Aggregate Data

[Norman Hovda (990826.10:06 MST)]

[From Rick Marken (990826.0900)]

Norman Hovda (990825.16:06 MST) --

> based on my observations and using my dynamic data distribution
> (DDD) model to measure _one_ stock's price data will yield
> similar perceptual control system phenomenon as demonstrated by
> aggregate data derived from the entire stock market index data

This sounds interesting. What are the perceptual control system
phenomena that are demonstrated in the behavior of the aggregate
and individual stock prices?

An animation may be worth many words?

Download player.exe (201K) at:

http://www.ultrasw.com/nth/pct/player.zip

Download animation file (972K) at:

http://www.ultrasw.com/nth/pct/dj990820.zip

Unzip both to the same folder. Open player, select the CAS.anm file
and enjoy the show.

In this example you will see about 2.75 days of mkt activity in about 70
seconds. The _mode_ of each distribution (about 600 in this animation)
is the Reference Signal. The changing trial and error RS's have varying
degrees of acceptance as measured by time.

Notice that the data start out rather stable or balanced, where the CV
(magenta line) is effectively controlling for the DDD mode or Reference
Signal within a relatively narrow high to low range or low error rate. Then
a big BUYER disturbance occurs as evidenced by range extension or
expansion.

Continue to watch as the DDDs Reorganize by trial and error for new
DDD modes or Reference Signals achieving various degrees of
acceptance. Eventually the CV is controlled more effectively at the end
of the sample, and the DDD is once again relatively balanced in a
narrow high/low range around the newly adjusted mode or Reference
Signal at higher prices.

To hopefully answer your question Mark, and offer some support for my
original claim, this aggregate data sample of the Dow Jones stk index,
comprises 30 stocks. I can create DDDs for the same time period for
any one of the 30 stocks and you would observe similar cycles of states
of balance, effective control or low error rate, then Disturbance, then
Reorganization through tested but relatively unaccepted Reference
Signals, back to a controlled balance around a more accepted low error
rate Reference Signal, etc. As you point out, the aggregate basket of 30
stks is "still irrelevant to" the DDD of any one particular stk and vice
versa. However, they do inform a specific context, as in "When they raid
the whore house they take the piano player too" or if the madame is
well connected the raid may never occur. <g>

Is this making any sense?

Thanks for reading.
nth

[From Rick Marken (990826.2010)]

Me:

What are the perceptual control system phenomena that are
demonstrated in the behavior of the aggregate and individual
stock prices?

Norman Hovda (990826.10:06 MST)--

An animation may be worth many words?

I got it. It's very pretty. But I'm not sure I understand
what I'm seeing.

In this example you will see about 2.75 days of mkt activity
in about 70 seconds. The _mode_ of each distribution...

What are these distributions of? What's the variable on the
vertical axis? Stock prices?

is the Reference Signal.

I don't understand this. Perhaps you mean that the mode is
the reference _state_ of the variable (stock price?) on the
vertical axis? The reference signal is a theoretical notion
that accounts for the observed reference state of a controlled
variable.

The changing trial and error RS's have varying degrees of
acceptance as measured by time.

I don't understand this either. What is an RS? Is degree of
acceptance measured _by_ (in terms of) time or is it measured
_over_ (at different points in) time?

Notice that the data start out rather stable or balanced, where
the CV (magenta line) is effectively controlling for the DDD
mode or Reference Signal within a relatively narrow high to
low range or low error rate.

I'm afraid this is completely incomprehensible to me. Is your
animation the behavior of data or the behavior of a model
control system?

Then a big BUYER disturbance occurs as evidenced by range
extension or expansion.

Ah. It must be data; you are guessing that there was a
disturbance (to what variable; the range of the distribution?)
because it was somewhat effective. Is that right?

Continue to watch as the DDDs

What are the DDDs?

Eventually the CV is controlled more effectively at the end
of the sample

What is the CV?

To hopefully answer your question Mark

Rick

I can create DDDs for the same time period for any one of
the 30 stocks and you would observe similar cycles of states

Again, what's a DDD?

