Stochastic Volatility Double Jump-Diffusions Model: The Importance of Distribution Type of Jump Amplitude

Abstract

This research examines if there exists an appealing distribution for jump amplitude in the sense that with this distribution, the stochastic volatility double jump-diffusions (SVJJ) model would potentially have a superior option market fit while keeping a sound balance between reality and tractability. We provide a general methodology for pricing vanilla options via Fourier cosine series expansion method, in the setting of Heston's SVJJ (HSVJJ) model that may allow a range of jump amplitude distributions. Example applications include the normal (N) distribution, the exponential (E) distribution and the asymmetric double exponential (Db-E) distribution, regarding to analytical tractability for options and economical interpretation. An illustrative example examines the implications of HSVJJ model in capturing option 'smirks'. This example highlights the impacts on implied volatility surface of various jump amplitude distributions, through both extensive model calibrations and carefully designed implied-volatility impacting experiments. Numerical results show that, with the Db-E jump distribution, the HSVJJ model not only captures the implied volatility smile and smirk, but also the 'sadness'.