Evaluation of volatility forecasts in an economic value framework

Abstract

The GARCH, moving average, previous period volatility and AVV models are compared relative to market implied volatility, first in a simulated market, and then using actual options data. The first set of analysis employs the simulation framework developed in a sequence of papers and published by Engle, Hong, Kane, and Noh (1993). This is an incremental profit methodology designed to rank volatility forecasts within a daily trading framework in an artificial options market. This type of analysis is important for exchanges and regulators when considering daily margin setting. The second set of analysis employs actual option trades in evaluating the ranking of volatility forecasts. This methodology accounts for the observed feature that options on index futures are typically held until maturity in the market considered. This type of analysis is important for institutions and large traders for position setting. The rankings based on daily trades are quite different from those based on positions held until expiration. Australian data on the Share Price Index (SPI) futures and options on SPI futures, traded on the Sydney Futures Exchange (SFE), from 1989 to 1995 are employed in this analysis.