Commodity prices have historically been the most volatile of all
global asset prices. Therefore, price forecasts are reliable only as far
as volatility is predictable. However, volatility in commodity exchanges
is far from constant and must be predicted before price forecasts can be
made. A model that incorporates both time series forecasts and
investors' expectations is proposed. This model results in
long-term volatility forecasts that are more accurate than those
obtained from other methods.