there is a 62% chance that the DJIA
would be below its average of 10.6% growth per year, it would have
predicted it correctly for the day, week, month, and year-but I would
have had to use a four function calculator to figure what the 10.6%
annual growth was at the day, week, and month. If I do that, I have a
problem that the prediction was 100% correct, and can't account for
not having 1 or 2 of the four predictions turn out to be wrong. The
62% number is actually quite a bit closer to 100% if you use n^0.6
instead of n^0.5. But that was not part of the "deal". I could not
use numerical data, (all I could do was look at a graph,) and could
not use a calculator, computer, or any analytical program. Even so, it
turned in more respectable numbers than the market gurus did. The
updates I mailed through January (from Reuter, API, and UPI,) had the
prognostications from the gurus-if you want them, let me know.
--
John Conover, john@email.johncon.com, http://www.johncon.com/
-------------------------------------------------------------------------------
The attached is more Domesday prognostications from the macroeconomics
community. The attempt utilized is to establish some sense of
credibility for such prognostications through a consensus of media
reports. Not being daunted by the selective democratic polling of a
the members of a profession which is founded on inductive reasoning,
(ie., statistics,) I worked my way through the numbers:
Chances of 1998 being a "bear" year? 62%, (and a 38% chance of
being a "bull" year.)
Chances of January, 1998 being a "bear" month? 41%, (and a 59%
chance of being a "bull" month.)
Chances of next week, (starting 12 January,) being a bear week?
29%, (and a 71% chance of being a "bull" week.)
Chances that the day of Monday, 12 January, will be a "bear" day?
45%, (and a 55% chance of being a "bull" day.)
Chances that the day of Monday, 12 January-if it is a "bear"
day-of the equity markets dropping at least 6%? 0.13%. (Note that
such small probabilities do not mean something will not
happen-only that there exists a remote possibility that it will. A
better way of looking at such things is that we would expect a 6%
daily drop in the equity markets every 769 business days, or about
once every 3 years.)
Although not cause for unbridled optimism, the probabilities certainly
do not justify cynical pessimism, either.
Is a US equity market disaster possible? Yes it is, (that's life.) Is
it probable? There is a little less than a 0.0001% chance that the US
equity markets will drop 33% in the month of January, 1998, (or, very
roughly, about the same chances of perishing in a commercial aviation
disaster,) which is about the average drop in the Asian Tiger's equity
markets for the month of December, 1997. (Note that such small
probabilities do not mean something won't happen-only that there
exists a remote possibility that it will.)
How about the probability of 1998 ushering in a sustained bear market
lasting at least 25 years, (as from 1930-1955, ie., a great depression
starting again?) 20%. (20% sounds almost eminent, huh? Well, not
exactly. There is also a 27% chance that the "bull" market that
started in 1984 will continue through 1998. There is, also, an 11%
chance that the current very "bull" market, that started in 1992 will
continue through 1998.)
John
--
John Conover, john@email.johncon.com, http://www.johncon.com/
-------------------------------------------------------------------------------
And, how has it turned out so far? So far, so good. Have to wait and
see how the year turns out ...
John
> John Conover writes:
> >
> > The attached is more Domesday prognostications from the macroeconomics
> > community. The attempt utilized is to establish some sense of
> > credibility for such prognostications through a consensus of media
> > reports. Not being daunted by the selective democratic polling of a
> > the members of a profession which is founded on inductive reasoning,
> > (ie., statistics,) I worked my way through the numbers:
> >
> .
> .
> .
>
> >
> > How about the probability of 1998 ushering in a sustained bear market
> > lasting at least 25 years, (as from 1930-1955, ie., a great depression
> > starting again?) 20%. (20% sounds almost eminent, huh? Well, not
> > exactly. There is also a 27% chance that the "bull" market that
> > started in 1984 will continue through 1998. There is, also, an 11%
> > chance that the current very "bull" market, that started in 1992 will
> > continue through 1998.)
> >
> > I just received some email requesting an assessment of the probability
> > of at least the following decade, starting 1 January 1998, being
> > "bear" or "bull."
> >
>
> The probability that the decade would be bear/bull is 50%/50%, almost
> exactly.
>
> So, starting with 12 January, 1998, the probabilities stack up as
> follows:
>
> Chances of at least a quarter century "bear" equity market: 20%.
>
> Chances of at least a one decade "bear" equity market: 50%.
>
> Chances of at least a one year "bear" equity market: 62%.
>
> Chances of at least a month "bear" equity market: 41%.
>
There was a 59% chance of a "bull" month. It was-barely. (NASDAQ,
Up. DJIA, up. NYSE, Jan 1 = 511.19, Jan 30 = 510.63, call it
neutral-maybe slightly "bear." S&P, Up.)
>
> Chances of at least a week "bear" equity market: 29%.
>
There was a 71% chance of a "bull" week. It was, for all indices.
>
> Chances of at least a day "bear" equity market: 45%.
>
There was a 55% chance of a "bull" day. It was, for all indices.
>
> John
>
> BTW, note that these probabilities only apply to 1 January 1998, and
> will be different for any other day/week/month/year/decade in the
> quarter century. Note also that the probabilities change over time.
> For example, the probability that the decade starting 1 January 1978
> would be "bear" was only 29%, (ie., a 71% chance that it would be
> "bull," which it turned out to be.) Note, also, that the
> probabilities are dependent on the time scale-although there is a
> likelihood that the quarter century will turn out to be "bull," which
> days in the quarter century will be "bull," (or "bear,") can only be
> forecast a few days in advance. Likewise for the weeks/months/years.
>
> As an interpretation of the probabilities for the US equity markets,
> beginning on 12 January, 1998, the near short term is somewhat
> optimistic, changing to optimistic for the week, and somewhat
> optimistic on the outcome of the month, but pessimistic on the year,
> neutral on the decade, and optimistic on the quarter century.
>
--
John Conover, john@email.johncon.com, http://www.johncon.com/
-------------------------------------------------------------------------------