Is this making any sense?

Not yet. My main problem is that I don't know what variable you
think is controlled and I don't know _why_ you think it is
controlled.

Best

Rick

···

--

Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[From Bruce Gregory (990827.0559 EDT)]

Rick Marken (990826.2010)

> Is this making any sense?

Not yet. My main problem is that I don't know what variable you
think is controlled and I don't know _why_ you think it is
controlled.

You are not the only one!

Bruce Gregory

[From Norman Hovda (990827.1045 MST)]

[From Rick Marken (990826.2010)]

Me:

> What are the perceptual control system phenomena that are
> demonstrated in the behavior of the aggregate and individual
> stock prices?

Norman Hovda (990826.10:06 MST)--

> An animation may be worth many words?

I got it. It's very pretty. But I'm not sure I understand
what I'm seeing.

Well, let's raise a glass to something coming into focus soon?

> In this example you will see about 2.75 days of mkt activity
> in about 70 seconds. The _mode_ of each distribution...

What are these distributions of? What's the variable on the
vertical axis? Stock prices?

Oooops. Yes.

> is the Reference Signal.

I don't understand this. Perhaps you mean that the mode is
the reference _state_ of the variable (stock price?) on the
vertical axis?

OK. Yes.

The reference signal is a theoretical notion
that accounts for the observed reference state of a controlled
variable.

Thanks for the clarification. Would this make sense then, assuming I
understand the distinction you're making?

If the observed Reference State is price X, and Greenspan is talking or
the Fed is about to make an announcement, the Reference Signal for
many mkt participants (crowd) may be disturbed in such a way as to be
observed as a change in the Reference State price X +/- Y?

> The changing trial and error RS's have varying degrees of
> acceptance as measured by time.

I don't understand this either. What is an RS?

If I understand your distinction made above... Let's go with "Reference
State".

Let's say the Reference Signal (A1) is in this case a theoretical notion
that the crowd, mkt movers and shakers, is collectively comfortable with
trade at current Reference State price levels (B1) because the control
variable of price (C) is narrowly hovering +/- around Reference State
price (B1).

Now there is a Disturbance (FED announces unexpected rate cut) for
which the crowd developes a new Reference Signal (A2). The market or
trading crowd then begins to Reorganize with trial and error price
volatility emerging and new Reference States (B2, B3, B4, etc.) are
tested for acceptance. Once again a cycle completes when acceptance
developes around... let's say (B3), which then implies an alignment with
the new Reference Signal (A2)?

Is degree of
acceptance measured _by_ (in terms of) time or is it measured
_over_ (at different points in) time?

I'm foggy about the terminology.

Hmmm... for shorter time frames I'd say _by_ time. The mode is
determined as the price traded most often (greatest acceptance) within
x minutes. For this animation 60 minutes.

For larger time frames, a day or longer, it seems to prove useful to
compare levels of acceptance _over_ time. By that I mean, if over the
past X days there are two emerging Reference States, A and B, and
more time is accumulating (greater acceptance) at B, which is higher
than A, then that suggests the trend is continuing up; the market crowd
is controlling for higher Reference Signal price levels.

> Notice that the data start out rather stable or balanced, where
> the CV (magenta line) is effectively controlling for the DDD
> mode or Reference Signal within a relatively narrow high to
> low range or low error rate.

I'm afraid this is completely incomprehensible to me. Is your
animation the behavior of data or the behavior of a model
control system?

I'm guessing "behavior of data"? From what I understand of PCT, my
data seem to exhibit correspondence with "behavior of a model control
system" based on what I've experience with PCT demos I've tried. I don't
know if that's fair and my ignorance is prolly showing big time. Thanks
for your patience. Can you help me understand the differences better?

Let me try describing the beginning of the animation again. The data
pattern begins with a normal bell distributions rotating a little up and a
little down, remaining relatively well balanced, and for a one hour window
the range is relatively contained and small, the Control Variable price is
narrowly fluctuating +/- the price mode or Reference State. This
continues for the first 20 seconds of the animation covering almost three
hours of trade, which implies an alignment between the Reference State
and the Reference Signal.

