How do I get coefficient of variation?

I've no programming experience but I've an idea.

So here is the deal, write this indicator and I'll tell you how to use it.

1. Calculate the mean of last 14 (X) candle stick based on closing price
2. Divide standard deviation of 14 candle stick by the mean
3. Multiply it with 100.

That would show the coefficient of variation on closing price.

Can anyone do it? It would be wonderful if 14 candle could be changed to any number of candles as a variable.

Thanks

ok it is written, now, how do you use it.

Wonderful magnumfreak!

I think you already know how to use it, coeff goes up = variability goes up and use a simple 5-9 ma cross to get the direction or use CCI or RSI or whatever other stuff you like. I use pure candlestick PA. By coeff we are converting stdev into percentile value based on previous data. So... by using stdev we can only know how strong the trend is but by using coeff we can know how strong the trend is in percentile :-D

I know many will disagree with it based on the interval and ratio scale thing but you can see from the chart itself that the probabilities of the trend continuation is almost in the 80% region when coeff is showing high variability.

So enter when coeff is going up and not going down. When its going down thats the retracement and unless it picks up again, we are either retracing or ranging.

Coeff can be used as a filter to your existing system to eliminate whipsaw.

Can I have the indicator now? You can post it here as attachment or mail me: im7i4z [at] hotmail

sure, just thought I would have a little fun. I made a little modification to the code. Due to the fact that the price change is relatively small, the standard deviation was ranging really low, 0.154 down to -.1 roughly, so I compensated for this by converting the standard deviation into pips. This did not change the way the curve looked but gave it a larger range to help pick out changes. If you want it back down to the low numbers just let me know.

Interesting concept by the way. I will do some forward testing this week.

thanks for posting. I can't open the ex4 with metaeditor. How do I edit anything if I want to? can you send me the code over e-mail?

I'll update here on this thread if I change any parameters. And yes I'll keep your name as the author but hope you will acknowledge me as well. If not please modify it back to the stdev divided by mean formula. I'm more interested in the value than the curve size :-)

Thanks a lots.