Abstract

We show that traditional measures of curvature and symmetry of the “smiles” improve volatility predictions in forex markets. We consider post crisis data at a daily basis for seven currencies vis a vis the American dollar: The British pound, the Euro, the Australian dollar, the Japanese yen, the Brazilian real and the Mexican and Chilean peso. While our results are robust to the option currency and maturity, they are particularly strong for latin-American currencies and options with longer maturity. We find that the simultaneous inclusion of skewness and kurtosis to a forecasting model significantly improves its predictive accuracy.