Does that help?

> Then a big BUYER disturbance occurs as evidenced by range
> extension or expansion.

Ah. It must be data; you are guessing that there was a
disturbance (to what variable; the range of the distribution?)
because it was somewhat effective. Is that right?

Yes. I think we may be on the same page.

> Continue to watch as the DDDs

What are the DDDs?

Dynamic Data Distributions is what I call the animation of price data
distributions

> Eventually the CV is controlled more effectively at the end
> of the sample

What is the CV?

Control Variable

> To hopefully answer your question Mark

Rick

Oooops. Sorry Rick.

···

> I can create DDDs for the same time period for any one of
> the 30 stocks and you would observe similar cycles of states

Again, what's a DDD?

*
**
***
*****
***
**
*

The above represents the first distribution (A) of price for x minutes. (In
the animation 60 minute *.bmp images, windows or snap shots.)

*
**
***
*****
******
*******
******
****

The above represents the second distribution (B) of price for x minutes.
A is 09:00 - 09:59, B is 09:01 - 10:00 and C would be 09:02 - 10:01,
etc. String them together in the animation, about 600 distributions in the
CAS.anm file and you have what I call the Dynamic Data Distributions.

> Is this making any sense?

Not yet. My main problem is that I don't know what variable you
think is controlled and I don't know _why_ you think it is
controlled.

Best

Rick

Ugh. I appreciate your laboring with my PCT ignorance here. Price is
the variable I believe is being controlled and I say so based on years of
studying and trading these price distributions and comparing what I
observe of the price distribution patterns with your tracking demo

http://home.earthlink.net/~rmarken/ControlDemo/BasicTrack.html

and Bill's tracking DOS demo too. Having played with these, and other
of your java demos, some _feel_ like they could be models of what goes
on in the mkts; trading crowds of human control systems all competing
for control of price. An interesting aside. The fellow who developed this
price distribution mkt analysis back in the late sixties, called the mode
of his price distributions the "point of control". <g>

Any better?

I suppose I'm wondering if, in your opinion, I've discovered (<g> so to
speak) that the markets are in fact perceptual control systems and if
so, is there potential and mutually beneficial collaboration possible that
may shed light on both HPCT and market understanding.

Thanks,
nth

[From Bruce Gregory (990827.1535 EDT)]

Norman Hovda (990827.1045 MST)

Thanks for the clarification. Would this make sense then, assuming I
understand the distinction you're making?

If the observed Reference State is price X, and Greenspan is talking or
the Fed is about to make an announcement, the Reference Signal for
many mkt participants (crowd) may be disturbed in such a way as to be
observed as a change in the Reference State price X +/- Y?

You are saying that the price of a stock is analogous to a controlled
variable. To the extent the price is relatively stable, it corresponds to
the reference state for this variable held by a majority of those trading in
the stock. The announcement by the Fed leads the traders to adopt new
reference levels for the stock. If the stock is trading below his or her new
reference level, he or she buys. If the stock is trading above his or her
new reference level, he or she sells. You observe that the stock price then
reaches a new state reflecting the aggregate new reference level.

I suppose I'm wondering if, in your opinion, I've discovered (<g> so to
speak) that the markets are in fact perceptual control systems and if
so, is there potential and mutually beneficial collaboration possible that
may shed light on both HPCT and market understanding.

If I understand you, you have demonstrated in the market a phenomenon some
have called the collective control of a perception. I'll leave it for them
to comment.

Bruce Gregory

[From Norman Hovda (990827.1320 MST)]

[From Bruce Gregory (990827.1535 EDT)]

Norman Hovda (990827.1045 MST)

> Thanks for the clarification. Would this make sense then, assuming I
> understand the distinction you're making?
>
> If the observed Reference State is price X, and Greenspan is talking or
> the Fed is about to make an announcement, the Reference Signal for many
> mkt participants (crowd) may be disturbed in such a way as to be observed
> as a change in the Reference State price X +/- Y?

