This is a Linear Reg fit using the equations from "Standard Math Tables"
published by CRC. The Y axis uses the array 'Prc', the X axis uses the array
'X'. You can put any values you want into those two arrays.
To get it to work you need to 'Insert' it into a Chart Window
then click on the last bar.
The answers come out in the Interpretation window.
This was coded for comfort - not for speed.
Watch out for line wraps
//Linear Regression Fit with Standard Error function;
// Ed Hoopes
// April 2008
// Rev A - Started with QFit
// ******************************* Linear Regression Fit
****************************;
Pers = Param("Periods", 20, 5, 200, 1 );
LookBack = Param("Look Back", 0, 0, 500, 1 );
SetBarsRequired( 100 );
//Prc = (Open + High + Low + Close ) / 4 ;
Prc = Close;
//Initialize variables;
SUMX = 0.00; SUMY = 0.00;
SUMXY = 0.00; SUMX2 = 0.00;
// X[0] = 0.00;
for (i = 0; i < BarCount; i++ )
{
X[i] = 0;
}
for (j = BarCount - 1 - Pers; j < BarCount; j++ )
{
X[j] = X[j-1] + 1;
}
//Calculate SUMS;
SumX = Sum(X, Pers );
SumY = Sum(Prc, Pers );
SumX2 = Sum(X*X, Pers );
SumXY = Sum(X*Prc, Pers );
// Calculate the Slope;
SlopeNumerator = Pers * SumXY - SumX * SumY ;
SlopeDenominator = Pers * SumX2 - SumX * SumX ;
Slope = SlopeNumerator / SlopeDenominator ;
// Calculate the Intercept;
InterceptTerm1 = SumY / Pers ;
InterceptTerm2 = Slope * SumX / Pers;
Intercept = InterceptTerm1 - InterceptTerm2;
// Calculate the Standard Error;
SumSE = Sum((Prc - (intercept + Slope * X))^2, Pers );
DenomSE = Pers - 2;
StErr = sqrt(SumSE / DenomSE );
printf("Slope = " + Slope + "\n" );
printf("Intercept = " + Intercept + "\n" );
printf("StdErr = " + StErr + "\n" );
Plot(Slope, "Prc", colorYellow, styleLine + styleThick );
--- In [email protected], "Mike" <sfclimb...@...> wrote:
>
> I haven't looked into either myself, but the two most common answers to
> alternative array manipulations are:
>
> Osaka DLL http://www.amibroker.org/3rdparty/
> R Plugin http://www.amibroker.com/members/
>
> Of course, you could always just write your own function:
>
> http://www.amibroker.com/guide/a_userfunctions.html
>
> Mike
>
> --- In [email protected], "brucet30" <brucet30@> wrote:
> >
> > I am looking for some AFL code that would perform a linear regression on
> > two arrays (x, y) and return the intercept and slope for this data set. THe
> > only funcitons I have been able to find so far in the AFL library and
> > elsewhere assume the x-axis is time and the y-axis is price of some other
> > indicator series. I need something more general, that could work with any
> > (X,y) data set. Any help on this would be very much appreciated.
> > Thanks
> >
>
