PureBytes Links
Trading Reference Links
|
Donny,
I totaly agree that because of this none of the
exponential stuff (or anything that involves a
counters or accumulators) will work unless you let the
program run for X bars.
Still, I don't use any of that stuff for my trading so
the program works decently enought for me. It would be
very nice if the program stepped through a number of
bars, though.
But I will say this: ANYTHING you do on your computer
-when talking data analysis - you could do by hand
with a piece of paper, a pencil, a ruler and a lot of
time. Only the computer does it much faster. If you
were calculating these indicators yourself, your would
have to start at some bar, and you would run into the
exact same problem you described the software has for
the first X bars, and you would not be able to do
anything about it other than use another formula to
calculate your averages.
And if you can write another formula to calculate your
averages by hand correctly from bar 1, you should be
able to do it in EasyLanguage as well. So just go do
it -if you can- and stop whining...
For example, if my indicator's formula is:
Indicator's value of yesterday + 1
Then what am I going to use on the first bar I start
calculating? If you do this in a computer, by hand, on
a brown paper bag (yes, I've been in this list for a
couple of years) you will still have a problem in bar
1.
IMHO If you don't know what your indicators are doing
(and how they are calculated) you probably shouldn't
be trading with them...
H
--- DonC <countach@xxxxxxxxxxxxxxxx> wrote:
> BZZZZZTT ! Wrong, Howie. In fact you have proven
> my point.
> Some indicators, such as exponential averages,
> require long
> periods to stabilize. Using the minimum bars back
> guarantees
> faulty results.
>
> AND... TS4 does not accumulate variables based on
> variables
> during the MaxBarsBack period. (I will bet that
> Omega did
> not change this in TS2K.) More unpredicatble
> results.
> Unless you spend a lot time testing each of your
> indicators.
>
> (I explained this less than a week ago. How long
> have you
> been on the list, Howie?)
>
> Thanks for the info, though. I was thinking that
> RadarScreen
> by itself might be useful. Now I see it would just
> be a lot
> more wasted time to discover the "features" and
> figure out
> work-arounds.
>
> donc
>
>
> Howard Jackson wrote:
> >
> > what is all this BS about RS?
> > Wrong theory there, and unjustified bash to go
> along
> > with it...
> > what RadarScreen does is apply the indicator ONLY
> to
> > the last bar of the "chart". So if you have a 10
> bar
> > moving average then RS will load 10 bars and run
> the
> > indicator through them, getting the value of the
> > average for those 10 bars but it will NOT step
> through
> > old data. Haven't you noticed that there is no
> "First
> > and last date" or "days back to load" option
> anywhere?
> >
> > This I discovered after working with the program
> and
> > EZL support for a while. Apparently they are
> trying to
> > save us from ourselves, not letting the program
> load
> > tons of historical data for hundreds of symbols
> all at
> > once...
> >
> > And by the way, bouncing ticks is a great
> feature...
> >
> > H
> >
> > --- DonC <countach@xxxxxxxxxxxxxxxx> wrote:
> > > >Subject: Re: RadarScreen Help
> > > > Date: Tue, 13 Jul 1999 19:13:07 -0400
> > > > From: "Richard Park"
> > > <rnpark@xxxxxxxxxxxxxxxx>
> > > > To: "Sanders, Jeff"
> > > <JSanders@xxxxxxxxx>, "Omega-List (E-mail)"
> > > <omega-list@xxxxxxxxxx>
> > > >
> > > >Jeff,
> > > >
> > > >You need to check with Omega on this, but I
> believe
> > > the reason is that
> > > >RadarScreen does not save the values of
> variables
> > > at the end of a bar.
> > > >They get reset to zero at the end of each bar.
> > > Thus your trend count
> > > >could never be more than one.
> > > >
> > > >Best regards,
> > > >
> > > >Dick
> > >
> > > Another great design "feature" of TS, along with
> > > bouncing ticks,
> > > alert-on-every-tick,
> > > don't-save-variables-during-MaxBarsBack,
> > > and
> don't-save-the-day's-open-in-case-of-a-crash.
> > >
> > > donc
> > >
>
|