You are saying that the price of a stock is analogous to a controlled
variable. To the extent the price is relatively stable, it corresponds to
the reference state for this variable held by a majority of those trading
in the stock. The announcement by the Fed leads the traders to adopt new
reference levels for the stock. If the stock is trading below his or her
new reference level, he or she buys. If the stock is trading above his or
her new reference level, he or she sells. You observe that the stock price
then reaches a new state reflecting the aggregate new reference level.

"reference state" relates to collective and "reference level(s)" relates to
the indivdual?

> I suppose I'm wondering if, in your opinion, I've discovered (<g> so to
> speak) that the markets are in fact perceptual control systems and if so,
> is there potential and mutually beneficial collaboration possible that
> may shed light on both HPCT and market understanding.

If I understand you, you have demonstrated in the market a phenomenon some
have called the collective control of a perception. I'll leave it for them
to comment.

Bruce Gregory

Great. I'm all eyes.

Thanks for the input.

nth

[From Rick Marken (990827.1420)]

Me:

What are the DDDs?

Norman Hovda (990827.1045 MST)--

Dynamic Data Distributions is what I call the animation of price data
distributions

Ok. So what I think I'm seeing in your animation is the distribution
of prices of the _same_ set of stocks varying over time. I think you
are saying that some parameter(s) (mean, mode, variance?) of this
distribution is controlled.

I agree that stock prices are very likely to be a controlled
variable. But I don't think the DDDs tell us this; it's what we
know about stocks that tell us this. Buyers want to pay as little
as possible for stocks and sellers want to charge as much as
possible. So buyers and sellers are controlling for stock prices
relative to different reference signals (in themselves); when
the reference signals of buyer and seller match there is a sale.
According to HPCT, buyers and sellers adjust their references for
stock prices as the means of controlling higher level perceptions
(like the perception of getting rich or appearing wise). So
your model of stock price control must include at least two
levels of control in both buyers and sellers.

I suppose I'm wondering if, in your opinion, I've discovered
(<g> so to speak) that the markets are in fact perceptual
control systems

Yes. Markets (which are collections of people who act in the
role of buyers and sellers; each person is often acting in both
roles simultaneously) are collections of control systems; as
Bruce Gregory (990827.1535 EDT) noted, you are observing (in
your DDD animations) the collective control of perceptions (of
stock prices). The questions you have to answer now (in order
to build your model) are _which_ perceptions (besides stock
price) are being controlled by buyer and sellers and
_how_ are these perceptions controlled.

is there potential and mutually beneficial collaboration possible that
may shed light on both HPCT and market understanding.

Sure. I think a model of a market (like the stock market) would
be a great PCT project. You might want to participate in the
CROWD modeling group for a while to get a sense of how your
project might proceed. Your DDD data is very similar to the
crowd data (collective behavior of people in various public
places) that the CROWD program is designed to imitate.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates mailto: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[From Norman Hovda (990828.1230 MST)]

At 14:21 Richard wrote about Re: Aggregate Data on 27 Aug 99,

[From Rick Marken (990827.1420)]

Me:

> What are the DDDs?

Norman Hovda (990827.1045 MST)--

> Dynamic Data Distributions is what I call the animation of price data
> distributions

Ok. So what I think I'm seeing in your animation is the distribution
of prices of the _same_ set of stocks varying over time. I think you
are saying that some parameter(s) (mean, mode, variance?) of this
distribution is controlled.

Yes.

I agree that stock prices are very likely to be a controlled
variable. But I don't think the DDDs tell us this; it's what we
know about stocks that tell us this. Buyers want to pay as little
as possible for stocks and sellers want to charge as much as
possible.

Ideally. But execution seldom lives up to inspiration. There are many
perceptions (reasons, desires, wants, fears, taxes, etc.) that drive
initiating buy sell decisions and many more for closing a position.

