SPEC model selection algorithm for ARCH models: an options pricing evaluation framework

Abstract

A number of single ARCH model-based methods of predicting volatility are compared to Degiannakis and Xekalaki's (2005) poly-model standardized prediction error criterion (SPEC) algorithm method in terms of profits from trading actual options of the S&P500 index returns. The results show that traders using the SPEC for deciding which model's forecasts to use at any given point in time achieve the highest profits.