This paper investigates the forecasting ability of several volatility specifications that aim to quantify market risk. Using an options’ trading strategy on volatility the comparison is implemented in a dynamic approach, applying the standardized prediction error criterion. The empirical findings of the paper suggest that the SPEC criterion outperforms all volatility models that assume normality on the data and exhibits similar forecasting ability with most of the models that assume skewed distributions of asset returns.
[ Download ]
