Re: [GT] SEMA

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

Weigert, Thomas wrote:
> This is probably not necessary as a separate indicator. I did that more
> as practice... Th.

humm, well a lot of this indicator discussion and correction came about
around the time i said (wrote) something about corner cases. sema doesn't
have a corner case in that if you want to graph the (S)EMAs of MSFT, AAPL,
IBM, T ... starting with the first date in your particular database you
can do it. but the other ema just gives you a completely blank graph from
start to end.

and that is a good segue into the question "why does an indicator 'abort'
-- completely stop testing for enough prior history as time moves onward?"

it would seem to me there ought to be a way to retry "are the dependencies
satisfied now?" periodically as time proceeds. i just don't have a complete
understanding of the interaction among the methods.

it seems to me (i don't know why) that SMA operates 'nicely', no abort,
but also no output until it has the needed history to compute results.

aloha

ras

ps -- i think SEMA is good, and unless you kibosh it, will commit it (someday)
-- ras


> > -----Original Message-----
> > From: Robert A. Schmied [mailto:ras
AT
acm.org]
> > Sent: Wednesday, March 12, 2008 12:37 PM
> > To: devel
AT
geniustrader.org
> > Subject: Re: [GT] SEMA
> >
> > Robert A. Schmied wrote:
> >
> > > thomas
> > >
> > > i think SEMA.pm is new, alternate for EMA.pm. but i'm not sure.
> > > bit confused here.
> > >
> > > ras
> >
> > duh! rtfm or pod in this case ;-)
> >
> > sema is starting ema (special ema), starts outputing results
>
> immediately
>
> > (or nearly so) and quickly converges with ema.
> >
> > ras