|
PureBytes Links
Trading Reference Links
|
On Sep 29, 2:22pm, Monte C. Smith wrote:
> Subject: Re: Exponential moving average
>
> Bob Fulks wrote:
> > At 5:52 PM -0700 9/28/00, Monte C. Smith wrote:
> > >Mark Jurik wrote:
[...]
> >
> > The simple moving average, the exponential moving average,
> > and the weighted moving average have no overshoot. More sophisticated
> > moving averages such as the T3 average and the Jurik moving average
> > can have overshoot as this makes them respond faster. On the moving
> > averages I design for myself, I usually include a parameter for this.
[...]
> > An exponential moving average and the weighted moving average put
> > more weighting on the most recent data and less and less on older
> > data. This makes them respond faster to changes, eliminates the
> > 20-bar null problem, and eliminates the problem with the false
> > response as old data drops off of the back of the filter. As a result
> > of the faster response, these tend to look noisier.
> >
> > The so called "end point moving average" is simply the end of a "best
> > least squares fit" linear regression line through all of the data
> > points. It has no lag at all on a sloping price line but overshoots
> > horribly at a turn. It also tends to be pretty noisy.
>
>
> It seems that some of these problems, such as lag and overshoot, could
> be put to good use, even though they are not desirable characteristics
> in many instances. The fact that the End-point MA overshoots drastically
> at turns might be used with an XMA to generate a raw signal. I've also
> "discovered" how the lag in an MA can be used to determine cycle period.
Sounds interesting. If you don't mind sharing the "discovery", would
you please explain how you use the MA lag to determine the cycle period?
[...]
> >
> > Then there are the adaptive moving averages that adapt their
> > characteristics dynamically to try and optimize the tradeoffs. These
> > have names such as the Kaufmann AMA, Kalman filters, Tushar Chande's
> > VIDYA, and the Jurik AMA. Some are very useful.
>
>
> Having talked about, or at least mentioned, some of the wide array of
> moving averages, I'll have to ask at this point if you or anyone on the
> List have any observations concerning the Volume Adjusted Moving
> Average, in which time does not play a role. The weighting is done
> 'dynamically' based upon volume. There is a description at
> http://trends-online.com/ind15691.htm if you or anyone else cares to
> have a look at it.
We've discussed this before, a little. Some people think that data
points with heavier volume should be given heavier "weight" in the
the MA. I note that reversals often happen on/near high volume bars,
and would argue that high volume bars should in fact be _negatively_
weighted, as a rule. Just a thought - haven't put it to the test.
As far as adpative avearges go, which vary the "speed" or look-back,
a similar question arises: should the MA place _less_ weight on
large moves, or _more_ weight? The Kaufman MA, for example assumes
that a big move implies "noise" and thus gives it less weight.
The univariate Kalman averge would give a big more more weight,
in an effort to minimize tracking error.
|