Re: [GT] DSS [fixed]

["Robert A. Schmied" <ras AT acm.org>]

i was about to commit dss.pm when i discovered there was a newer version.
after messin' about with it some i have a couple of questions and or
comments, mostly of the trivial, non-technical variety.


I)
array @NAMES does not include the last 3 indicator series
(e.g {I:Prices HIGH}, {I:Prices LOW}, {I:Prices CLOSE} as ,#4,#5,#6).
as a result
%  display_indicator.pl I:DSS CIEN
shows only
DSS-BLAU[5,7,3]     [2007-05-24] = 93.2404
...

instead of
DSS-BLAU[5,7,3,{I:Prices HIGH},{I:Prices LOW},{I:Prices CLOSE}][2007-05-24] = 93.2404
...

was that omission intentional?

personally i frequently use display_indicator.pl to 'see' the default
arguments when they are not shown in the pod, and sometimes when the are too.
but it does make for a very long output line ...


II)
i don't think this pod sentence provides any useful information and
recommend it be removed or re-written to be useful (it is likely a
cut/copy/paste hold-over):
"The handling and the calculations of signals is similar to the Stochastic-Indictor."

regardless the Indictor typo will be fixed.


III)
i have also added the following sentence to the 'High, Low, and
Close of Source' pod item.
"By default these are {I:Prices HIGH}, {I:Prices LOW}, {I:Prices CLOSE} respectively."



IV)
topic for general ta discussion. the question of 'which data point' to use
in an indicator comes up from time to time, as it has here and again in
the aroon indicator. in the aroon case the user is able to alter the default
values by specifying alternatives, in this instance, unfortunately the user
has no option (that i can see), the algorithm is fixed.

so the questions include:
is there a place for an alternate dss that implements the alternate algorithm?
or can on dss module easily accommodate the two different algorithms?


aloha

ras


Weigert, Thomas wrote:
> I have made some updates to the DSS as rewritten by kw.
>
> 1. The computation of the "period low" and "period high" really refer to
> The lowest low and highest high, I believe. Therefore, this indicator
> should take 3 series as (default) arguments: {I:Prices HIGH}, {I:Prices
> LOW}, and {I:Prices CLOSE}.
>
> 2. The high, low, and close are all used in the computation of the
> stochastic.
>
> 3. The dependency should just be 1 day on the smoothing intermediate
> series, as the dependency mechanism takes care of the rest. This is
> important, as the EMA gives quite different results if a longer
> dependency is requested (in effect, it lengthens the smoothing period).
>
> 4. I added a check in the calculation of DSS to avoid divide by zero
> errors.
>
> 5. Added a check for constant input.
>
> Th.

<< big snip >>