Is there a standard method for getting a continuous time series from futures data?

I would like to be able to analyse futures prices as one continuous time series, so what kinds of methods exist for combining the prices for the various delivery dates into a single time series?

I am assuming you would just use the front date for most of the time, and then combine this into some kind of weighted combination of the front and second month to simulate the roll.

A quick Google search has shown that there seems to be a number of methods for doing this. Are any considered standard?

In terms of my ultimate goal, I would like a single time series of prices (I can live with just using closing prices rather then OHLC data) so that I can estimate historical volatilities in the prices.

Answer

If your intention is to use the resulting continuous contract time series to perform return based calculation as would be the case with a volatility analysis then what you want to use is the ratio-adjustment method.

If you are happy to roll on a single day this is trivially implemented by taking the ratio of next contract settlement price to leading contract settlement price on the roll day, then multiplying all historical leading contract prices by that ratio and repeating the process backwards along the contract history.

This will result in a time series that exhibits the returns of an excess return commodity index. Note that benchmark commodity indices roll on a multi-day window, e.g. DJ-UBS or GSCI roll in 20% increments over 5 days.