|
PureBytes Links
Trading Reference Links
|
Donald and List Members,
One of the worst thing we can do is to fall into the "statistics trap".
With respect to CYCLES, cyclic phenomena in the market is
typically short in duration of an observed AMPLITUDE even
though observation of period length might persist for a much longer
time.
I'm not the least bit interested in validity, statistically speaking, but
in validity of what happens in the market in the next time period of
the cycle that I am analyzing.
At most, to get a reasonable approximation of what influence a
particular cycle might have on the market over some time period
forward, my observation is that a period of twice the length of the
cycle under question is normally sufficient for longer periods and
never is it necessary to have more than 5 cycles of data for the
shorter periods.
EMPIRICAL OBSERVATIONS -- NOT STATISTICS.
Clyde Lee
Donald Thompson wrote:
> Clyde,
>
> I noted the fourier wave work that you posted. One conclusion is that
> the composite implies price trend is down. My only question and it is
> also put to the rest of the group.
>
> What is the appropriate amount of data needed to find a cycle of any
> given length?
>
> In statistics I have found one referance, that if one wanted to get a
> decent standard deviation for one year that ten years of data is needed
> to get a SD of statistical credibility.
Attachment Converted: "c:\eudora\attach\vcard141.vcf"
|