Re: calculation of price/frequency distributions

["Carroll Slemaker" <cslemaker1@xxxxxxxx>]

Thanks for the reply.  I must respectfully disagree, however, with your
statement that this kluge is the only thing that can allow real-time results
to match historical results.  First of all, this kluge does  NOT  cause ALL
real-time results to match historical results as many on this list have
often complained.  The only way to guarantee such consistency would be for
the historical run ALSO to use the tick data.  And if the tick data have not
been saved, then one must face the consequence that it is logically
impossible to duplicate real-time results, in all cases, using abbreviated
data.

Actually, the best solution would be to provide the user with two types of
variables and let the user decide which behavior he desires.  But,
especially in the absence of documentation of this "feature", a programmer
has the right to expect that a variable will retain its value until he, the
programmer, decides to change it.

Regards,
Carroll Slemaker


----- Original Message -----
From: Howard Jackson <hrjf4@xxxxxxxxx>
To: Carroll Slemaker <cslemaker1@xxxxxxxx>; Tim McCaughey <TimM@xxxxxxxxxx>;
<omega-list@xxxxxxxxxx>
Sent: Tuesday, July 27, 1999 11:22 AM
Subject: Re: calculation of price/frequency distributions

<snip>
>
> and Carrol, that 'kludge' is the only thing that can
> allow indicators that are updating real time to match
> historical results. As far as i am concerned it works
> fine and it is very consistent with the rest of the
> program's behaviour... what it seems like you want is
> the never-implemented mutliple data charts with tick
> bars... something that should have been out a long
> time ago.
>
> H
>
> --- Carroll Slemaker <cslemaker1@xxxxxxxx> wrote:
> > Unless they have fixed this since I last tested it
> > long ago (not likely,
> > since Omega considered it a "feature"), you will not
> > be able to do this.
> > From your description of the problem it seems that
> > you want to collect data
> > on a tick-by-tick basis (increase a tick counter for
> > each possible price
> > when a tick occurs at that price).  I assume that it
> > is still not possible
> > to attach an indicator or system to a tick chart, or
> > to produce a
> > multi-symbol tick chart.  If this assumption is now
> > wrong, please disregard
> > the following.
> >
> > Unfortunately, EL "variables" behave like no
> > variables in any programming
> > language you have ever seen.  In a reasonable
> > language, once you set a
> > variable to a value, it retains that value until you
> > set it to a different
> > value.  But not Omega's EL - that is, not if you are
> > updating these
> > variables in a time-bar chart on a tick-by-tick
> > basis.  A variable WILL
> > update correctly on each tick, but at the end of the
> > bar's time slot the
> > value will be "restored" to some prior value and
> > there is no way for you to
> > inhibit this "feature".  This was actually a kluge,
> > undocumented, to cause a
> > real-time run to match results of a run on
> > historical data where only bar
> > values are considered, not individual ticks.
> >
> > Carroll Slemaker