Sorry,
Forgot that the rich text editor fails when using Google Chrome.
What I tried to say was that you can push Howard's code inside the
custom backtester to get your value. When working with a single symbol
it should be the same thing (though I note that it is not for some
unknown reason). When working with a portfolio it will be for the entire
portfolio.
Buy = Month() != Ref( Month(), -1 );
Sell = BarsSince( Buy ) >= 10;
Buy = ExRem( Buy, Sell );
Sell = ExRem( Sell, Buy );
e = Equity(1);
Refline = Cum( 1 );
rsquared = Correlation( e, Refline, 21 ) ^ 2;
Plot( C, "C", colorBlack, styleCandle );
shape = Buy * shapeUpArrow + Sell * shapeDownArrow;
shapecolor = IIf( Buy, colorGreen, colorRed );
PlotShapes( shape, shapecolor );
Plot( e, "e", colorGreen, styleLine | styleLeftAxisScale );
Plot( Refline, "refline", colorBlue, styleLine | styleOwnScale );
Plot( rsquared, "rSqr", colorRed, styleLine | styleOwnScale );
// Added these lines
Plot( Foreign( "~RSQ", "X" ), "RSQ", colorDarkRed, styleDashed |
styleOwnScale );
SetCustomBacktestProc( "" );
if ( Status( "action" ) == actionPortfolio )
{
bo = GetBacktesterObject();
bo.Backtest();
e = Foreign( "~~~Equity", "C" );
refline = Cum( 1 );
rsquared = Correlation( e, refline, 21 ) ^ 2;
bo.AddCustomMetric( "R-Squared", LastValue( rsquared ) );
AddToComposite( rsquared, "~RSQ", "X", atcFlagDefaults |
atcFlagEnableInPortfolio | atcFlagEnableInBacktest);
}
Mike
--- In [email protected], "sfclimbers" <sfclimb...@...> wrote:
>
>
> --- In [email protected], "ramoncummins" ramoncummins@
> wrote:
> >
> > Hi Howard, thanks for your reply.
> >
> > However, the reason I want to use the custom backtester is so that
the
> r-squared metric appears as a column in the system statistics when I
> analysis a portfolio of individual stocks (say the S&P 500 for
example)
> - this would allow me to find the best equity curves quickly.
> >
> > Any further input would be greatly appreciated!
> >
> > Ramon
> >
> >
> > --- In [email protected], Howard B howardbandy@ wrote:
> > >
> > > Hi Ramon --
> > >
> > > You do not need custom backtester to get the statistic you want.
> Try this
> > > code:
> > >
> > > ///////////////////////
> > >
> > > // EquityRsqr.afl
> > > //
> > > Buy = Month() != Ref( Month(), -1 );
> > > Sell = BarsSince( Buy ) >= 10;
> > >
> > > Buy = ExRem( Buy, Sell );
> > > Sell = ExRem( Sell, Buy );
> > >
> > > e = Equity();
> > >
> > > Refline = Cum( 1 );
> > >
> > > rsquared = Correlation( e, Refline, 21 ) ^ 2;
> > >
> > > Plot( C, "C", colorBlack, styleCandle );
> > > shape = Buy * shapeUpArrow + Sell * shapeDownArrow;
> > > shapecolor = IIf( Buy, colorGreen, colorRed );
> > > PlotShapes( shape, shapecolor );
> > >
> > > Plot( e, "e", colorGreen, styleLine | styleLeftAxisScale );
> > > Plot( Refline, "refline", colorBlue, styleLine | styleOwnScale );
> > >
> > > Plot( rsquared, "rSqr", colorRed, styleLine | styleOwnScale );
> > >
> > > /////////////////////
> > >
> > > Thanks,
> > > Howard
> > >
> > > On Mon, Aug 31, 2009 at 3:12 PM, ramoncummins ramoncumm...@wrote:
> > >
> > > >
> > > >
> > > > Hi there
> > > >
> > > > I am trying to calculate the r-squared value of my equity curve
> using the
> > > > custom backtester (this is a measure of goodness of 'fit' -
> basically so I
> > > > can test for a nice smooth equity curve).
> > > >
> > > > The generic r-squared formula is :
> > > >
> > > > r-squared = Correlation(array1, array2, periods) ^ 2
> > > >
> > > > However I don't know what arrays to pass this equation. One of
the
> arrays
> > > > will need to be the equity curve and the other just a counter
type
> array -
> > > > like barindex(). Any help very much appreciated, my experience
> using the
> > > > custom backtester is very limited!
> > > >
> > > > cheers
> > > >
> > > > Ramon
> > > >
> > > >
> > > >
> > > >
> > >
> >
>
