tman, I have only had a quick look over your indicator, and offer: First, you don't need to take the Log() of the numerator and the divisor. Its redundant in the big scheme of things. Second, to allow users more flexibility in the application of this indicator I would add an Input() variable to be able to easily change the period length. Third, I would include some error handling for Divide by Zero errors whenever you have any division operation taking place in your code. All in all, its good clean code. How do you intend trading this indicator? Do you smooth the output with a MA? Two MA crossover? Anyway, this is more like I would use: --8<------------------------------------------ {tman Volatility Indicator} prd:=Input("Periods",2,100,20); Std(H/L,prd)/Max(Std(C/Ref(C,-1),prd),0.0001); --8<------------------------------------------ Hope this helps. wabbit :D

Hi tman, I've been using volatilyty but to help me with the markets.Your comments raises a question and please forgive me if I'm not thinking right-I'm not feeling that bright right this moment.I always thought that wether you looked at a 5 min,or a daily the high for that day is the same #.Yes it does adjust intraday depending if the price spikes to a new low or a new high but the high for that day is the same.

Sportrider, Sorry for the tardy reply. I am looking for differentials in the close-close volatilty vs the high-low volatility. In other words, a stock may have a many large intraday swings, yet close almost unchanged on the day. Options on a stock like this may not reflect the large intraday volatility. I use this as a screening tool and then narrow the list further by looking at the 30 day implied vol vs the 52 week range of implied vol-----looking to purchase "cheap" straddles. Again, I apologize the the delay in responding. tman

Actually one does need to take the Log if one is going to follow the standard method for calculating volatility. However I think you would also need to calculate the volatility on an annualized basis (252 days) and then look back to the period you are interested in exploring.

This is from CRB

HISTORICAL VOLATILITY (20-day):

STEP 1
For the past 20 days, calculate: today's close / previous close (requires 21 days of data)

STEP 2
Calculate the natural log of the results calculated in STEP 1.

STEP 3
Calculate the sum of the natural logs over the past 20 days.
Calculate the sum of the squares of the natural logs over the past 20 days.

STEP 4
Divide the sum of the natural logs by 20.......................#1
Divide the sum of the squares of the natural logs by 20........#2
Calculate: RESULT 2 - the square of RESULT 1...................#3
Calculate the (square root of RESULT 3) x (sq. root of 252) x 100
This is the 20-day historic volatility for today.

I would highly recommend that the user have a look at both QC volatility fields for a given instrument and also at a site such as iVolatility where you can find graphs of volatility to compare to your results.

By the way, it should be noted what the user is looking at calculating here is historical volatility using the standard method. There is also Parkinson's method which uses ratios. You will get a similar looking volatility shape but very different numeric values.

Bottom line? If you are trying to use the industry standard measure of historical volatility you need to use this sort of formula:

pds:=Input("Volatility Period",0,252,5);
volatility:=Stdev(Log(C/Ref(C,-1)),pds)*Sqrt(252)*100;
volatility;

There are many ways to skin the Volatility cat. This is just the basic approach to historical vol. But this will give you the base line result that is the industry norm. Remember there are four types of Volatility -- historical, future, expected, implied. You will have to do completely different calculations for the other three forms.