|
PureBytes Links
Trading Reference Links
|
Thomas,
Yes, that's the key. If you use ParamOptimize then it behaves
like PARAM function - i.e. you need to use Parameters dialog
to reset values to defaults (Press "RESET" button), otherwise
ParamOptimize will work like Param - i.e. will return (old values).
Also each indicator window and AA window parameters are
INDEPENDENT (which means that if you change indicator
parameters, AA parameters are NOT affected).
Best regards,
Tomasz Janeczko
amibroker.com
----- Original Message -----
From: "Thomas Ludwig" <Thomas.Ludwig@xxxxxx>
To: <amibroker@xxxxxxxxxxxxxxx>
Sent: Thursday, December 14, 2006 11:45 AM
Subject: Re: [amibroker] ParamOptimize - different backtest results
Thanks, Tomasz,
Below is the code I was using. As mentioned, I backtested this formula with
the default values, replaced "Optimize" with "ParamOptimize" and got a
different backtest result.
However, your remark:
>Note that BACKTEST in second case will use current param values,
> not the defaults.
made me think about this matter once more. I'm not sure if this system was
plotted as an indicator when I performed these backtests. If the meaning of
your comment is that in this case "ParamOptimize" uses the (different) param
values of the plotted indicator and not the default values (=optimal values
from the optimization which I had manually entered) then this is a trap I
wasn't aware of. In order to evaluate this, I repeated the tests today
without the indicator plotted with different param values - and this time the
results were identical. So this is obviously the explanation I was looking
for.
Thanks again!
Greetings, Thomas
function ParamOptimize( TitleName, default, minv, maxv, step )
{return Optimize( TitleName, Param( TitleName, default, minv, maxv, step ),
minv, maxv, step );}
function LinearRegReversal( array, period )
{
Lr = LinearReg( array, period );
return IIf( Lr >= Ref( Lr, -1 ), 1, -1 );
}
p=Optimize("ER-Berechnungszeitraum",17,10,40,1);
z=abs(C-Ref(C,-p));
n=Sum(abs(C-Ref(C,-1)),p);
ER=z/n;
d=Optimize("Glättung",21,15,30,2);
ER=EMA(ER,d);
per=Optimize("LRR-Period",120,60,120,2);
per=round(per/ER);
LRR=LinearRegReversal(C,per);
Plot( LRR, "LR Reversal mit ER", colorBlue);
PlotGrid(1);
PlotGrid(-1);
GraphXSpace=10;
SetTradeDelays(1, 1, 1, 1);
Buy=LRR ==1;
Sell=LRR ==-1;
Cover=Short=0;
Buy1=ExRem(Buy,Sell);
Sell1=ExRem(Sell,Buy);
shape = Buy1 * shapeUpArrow + Sell1 * shapeDownArrow;
//Plot( Close, "Price", colorBlack, styleCandle );
PlotShapes( shape, IIf( Buy1, colorGreen, colorRed ), 0, 0 );
> Send full code you are using. The reason may be invalid formula
> (for example using same names for different params).
> You are not saying whenever you run BACKTEST or OPTIMIZE.
>
> Note that BACKTEST in second case will use current param values,
> not the defaults.
>
> Best regards,
> Tomasz Janeczko
> amibroker.com
> ----- Original Message -----
> From: "Thomas Ludwig" <Thomas.Ludwig@xxxxxx>
> To: <amibroker@xxxxxxxxxxxxxxx>
> Sent: Wednesday, December 13, 2006 9:02 PM
> Subject: [amibroker] ParamOptimize - different backtest results
>
> > In the Amibroker Feedback Center is a request for an "Optimizable
> > Param()" . Graham and Tomasz pointed out that this functionality already
> > exists via the following function:
> >
> > function ParamOptimize( TitleName, default, minv, maxv, step )
> > {return Optimize( TitleName, Param( TitleName, default, minv, maxv, step
> > ), minv, maxv, step );}
> >
> > As a matter of fact I had used this function before but had had problems
> > with it. Now I took the opportunity to try it out again. I tested a
> > trading system at first with using "Optimize" and performed a second
> > backtest where I replaced "Optimize" with "ParamOptimize" without any
> > other changes. The results of both backtests were completely different!
> > This is true regardless if I copy above function into the formula editor
> > or use it via #include <ParamOptimize.afl> (this file is in my
> > Formulas/Include" folder).
> >
> > Could anybody try out if he/she is confronted with the same problem? If
> > yes - what could be wrong here?
> >
> > Thanks in advance.
> >
> > Greetings, Thomas
> >
> >
> > Please note that this group is for discussion between users only.
> >
> > To get support from AmiBroker please send an e-mail directly to
> > SUPPORT {at} amibroker.com
> >
> > For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
> > http://www.amibroker.com/devlog/
> >
> > For other support material please check also:
> > http://www.amibroker.com/support.html
> >
> > Yahoo! Groups Links
>
> Please note that this group is for discussion between users only.
>
> To get support from AmiBroker please send an e-mail directly to
> SUPPORT {at} amibroker.com
>
> For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
> http://www.amibroker.com/devlog/
>
> For other support material please check also:
> http://www.amibroker.com/support.html
>
> Yahoo! Groups Links
>
>
>
Please note that this group is for discussion between users only.
To get support from AmiBroker please send an e-mail directly to
SUPPORT {at} amibroker.com
For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
http://www.amibroker.com/devlog/
For other support material please check also:
http://www.amibroker.com/support.html
Yahoo! Groups Links
Content-Description: "AVG certification"
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.15.18/586 - Release Date: 12/13/2006 6:13 PM
|