I am not sure if you were wanting to know how to actually code something like this or if you just want the functionality. I can't help you on the coding part, but I believe there is something available that sounds like what you want. Check this link. I use it as a stop for both long and shorts.
I tried attaching a pic, but it doesn't look like it made it...

John:

An interesting problem. Here is one way I have found of setting the initial stop loss, illustrated by some code taken from another post (# 22224) discussing the RMO system:

{variable a7 is the initial stop loss set at "a couple of ticks" below the low of the bar preceding the entry signal}

a5:=(Fml("Rahul Mohindar Osc (RMO)") > 0) AND

(Fml("SwingTrd 2")>0) AND

(Fml("SwingTrd 2")>Fml("SwingTrd 3"));

a6:=a5=1 AND Ref(a5,-1)=0;

a7:=ValueWhen(1,a6,Ref(LOW,-1)-0.25);

Variable a5 is the "long position" indicator for this variation of the RMO system - a5 is "true", or a long position is recommended, when all three RMO indicators are bullish.

Variable a6 is our signal for entry - it is only "true" on bars where all the conditions for a long position are true, but were not true on the previous bar.

Lastly, variable a7 is the initial stop itself - calculated as the "low less a couple of ticks" for the bar immediately preceding the most recent entry signal.

This method is a bit clumsy, and please note that the "initial" stop loss will be reset every time a new "buy signal" is generated. But the idea of refering back to your most recent entry signal as a fixed point for calculating stops (or profit targets) is one way to deal with the problem you mention.

I got this idea from examining the code for the MS PF Fractal Trading Systems, and how they use a "State=1" variable for showing that the system is in a trade. Thus the initial stop losses and profit targets could be based on the most recent point at which the system switched from having no position ("State=0") to having a position ("State=1").

Shave