|
PureBytes Links
Trading Reference Links
|
Lionel,
> Some time ago I tried to use Metastock Fourier analysis. The results were
> poor as the dominant cycle was always close to the time interval of the
> displayed data.
This is my perception as well of what Metastock Fourier analysis can do. To
be fair, I should add that the sensitivity to the time interval is actually
inherent in Fourier analysis, whether it's financial series or anything
else, by Metastock or some university.
Without elaborating much, I think an analogy with simple averaging versus
EMA is very much appropriate here: SMA is sensitive to what happens at the
far end of the time interval. This is why one tends to choose EMA when
dealing with long time intervals. So what we would need here instead is a
sort of "exponentially modulated" Fourier analysis. I don't know what it
really can be :-) .
Cheers, Vitaly
----- Original Message -----
From: "Lionel Issen" <lissen@xxxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: Wednesday December 13 2000 11:06 AM
Subject: Re: cycles
> When Peters books were reviewed after they came out, the main criticism,
by
> the reviewers, was that they are mathematically demanding and they do not
> provide a trading method. Peters' books are interesting and he has enough
> text so that a non-mathematician can sort of understand the thrust of his
> ideas.
>
> Peters references Hurst's paper (Hurst numbers) which was published in the
> Transactions of the American Society of Civil Engineers around 1942. I
> looked up his paper and I suspect that Hurst must have been as much of a
> statistician as he was a hydraulic engineer. Considering the state of the
> art of statistics back in the 1930's, Hurst's paper is most impressive.
>
> Some time ago I tried to use Metastock Fourier analysis. The results were
> poor as the dominant cycle was always close to the time interval of the
> displayed data.
> Lionel Issen
> lissen@xxxxxxxxx
> ----- Original Message -----
> From: "Jeff Haferman" <haferman@xxxxxxxxxxxxxxxxxxxxxxx>
> To: <metastock@xxxxxxxxxxxxx>
> Sent: Tuesday, December 12, 2000 4:10 PM
> Subject: Re: cycles
>
>
> >
> > Hi Dave,
> > I've never tried using the Metastock Fourier stuff on an
> > exploration, like you I've just done charting with it. I'm
> > involved with refining my own systems, so I have not devoted
> > anytime to this.
> >
> > As for the fractal approach, Edgar Peters books are very good
> > at explaining the underlying ideas (but the books are not
> > for the mathematically faint-of-heart). The *basic* idea is
> > that market prices are cyclical, but non-periodic. A practical
> > example from nature of a non-periodic cycle is the sunspot
> > cycle... see for example
> > http://www.sunspotcycle.com
> > http://science.nasa.gov/ssl/pad/solar/predict.htm
> > or especially the plot at
> > http://www.sunspotcycle.com/images/zurich.gif
> >
> > Anyway, Peters describes how to use "rescaled range" analysis
> > (also called "R/S" analyses) to determine cycle lengths for
> > the sunspot cycle, and then moves on to the stock market.
> >
> > I can go into more detail, but the bottom line (as far as
> > I'm concerned) is that while Peter's books are extremely
> > fascinating, there is nothing in them that will help people
> > put together a practical trading system unless you are
> > extremely creative. That being said, there is a website
> > devoted to Edgar Peters at
> > http://oara.org/mpc/fma
> >
> > Some programs that implement some details of "R/S" analysis
> > have been put together by John Conover and are available at
> > http://www.johncon.com/ntropix
> >
> > I've been using the programs at the above site for the last
> > couple of years, and while I have a much better understanding
> > of the theory of fractal "behavior" of markets, I have come
> > to believe that as a practical matter, T/A concepts such
> > as support and resistance, momentum, and price/volume
> > relationships are easier to understand and implement, and
> > can also produce profits.
> >
> > Sorry for the long-winded reply.
> >
> > Jeff
> >
> >
> > Dave Nadeau wrote:
> > >Jeff,
> > >
> > >I'm a big fan of Fourier Analysis, as I've used it in other
> > >work. But I have been unsuccessful in getting it to work
> > >in anything other than a MetaStock chart. In an
> > >exploration, any variations I've tried or number of bars
> > >loaded has produced a result of N/A. I'm wondering if you
> > >have any ideas or have had any success working around this.
> > >
> > >Can you elaborate on the fractal approach? It sounds like
> > >an interesting piece of work. I will look up the book as
> > >well.
> > >
>
>
|
