Title: Forecasting the daily spot volatility paths of equity indices via functional autoregressive models: An empirical study
Authors: Isao Ishida - Konan University (Japan) [presenting]
Abstract: The performance of functional autoregressive models in forecasting daily spot volatility paths is empirically investigated using high frequency intraday data of S\& P 500 and Nikkei 225 equity indices. The functional data analysis involves smoothing in obtaining functional representations from discretely observed data. In this step, we apply some of the methods, such as the Fourier transform method, developed specifically for estimating daily spot volatility paths to the observations of high frequency equity index returns, and treat the obtained daily functions as observed realizations of the daily spot volatility paths (unlike in the case of the functional ARCH/GARCH models). We then use the functional autoregressive models for modeling and forecasting the daily spot volatility paths, and find some improvement in forecast accuracy over several alternative procedures for spot volatility forecasting.