Volatility Prediction

This chapter reviews the applications of volatility prediction. Volatility prediction can be applied in variance and volatility trading, in covariance trading, in portfolio selection, in quantile estimation, and in option pricing. In addition, prediction of volatility can be applied by credit institutes to measure risk and to set the risk premium. Volatility estimation can be applied in quantile estimation, because a standard deviation estimate can be used to construct a quantile estimate. The chapter discusses the measurement of performance of volatility predictors. It explains autoregressive conditional heteroscedasticity (ARCH) and generalized ARCH predictors of volatility. GARCH predictors and moving average predictors lead often to good predictions of volatility, for short prediction horizons. Finally, the chapter explains the state space predictors of volatility and illustrates the application of linear and kernel regression in volatility prediction.