|
PureBytes Links
Trading Reference Links
|
Hi David
No problem
Time series data is any data that has 'regular' dates. Your downloaded H,L,C
prices that you use in Metastock are time series data. Monthly Industrial
production data from Northern France is time series data. They both contain
"hidden" elements.
Economists who work in the area of econometrics have developed methodology
and software programs to specifically "deal with" the unobserved components
in the time series data. If they don't deal with it, it can mess up their
results pretty quickly.
You have lots of indicators in Metastock that supposedly do the same thing,
only most Metastock users don't have a clue what the indicators are really
doing ... or why they are doing it ... or if the indicators are really doing
what they are suppose to do.
Most users put together systems ad-hoc ... a little bit of this, an idea
from TASC, anything with a Fibonacci number, a moving average etc. Nothing
is developed from a consistent methodology or from the knowledge of the
statistical characteristics of the price data.
The unobserved components of the time series data (price data) are trend,
seasonality, cyclicality, and noise or random shocks. The various time
series programs try to separate out these 4 unobserved components to arrive
at the "core" or adjusted data.
Some people use the unobserved "bits" in their trading systems, i.e.,
"trend" ... others use the corrected data to get a "true picture" of price
movement and trading system results.
The "big news" is that previous "seasonal adjustment" software, i.e., BV4
from Germany, X-11 and X-12 from the US Census, SABL from Bell Labs, etc.
were geared to specialist use and monthly/quarterly data.
Now TRAMO/SEATS originally from the Bank of Spain and now from Eurostat is
available. It is Windows user friendly for daily and weekly data which is
inputted directly from Excel.
So we have three linkable and powerful programs to use in our trading:
Metastock, Excel and Demetra.
Best regards
Walter
----- Original Message -----
From: "David Bozkurtian" <dbozkurtian@xxxxxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: Wednesday, January 26, 2000 11:58 AM
Subject: Re: stuff
| Walter,
|
| Can you clue me in as to what you guys are talking about?
|
| Thanks
|
| David
|
|
| From: "Walter Lake" <wlake@xxxxxxxxx>
| Reply-To: metastock@xxxxxxxxxxxxx
| To: "Metastock bulletin board" <metastock@xxxxxxxxxxxxx>
| Subject: stuff
| Date: Wed, 26 Jan 2000 10:55:03 -0500
|
| Thanks for your emails and the autocorrelation stuff
|
| "... I went crazy trying to get the X11 to do anything. Demetra does so
much
| I am still trying to figure it out. It is a great way to learn about time
| series analysis. They also offer so much documentation for free. If this
was
| in book form it would be very expensive. And once you figure out how to
use
| this you can start putting out your own econometric forecast's. <G> With
all
| the
| data you can get from the St Louis Fed this is a very powerful
combination.
| ..."
|
| ============
|
| I'll stick with the components of the seasonal adjustments. I'm sure that
| the S&P boys will love all the St.Louis Fed stuff. I'm just stunned at the
| automated processing power that is there ... compared to the skill and
| knowledge required to manipulate all those Moving Averages in X-11 and
X-12.
|
| What I'm really enjoying now, is choosing the model from the statistical
| analysis of the price data, then running it back into Metastock. At least
| now I know what I'm dealing with.
|
| My old way of running different systems ad-hoc on the data in Metastock
then
| doing the analysis seems very haphazard now.
|
| Will keep plowing ahead on the Math, my Junior level Math skills are still
| having a hard time with all those funny looking signs.<G>
|
| Best regards
|
| Walter
|
|
| ______________________________________________________
| Get Your Private, Free Email at http://www.hotmail.com
|
|