So buyers and sellers are controlling for stock prices
relative to different reference signals (in themselves); when
the reference signals of buyer and seller match there is a sale.

I wouldn't put it like that. My appraisal of the mkt, as a seller, may
suggest a Ref Signal quite different from the buyer's Ref Signal. Then
again I'm confused about a Reference State v Reference Signal v.
Reference Level.

According to HPCT, buyers and sellers adjust their references for
stock prices as the means of controlling higher level perceptions
(like the perception of getting rich or appearing wise). So
your model of stock price control must include at least two
levels of control in both buyers and sellers.

I can separate buyer data from seller data but why at least two levels of
control for each? By "levels" are you referring to _hierarchy_ as shown
in Bill's model for controlling the inverted pendulum?

> I suppose I'm wondering if, in your opinion, I've discovered
> (<g> so to speak) that the markets are in fact perceptual
> control systems

Yes. Markets (which are collections of people who act in the
role of buyers and sellers; each person is often acting in both
roles simultaneously) are collections of control systems; as
Bruce Gregory (990827.1535 EDT) noted, you are observing (in
your DDD animations) the collective control of perceptions (of
stock prices). The questions you have to answer now (in order
to build your model) are _which_ perceptions (besides stock
price) are being controlled by buyer and sellers and
_how_ are these perceptions controlled.

This brings up the problem of the data I have are not the data I want.
The data I want are not the data I need. The data I need are not the data
I can get. The data I can get costs more than I want to pay.

I'm not convinced such an undertaking is necessary but if I can show it
would improve trading performance... hmmmm.

> is there potential and mutually beneficial collaboration possible that
> may shed light on both HPCT and market understanding.

Sure. I think a model of a market (like the stock market) would
be a great PCT project. You might want to participate in the
CROWD modeling group for a while to get a sense of how your
project might proceed. Your DDD data is very similar to the
crowd data (collective behavior of people in various public
places) that the CROWD program is designed to imitate.

I'm following along, peddling as fast as I can. Where can I download the
CROWD program?

Best

Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates mailto: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

Thanks for your encouragement.

nth

[From Bruce Gregory (990828.1539 EDT)]

Norman Hovda (990828.1230 MST)

Then
again I'm confused about a Reference State v Reference Signal v.
Reference Level.

I don't know anything about Reference States. The Reference Level is the
instantaneous value of the Reference Signal. You adjust the Reference Level
of the Reference Signal for a household thermostat by setting the register
to maintain a particular temperature (the Reference Level).

Bruce Gregory

[From Rick Marken (990828.1700)]

Me:

I agree that stock prices are very likely to be a controlled
variable. But I don't think the DDDs tell us this; it's what we
know about stocks that tell us this. Buyers want to pay as little
as possible for stocks and sellers want to charge as much as
possible.

Norman Hovda (990828.1230 MST)

Ideally. But execution seldom lives up to inspiration. There
are many perceptions (reasons, desires, wants, fears, taxes,
etc.) that drive initiating buy sell decisions and many more
for closing a position.

I think you've mentioned one possible higher order controlled
variable (taxes) that might be involved in stock buying and
selling (for some individuals). I think buyers (and sellers) set
their references for what they pay and get, respectively, for
stocks as the means of controlling higher level variables, like
taxes, liquid assets, transaction state (successful or not), etc.

I think a successful transaction occurs when the buyer and seller,
for whatever their seperate higher level reasons (higher level
perceptions they are trying to control), have the same reference
for the lower level "stock price" variable. The buyer, at some
moment, might want to pay, say, $120/share and the seller might
wants to get $120/share; in this case there will be a sale (a
successful transation) If the buyer's reference for stock price
is at $1/share and the seller's is at $1000/share there will
be no sale.

My appraisal of the mkt, as a seller, may suggest a Ref Signal
quite different from the buyer's Ref Signal. Then again I'm
confused about a Reference State v Reference Signal v. Reference
Level.

