Volatility forecasting in the framework of the option expiry cycle

The paper presents new UK evidence on the relative predictive performance of several implied and historical volatilities. The Datastream combination of historical and implied volatilities is also tested empirically for the first time. Daily observations are used to increase the power of the tests, and particular attention is paid to forecasting over the life of options. A further contribution of the paper is to examine relative accuracy for several different horizons, and matching the amount of past data to the forecast horizon is found to be effective when forecasting over longer horizons. Historical volatility estimators are found to have greater forecast accuracy than implied volatilities. Although implied volatility is a biased estimator of realized volatility, regression tests show that it contains more information than historical volatility. Also, a simple trading rule using historical volatility estimators is unable to exploit the forecast improvements since it fails to earn abnormal profits after transactions costs.