When i put a volatility cone together for an equity index, my output isn't as smooth or regular as the examples i've seen everywhere else [including the original Burghardt & Lane paper]. Here is my simple methodology:
-Get the daily closing prices.
-Calculate the series of log returns.
-For the n-day HV, i take the STDEV(last n days)*SQRT(256)
I do that for 30, 60, 120, 180 day horizons. Then for each of those series i get the 90th/75th/median/25th/10th percentiles and chart them.
I have some ok looking results for some of the major FX pairs, but not when i do a stock index. Obviously i expect it to change with the length of the sample data, but even when i vary that i get ugly results. The most noticable difference is that the 90th and 75th percentile lines are wavy and don't conform to the cone shape irrespective of whether i take data from the last 30yrs or 2004-6.
Am i going wrong somewhere?
-Get the daily closing prices.
-Calculate the series of log returns.
-For the n-day HV, i take the STDEV(last n days)*SQRT(256)
I do that for 30, 60, 120, 180 day horizons. Then for each of those series i get the 90th/75th/median/25th/10th percentiles and chart them.
I have some ok looking results for some of the major FX pairs, but not when i do a stock index. Obviously i expect it to change with the length of the sample data, but even when i vary that i get ugly results. The most noticable difference is that the 90th and 75th percentile lines are wavy and don't conform to the cone shape irrespective of whether i take data from the last 30yrs or 2004-6.
Am i going wrong somewhere?