Hello,

I explained it on -at list I think. Custom backtest phase (i.e. 2 nd phase) 
operates not on your symbol(s), but on the portfolio equity ticker.
What you see on the chart displays symbol-derived data. This is different if 
you apply   calculation of portfolio equity ~~~EQUITY.
If you want to pass variables from 1-st pass to  2-nd pass you need to use 
STATIC variables, store in 1st pass and read in 2nd pass (CBT)

Best regards,
Tomasz Janeczko
amibroker.com
  ----- Original Message ----- 
  From: Steve Dugas 
  To: [email protected] 
  Sent: Thursday, February 14, 2008 9:16 PM
  Subject: Re: [amibroker] Custom Backtest doubles opt time?


  OK, thanks a lot TJ.  I guess I have a bigger problem anyway if you or anyone 
else knows the answer...
  I am already calculating a stat in regular AFL that I display on my chart, 
and I am just trying to add this stat to the optimization report so I can sort 
on it.  Using bo.AddCustomMetric, If I define MyStat like this it works fine

  MyStat = LastValue( Cum( 1 ) );

  But if I change it to this, my new column always shows zero

  MyStat = LastValue( Cum( MyBooleanArray ) ); // array definitely contains 
some 1's and some 0's

  Any idea what could be wrong? Thanks very much for any help!

  Steve

  ----- Original Message ----- 
    From: Tomasz Janeczko 
    To: [email protected] 
    Sent: Thursday, February 14, 2008 5:08 AM
    Subject: Re: [amibroker] Custom Backtest doubles opt time?


    Hello,

    If you run on SINGLE ticker and actual system formula is very very basic 
(few lines), then adding CBT will have significant impact. 
    But if you run on any larger list of symbols then it won't be significant 
at all (there is only one execution of CBT (2nd phase of backtest) regardless 
of symbol count, 
    while there are (number of symbols) executions of 1st phase).  

    Bottom line: in all cases except most trivial ones, it does not add more 
than 10%.

    Best regards,
    Tomasz Janeczko
    amibroker.com
      ----- Original Message ----- 
      From: Steve Dugas 
      To: [email protected] 
      Sent: Thursday, February 14, 2008 4:08 AM
      Subject: Re: [amibroker] Custom Backtest doubles opt time?


      Hi Graham - That run was done on current ticker only   8 - )
      There is a lot of charting code, parameters, etc in there, I guess I 
could comment it out or create a shorter version to reduce the opt 
time...thanks!

      Just wondering though, in your experience, does accessing the CBI add 
100% to the opt time? Does the length of the non-CBI code affect whether it 
adds 10% or 100% ?  Thanks for the advice!


      Steve
        ----- Original Message ----- 
        From: Graham 
        To: [email protected] 
        Sent: Wednesday, February 13, 2008 9:35 PM
        Subject: Re: [amibroker] Custom Backtest doubles opt time?


        Suggest you see if you can reduce your basic code to be as efficient as 
possible
        Can you reduce the number of symbols used in the backtest by including 
only those you are likely to trade?


        -- 
        Cheers
        Graham Kav
        AFL Writing Service
        http://www.aflwriting.com 



        On 14/02/2008, Steve Dugas <[EMAIL PROTECTED]> wrote: 
          Hi All - I want to use the custom backtester interface for the first 
time, to add a custom metric. Without accessing the CBI, my AFL is 1,900 lines 
of code and takes 38 minutes to run 42,000 opt steps. Even adding the tiny test 
code below doubles the opt time to 1 hour and 12 minutes. Just wanted to ask 
any of the more experienced CBI users if this sounds normal - I am wondering if 
maybe there is something I should change...  Thanks very much!

          Steve

          SetCustomBacktestProc

          ( "" ); 
          if

          ( Status( "action" ) == actionPortfolio ) 
          {

          bo = 

          GetBacktesterObject(); 
          bo.Backtest();

          TotalModeRevs = 

          5; 
          bo.AddCustomMetric( 

          "Total Mode Reversals", TotalModeRevs ); 
          }






   

Reply via email to