thanks for your help GP. I will look at your suggestion but it should be 
possible to program this exactly using the CBI. Have little time right now and 
will look at it later,

rgds, Ed



  ----- Original Message ----- 
  From: gp_sydney 
  To: [email protected] 
  Sent: Sunday, August 19, 2007 3:57 AM
  Subject: [amibroker] Re: sigScaleOut


  Ed,

  Can't see your attachment, but where the two scale-outs are at the
  same bar, can't you just combine the position sizes for that bar so
  that the one scale-out signal has a larger position size?

  Similarly for the buy and scale-out at the same bar, can't you reduce
  the position size for the buy by the position size for the scale out
  and just take a smaller position for the buy signal?

  Regards,
  GP

  --- In [email protected], "Edward Pottasch" <[EMAIL PROTECTED]>
  wrote:
  >
  > hi,
  > 
  > I wonder if somebody has been using the CBI (Custom backtester
  Interface) to perform a sigScaleOut (or sigScaleIn). Here is the
  problem: In a test setup I enter 3 contracts of ES futures. The code
  uses 3 targets and a trailing stop loss. In some cases it can work
  without using the CBI. However if the first sigScaleOut signal falls
  at the same bar where the trade is entered there is a problem because
  both the entry and the scaleOut are stored in the Buy array. 
  > 
  > Next problem that occurs is that if the first and the second
  scaleOut targets are hit during the same bar. Again we can only store
  one or the other.
  > 
  > Using the low level CBI (I assume using ScaleTrade() ) can overcome
  this problem. Anyone know how to implement this? A simple example will do.
  > 
  > Illustrated is the left entry signal scales out nicely (blue arrows)
  and the last contract is sold at the trailing stop. The second signal
  however both the first and the second scaleOut are at the bar
  following the entry bar (in this setup I do not allow to exit during
  the bar of entry). The backtester will sell only 1 contract at the
  scaleOut point and will sell 2 contracts at target number 3. This is
  wrong.
  > 
  > thx, rgds, Ed
  >



   

Reply via email to