Empirical performance of stochastic volatility option pricing models

摘要

This paper examines the empirical performance of four stochastic volatility option pricing models: Heston, Heston++, Bates and Heston-Hull-Whitc. To compare these models, we use individual stock options data from January 1996 to August 2014. The comparison is made with respect to pricing and hedging performance, implied volatility surface and risk-neutral return distribution characteristics, as well as performance across industries and time. We find that the Heston model outperforms the other models in terms of in-sample pricing, whereas Heston++ model outperforms the other models in terms of out-of-sample hedging. This suggests that taking jumps or stochastic interest rates into account docs not improve the model performance after accounting for stochastic volatility. We also find that the model performance deteriorates during the crises as well as when the implied volatility surface is steep in the maturity or strike dimension.