ABSTRACT
In considering daily commodity data for use in practical trading systems, mapping that accounts for rollover at contract expiry is required to modify the data. This is because the individual contract data that constitute a conventional time series do not account for contract expiry and the roll that is inherent in it. Both mapped and unmapped data series for certain key commodities are investigated using econometric PARCH models. The significance of such an approach is that the creation of time series that account for roll will allow more accurate back testing of any algorithmic trading system.