The reference signal is (according to PCT) a neural signal
in your brain, the magnitude of which specifies the reference
_state_ of some perceptual variable (in this case, the perceived
price of the stock). If you reference signal (as a seller)
specifies a stock price of $1000 (the desired reference state
of the stock price variable) and all the buyers have a reference
signal that specified a stock price reference state of $1 then
you are not going to sell the stock.

The stock market is a place where buyer and seller control systems
can get together and see what state of the variable stock price
of a bazillion different stocks is mutually acceptable (mutual
acceptability occurring, in PCT, when different control systems,
controlling the same variable, set the same reference signal
specification for the state of that variable).

I can separate buyer data from seller data

How do you do this? This would be very interesting data to see.
Is the seller data the average asking price for a stock over
time? Is the buyer data the average offered payment for the
same stock over time? This would be very intereting to see.
Wouldn't these curves lie on top of each other?

but why at least two levels of control for each? By "levels"
are you referring to _hierarchy_ as shown Bill's model for
controlling the inverted pendulum?

Yes. Levels of the hierarchy. I think you need some higher level
control systems to adjust the reference signal for stock price
as the means of controlling their higher level perceptions (of
liquid cash, taxes, for example).

I'm following along, peddling as fast as I can. Where can I
download the CROWD program?

Press "Crowd simulation" at the bottom of Bill Powers' home
page

http://www.frontier.net/~powers_w/INDEX.HTML

If your browser supports ftp and your are on a PC you are
in for a treat.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[From Norman Hovda (990828.1140 MST)]

[From Rick Marken (990828.1700)]

[snip good stuff]

The reference signal is (according to PCT) a neural signal
in your brain, the magnitude of which specifies the reference
_state_ of some perceptual variable (in this case, the perceived
price of the stock). If you reference signal (as a seller)
specifies a stock price of $1000 (the desired reference state
of the stock price variable) and all the buyers have a reference
signal that specified a stock price reference state of $1 then
you are not going to sell the stock.

For the "here and now" what's missing in this example is the current
mkt price. Neither will prolly participate given their respective Reference
Signals and Ref States. Without a current mkt price, neither party has
any relationship; nothing but an abstract value.

Let's say the mkt is trading $500 +/-. What if I, as seller, surmise a
Reference Signal for a _future_ price of a $1 and you, the buyer, hold a
Reference Signal for the _future_ price of $1000. So, in your language,
we may share a Reference State that $500 is a desireable and
acceptable opportunity for both buyer and seller? Voila, a trade with
each participant having a different Ref Signal, no?

How'm I doing?

The stock market is a place where buyer and seller control systems
can get together and see what state of the variable stock price
of a bazillion different stocks is mutually acceptable

ok...

(mutual
acceptability occurring, in PCT, when different control systems,
controlling the same variable, set the same reference signal
specification for the state of that variable).

Hmmmm... "controlling the same variable" gives me pause. Would you
say that about football or basketball team not in posession of the ball?
Two or more parties may seek to control the same variable, but can
they all control it at the same time?

> I can separate buyer data from seller data

How do you do this?

Easier said then done but prolly off topic. <g> Perhaps a better
understanding of PCT will help me in this regard.

This would be very interesting data to see.
Is the seller data the average asking price for a stock over
time? Is the buyer data the average offered payment for the
same stock over time? This would be very intereting to see.
Wouldn't these curves lie on top of each other?

Averages have nothing to do with buyer v. seller analysis. It has to do
with competing time frames IMO.

> but why at least two levels of control for each? By "levels"
> are you referring to _hierarchy_ as shown Bill's model for
> controlling the inverted pendulum?

Yes. Levels of the hierarchy. I think you need some higher level
control systems to adjust the reference signal for stock price
as the means of controlling their higher level perceptions (of
liquid cash, taxes, for example).

Xlent. I conceptualize this as a function of separating out different
timeframes; larger timeframes tend to dominate being higher up the
hierarchy (perhaps relating to "intrinsic" variables?). However, every
participant regardless of timeframe Reference Signal, must trade in the
shared present tense, the playing field (free mkt exchange) where every
timeframe competes for control of price; for possesion of the ball so to
speak.

