Re: General Thery seminar

["J. Barkley Rosser, Jr." <rosserjb@xxxxxxx>]

Ronald,
      Guess I don't find your "true" versus "false"
traders story too appealing.  Using these two
kinds of agents, how do you get a bubble?  Is
it a particular large positive exogenous shock
pushing both kinds of traders into buying?  I
don't get it, I'm afraid.
       The fundamentalist/chartist distinction, for all
its problems, does give us a mechanism for
dynamics.  Given a perceived fundamental, let it
be the average opinion of the "well-informed"
fundamentalist traders, an arbitrarily defined group
perhaps (and certainly a Keynesian beauty contest
story); then one can predict behavior of agents.  One
can easily see fundamentalists and chartists doing
different things in different situations (fundamentalists
selling when price is far aboved the "perceived fundamantal"
while chartists keep buying if price is moving up fast enough).
This allows for an understanding of market dynamics with
different ratios of the kinds of traders.
        The literature on this is growing rapidly, but the
general perception that an increase in the percent of
traders following chartist strategies destabilizes markets
and complicates dynamics rather neatly holds (I also
recognize that there are other traders who are not so
neatly categorized, complicated program traders working
for Henry and his competitors; the behavior of whom is
neither clearly stabilizing or destabilizing, and may in fact
be locally stabilizing but potentially globally destabilizing
as the LCTM outcome and the 1987 crash all may show).
Barkley Rosser
-----Original Message-----
From: Ronald Calitri <calitrir@xxxxxxxxxxxxx>
To: POST-KEYNESIAN THOUGHT <pkt@xxxxxxxxxxxxxxxx>
Date: Saturday, January 29, 2000 6:59 PM
Subject: Re: General Thery seminar


>Paul Henry Barkley Ted
>    Henry's model is actually quite good. It is however, a crowded terrain
>compiled from many papers. The financial market emulates academic finance;
>and there wouldn't be a game if everybody wasn't able to win at least
>somewhat consistently. By the way, Henry you are right that returns are
>falsifiable; you neglect to note this is the standard practice in most
>living corporations. The indices are compiled with error.
>    So Barkley, let's try a different "game". Suppose there are two types
of
>traders in the market: True and False. The True traders adhere to their own
>fundamentals. That is, they cast their balance sheets and income statements
>after every trade. That is for ordinary investors, whose borrowing or
>lending depends on present circumstances.
>The False traders cast their balance sheets only when they are overbought
or
>over-sold. The distinction is designed to allow them to appear less hedged
>when appealing to borrowers and more hedged when going after investors.
That
>is for corporate investors with accounts on many sides that may be loaded
>off.
>    This model contains only "internal" features. Put in a "fundamentals"
>dimension and the "False" traders seem to be farther from the
>security-market frontier, and would presumably be in line to accrue
>differentially any benefit from systematic forces attracting them towards
>the norm.
>    The question of interest then, concerns what will be the statistics
>generated by the false and true mixture of positions. The question is not
>whether the future is indeterminate, but rather whether the "true" data
>about it may be distinguished from "falsified" data. Or more prosaically
>whether heteroschedasticity in measurement error may be distinguished from
>variance in the underlying process.
>    The speculation arises whether the undetectable trading off of
>falsifications is involved in a bubble process, trapping true investors on
>an exponential copula.
>
>    By the way, it is not necessary the falsification be indetectible, just
>relatively intractable. There could be several layers relatively T/F and
>turbulent stability. What I was thinking about was a simple example where
>firms that acquire partial insight into the future like Henry's could
>coexist with firms that distort their accounting data, without engendering
>systematic instability. The conclusion is that they could coexist so long
as
>they both drained resources from a third group with stricter constraints on
>positions. Rock-solid stability could be acquired if the "true" traders
were
>entitled to falsify returns against another group not allowed entree to the
>market at all.
>
> Chiao,
>Ronald Calitri
>
>
>
>----- Original Message -----
>From: Paul Davidson <pdavidson@xxxxxxx>
>To: POST-KEYNESIAN THOUGHT <pkt@xxxxxxxxxxxxxxxx>
>Sent: Friday, January 28, 2000 5:14 PM
>Subject: Re: General Thery seminar
>
>
>> At 10:05 PM 01/27/2000 -0500, Henry wrote:
>> >Believing that a computer program can be a good tool for consistently
and
>> >systematically generating stock anaylsis for stocks, our group have
built
>> >a model which we call  Artificial Intelligence Live Stock Commentary
>> >Engine (AI LSCE).
>> >We actually have a website:  www.tradetrek.com  (This is not a plug, but
>> >all are welcome to visit it, free).
>> >
>> >
>> >                    Numerous studies have indicated that the stock
>market,
>> > as well as other financial markets, are predictable to a certain
extent.
>>
>>
>> Is that,  Henry, how you "knew"  the NYSE Dow Jones would go down by more
>> than 300 points today. Right?
>>
>>
>> >                    The movements of stock prices (as well as the prices
>> > of other financial instruments) generally have a deterministic trend
>> > superimposed by some "noise" signals, which, in turn, are composed of a
>> > truly random signal and a chaotic signal.
>>
>> Ex poste Henry -- everything has a deterministic trend.  All that means
is
>> that you are calculating a moving average of a series of numbers. By
using
>> the mathematical formula you to fit the data, you assure that there is
>> always a moving average (i.e., a deterministic trend) and some vairance
>> around the moving average.  That is the nature of how a moving average is
>> calculated, it has nothing to do with the real world-- if you believe the
>> future is encased in an nonergodic environment.
>>
>> As a test Henry, generate a column  of Random numbers -- or copy a series
>> off of a randon number Table-- and see that you can always find a moving
>> average (deterministic )trend and some random noise variance.  So what
>does
>> it mean?
>>
>> I wish economists would learn "design of experiments" processes before
>> rushing off  for computer fishing expeditions.
>> The computer has no intelligence-- I wish people who manipulate computers
>> would use their natural intelligence.
>>
>> PAUL
>> Paul Davidson
>> Holly Chair of Excellence in Political Economy
>> Editor, JOURNAL OF POST KEYNESIAN ECONOMICS [JPKE]
>> Economics Department -- 523 SMC
>> University of Tennessee
>> Knoxville, Tennessee 37996-0550
>> email: Pdavidson@xxxxxxx;   phone: (865)974-4221;    fax: (865) 974-4601
>> http://econ.bus.utk.edu/Davidson.html
>>
>>
>
>