|
PureBytes Links
Trading Reference Links
|
------------------------------------------------------------------------
You cannot reply to this message via email because you have chosen not
to disclose your email address to the group.
To reply: http://groups.yahoo.com/group/equismetastock/post?act=reply&messageNum=5277
------------------------------------------------------------------------
James,
Thanks for sharing this information!
Preston
--- In equismetastock@xxxx, jdltulsa@xxxx wrote:
> Someone Wrote.....
>
> "Might it be possible to somehow extrapolate the latest cycle from
a shifted
> indicator into today's data? Has anyone done any work on this
concept? "
>
> I have done extensive forecasting work using Hurst's Centered and
Inverse
> Moving Averages. I have attempted these forecasts using multiple
regression
> and ARIMA models with little success. However, the problem I
discovered,
> which is well documented in other readings, i.e., Channels &
Cycles: A
> Tribute to J.M. Hurst by Brian Millard (a good book by the way), is
that you
> cannot arbitrarily choose any length moving average to shift. If
you do,
> your choice could be completely out of phase with the dominant
cycle and
> market. This of course can cause financial disaster. What you
must first do
> is develop an approach that allows you to isolate and identify the
current
> dominant cycle in the time series under investigation.
>
> To select the dominant cycle you could rely on Fourier analysis or
John
> Ehler's MESA, or you could perform a decomposition of the time
series
> (stocks, futures, mutual funds, etc.). Basic theory suggests a
time series
> is comprised of four components, namely,
>
> Trend
> Seasonal
> Cyclical
> Random
>
> You can start by detrending the series. There are a number of
approaches
> that can be used, e.g., an Inverse Moving Average (Close - Centered
Moving
> Average), some momentum function, etc. However, the selection of
the look
> back period is the most critical choice to make at this point.
After
> detrending, you are left with a new time series that only has the
Seasonal,
> Cyclical and Random components. Next, if you are ready for more
math, you
> can remove the seasonal component. Most basic forecasting or
Econometric
> texts can give you guidelines on isolating and removing seasonal
variations
> in a time series. If you make it to this point, subtract the
seasonal
> component from the detrended series and you are left with the
Cyclical plus
> Random components.
>
> I hope this is enough to stimulate some new thought and
appreciation to the
> problem your trying to solve. Like you, I hope others can
contribute to the
> dialog.
>
> James
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Sell a Home for Top $
http://us.click.yahoo.com/RrPZMC/jTmEAA/jd3IAA/BefplB/TM
---------------------------------------------------------------------~->
To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|