|
PureBytes Links
Trading Reference Links
|
Hi Jose,
The small errors compound and eventually become unacceptable. LRS is the real problem, where values
on a 2500 bar chart can be off by 20% or so. My belief is that a calculation based on just the
required number of periods is not going to be subject to the distortions that arise from using
Cum().
Thanks for your efforts, but this is not the solution that I'm looking for.
Regards
Roy
> Roy, both LinRegInd and LinRegSlp plot within a fraction of
> MetaStock's LRI and LRS values on my charts, minor inaccuracies
> courtesy of the Cum(1) errors accumulated by MetaStock.
> Applying the same exact code with an external dll which calculates
> values using double-precision numbers, the LRI & LRS values match
> precisely.
>
> So basically, it just comes down to replacing Cum(1) to get around
> MetaStock's inadequacies. ;)
>
>
> jose '-)
>
>
> --- In equismetastock@xxxxxxxxxxxxxxx, "Roy Larsen" <rlarsen@xxxx>
> wrote:
> > Hi Jose
> >
> > > Roy, maybe this will help:
> >
> > No, not at first glance anyway. Neither LinRegInd nor LinRegSlp
> appear to agree with the standard
> > MetaStock LR and LRS values, but besidesthat I want to get away from
> using Cum(1). Tom has
> > demonstrated (if I understand his code correctly) that it's not
> necessary to cumulate anything from
> > bar one. A 21 period LR only needs the 21 most recent bars of data
> to calculate, surely. I reason
> > that LRS should not need any more bars than that either.
> >
> > Thanks
> >
> > Roy
> >
> > > =======================
> > > Lin Reg Indicator/Slope
> > > =======================
> > > ---8<---------------------------
> > >
> > > { User inputs }
> > > plot:=Input("[1]LinReg Indicator, [2]LinReg Slope)",1,2,1);
> > > pds:=Input("LinReg periods)",2,2520,21);
> > > x:=Input("use Open=1, High=2, Low=3, Close=4, Volume=5",1,5,4);
> > > x:=If(x=1,O,If(x=2,H,If(x=3,L,If(x=4,C,V))));
> > >
> > > { Linear Regression components }
> > > y:=pds*Sum(Cum(1)*x,pds)
> > > -Sum(Cum(1),pds)*Sum(x,pds);
> > > z:=pds*Sum(Pwr(Cum(1),2),pds)
> > > -Pwr(Sum(Cum(1),pds),2);
> > >
> > > { Linear Regression Slope }
> > > z:=If(z=0,.000001,z);
> > > LinRegSlp:=y/z;
> > >
> > > {Linear Regression Indicator (LinReg end-point)}
> > > LinRegInd:=LinRegSlp*Cum(1)
> > > -LinRegSlp*Mov(Cum(1),pds,S)+Mov(x,pds,S);
> > >
> > > { Plot }
> > > If(plot=1,LinRegInd,LinRegSlp)
> > >
> > > ---8<---------------------------
> > >
> > >
> > > jose '-)
> > >
> > >
> > > --- In equismetastock@xxxxxxxxxxxxxxx, "Roy Larsen" <rlarsen@xxxx>
> > > wrote:
> > > > Hi All
> > > >
> > > > Can anyone help with the mathematical relationship between
> Linear
> > > Regression and Linear Regression
> > > > Slope. I have found that the formula created by Tom (Sprunger)
> for
> > > Linear Regression, using a pseudo
> > > > loop, to be more true to the standard Linear Regression
> indicator in
> > > MetaStock than the common LR
> > > > formula using a number of Cum() functions. I would like to carry
> > > that greater accuracy through to a
> > > > Linear Regression Slope indicator for plotting the weekly value
> on
> > > daily charts.
> > > >
> > > > Any assistance would be appreciated.
> > > >
> > > > Thanks
> > > >
> > > > Roy
> > >
> > >
> > >
> > >
> > >
> > > Yahoo! Groups Links
> > >
> > >
> > >
> > >
> > >
> > >
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
------------------------ Yahoo! Groups Sponsor --------------------~-->
Make a clean sweep of pop-up ads. Yahoo! Companion Toolbar.
Now with Pop-Up Blocker. Get it for free!
http://us.click.yahoo.com/L5YrjA/eSIIAA/yQLSAA/BefplB/TM
--------------------------------------------------------------------~->
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/equismetastock/
<*> To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|