Hi all. I'm looking at a Channel Breakout system at the moment. Most of what I've read takes the optimisation approach to determining the length of the channel; the answer seems to be 40-80, but I want to have a look at something more flexible to prevailing conditions. What I'm thinking about is to use an adaptive measure of channel length. From my past life I can see it as a loop such that: cylen:=fml 'cycle length'; {this will be a relatively fast-moving variable} chanlen:=forum.dll (hhv(cylen,cylen)); {hopefully this one is slower, and step-change in nature} {this is where I have to step outside msfl} loop:=if(chanlen>cylen, cylen=chanlen and redo chanlen calc, chanlen); sig:=c>forum.dll (hhv(h,chanlen)); etc I realise that the syntax isn't here yet, but what I ultimately want to get is a measure of channel length for my breakout. Maybe this is an overkill on what I really need, but I'd appreciate being able to know for sure. Any help greatly appreciated, and thanks in advance.

Hi Jose Prompt and helpful as ever; but perhaps you missed my difficulty? Whether one chooses to adapt your code - something I've done in the past, thank you - or use another measure of cycle length, my prime difficulty remains that I want to determine a 'justifiable'(?) channel length on the basis of immediate past behaviour. My own biases lead me towards a cycle-based determination of that length. With that in mind, I want to make sure that I don't end up with an arbitrarily low channel length which had been derived from the then-current instantaneous cycle length. Can I draw you a word picture? Say we have a price series which starts with a firm (linear?) trend down: this will have a long instanteous cycle length. As the trend slows and begins to reverse, the 'curvature' of the series increases and the cycle length will decrease commensurately. As another trend becomes established (up or down doesn't matter), the cycle length will begin to increase again. Depending on where we get our signal, we may see a breakout over a longer- or shorter- period channel, dependent on the instantaneous value of the cycle length. What I want to do is assume that I'm looking at the lowest cycle length present. Then any HHV measure of that cycle length will return a value greater than the current value. If I were now to look back further using the new value and redo the HHV calculation, I will again get a value equal to or greater than my first iteration. If it's equal, then great: run with that as the channel length. If not, then redo the HHV calculation until we get equality. It's the iteration thing that gets me: no problem in the languages I learnt way back, but not so in MSFL unless one knows how. Perhaps I could just double the length until I get a result, but then I'm left with a channel that may be much too long. That would make for a reeeeaaallly long-term system. :P Essentially, what I'm trying to do is capture the length of a consolidation period, but without penalising myself too much. An HHV lookback over the length of a cycle seemed a good first approximation, but I wanted to refine that calculation by 'running it throught the washer until it was clean'! So there's my 'need' explained as fully as I can. As always, my ideas run ahead of my code skills. #-o Thanks for your help. D