--- In [email protected], "Howard B" <[EMAIL PROTECTED]> wrote:
> For this statement:
> somevar = CCI(parm);
> "somevar" will be an array with one element for every bar of the
data array,
> the value of that element the result of applying the CCI function to the
> "average" of that bar ((H+L+C)/3), for the lookback length of "parm".
> 
> What problem are you trying to solve?
> 

I want parm to be an array.

What I am doing is trying to 'tune' the CCI to the dominant cycle in
the market.

Let me give you a simplistic example using moving averages. 

If you are trading a trending market (longer period dominant cycle),
you want a longer period moving average to filter out the small (high
frequency) corrections that occur.

In a trading range market (shorter period dominant cycle), you want a
shorter period moving average that can react more quickly to the
shorter term changes in direction.

I started by using the fft to estimate the dominant cycle but I had a
lot of trouble coding something useful so I switched to Ehler's
estimate (using Laguerre filter, it's in the afl library). 

As crude as that estimate is, it improves the performance of the
indicators I tried it on. If I could get a more accurate measure of
the dominant cycle, I am confident it would improve performance even more.

That's why I want the period of the CCI to vary.

I also don't see any point to include the C of a bar for intraday
charts. I use CCIa((H+L)/2,period).

Bill

Reply via email to