I have a sample set of values that were taken over a period of time. However, the delta time between each sample is different.
Do I need to account for the different time deltas in the std-dev?
Is std-dev even appropriate for this kind of data?
More info…
The data are temperature samples.
The time range is from 1 hour to multiple days.
Best Answer
Yes, you do need to account for the irregularity of the time series because volatility scales with time. Depending upon the distribution and independence assumptions, sometimes a "square root of time" rule can be appropriate.
Is this data sampled irregularly intraday or across a longer time period? What kind of data is it?
For dealing with high-frequency financial data, you can apply a realized volatility measure, which is available in R in the realized package.
Similar Posts:
- Solved – Realized volatility vs close to close return and open to close return
- Solved – How to measure the true underlying daily volatility from daily data
- Solved – On forecasting, the mean squared error and realized volatility
- Solved – Why is a GARCH model useful
- Solved – Rolling period volatility forecasts