> I'm following along, peddling as fast as I can. Where can I
> download the CROWD program?

Press "Crowd simulation" at the bottom of Bill Powers' home
page

http://www.frontier.net/~powers_w/INDEX.HTML

If your browser supports ftp and your are on a PC you are
in for a treat.

Got'em. Planning to spend some time with them this week.

Thanks much,

nth

[From Rick Marken (990829.1830)]

Norman Hovda (990828.1140 MST)--

For the "here and now" what's missing in this example is the
current mkt price.

As I understand it, the market price at any time during the
trading day is the price at which the stock sold (for some number
of transations or within some time window; I don't know). So
the market price is the state if a controlled variable (stock
price) which some proportion of the buyers and sellers want
(have set their reference signals for) at each instant during
the trading day.

Time variations in market price trace out where the buyer/seller
mutually agreed on reference state for stock price is at each
point in time. Actually, market price is the variable whose behavior
you would want to explain with a control model; I imagine you would
explain this behavior by modeling buyers and sellers as input
control systems controlling their perceptions of stock price
relative to a varying reference for stock price) as the _means_ of
controlling their own higher level variables; the most obvious
possibilities for controllled higher level variables in this model
(that I can thing of at the moment) are 1) available liquid cash
(can't buy more than you can afford) 2) rate of growth in non-liquid
assets and 3) growth potential of company whose stock is being
bought or sold.

Let's say the mkt is trading $500 +/-. What if I, as seller,
surmise a Reference Signal for a _future_ price of a $1 and
you, the buyer, hold a Reference Signal for the _future_
price of $1000. So, in your language, we may share a Reference
State that $500 is a desireable and acceptable opportunity
for both buyer and seller? Voila, a trade with each participant
having a different Ref Signal, no?

No. The reference signals of both buyer and seller specify
a reference state for stock price of $500. If you have a
reference signal specifying a future price you will be sorely
disappointed becuase you cannot control stock price (legally).
That's why people often have to vary their reference signal
specification for stock price in order to complete a transation;
the buyer can't control the price (get it to some particular
reference state) because she can't force the seller to sell it
at that price; same for the seller, who can't force the buyer
to buy at the seller's desired price.

The buyer and seller might end up setting compatable references
for stock price becuase of what they _imagine_ the stock will
do in the future; but the transation won't happen unless the
buyer and seller agree on (set the same reference signal
specification for) stock price. That's not really a theoretical
point; that's how the stock market works.

Hmmmm... "controlling the same variable" gives me pause. Would
you say that about football or basketball team not in posession
of the ball?

Yes. Of course. Sports is all about seeing which team can force
one variable (the score) to match the reference signal shared by
each member of the team (unless the fix is in). When the Dodgers
play that Cards I imagine that every Dodger wants the score at the
end of the game (the controlled variable) to be in the state
"Dodger score > Card score"; and every Card wants the same variable
to be in the state "Card score > Dodger score".

Two or more parties may seek to control the same variable,
but can they all control it at the same time?

No. There is a conflict (a uniquely control system phenomenon).
There is conflict in the stock market because the seller wants
the "stock price" variable to end in a different state than does
the buyer.

Got'em. Planning to spend some time with them this week.

Great. Pay particular attention to how the model individuals
actually work. I think you really have to be very clear about
what a perceptual, reference and error signal are, and what
an action, controlled and disturbance variable are. Read
Behavior: The Control of Perception and try to write up some
working models.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[Norman Hovda (990829.2315 MST)]

[From Rick Marken (990829.1830)]

> Got'em. Planning to spend some time with them this week.

Great. Pay particular attention to how the model individuals
actually work. I think you really have to be very clear about
what a perceptual, reference and error signal are, and what
an action, controlled and disturbance variable are. Read
Behavior: The Control of Perception and try to write up some
working models.

Best

Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

Ordered last week. It's in the mail.

nth