A0349
Title: Jump modeling with Dirichlet process mixture
Authors: Yuru Sun - Monash University (Australia) [presenting]
Ole Maneesoonthorn - Monash University (Australia)
Yong Song - University of Melbourne (Australia)
Wei Wei - Monash University (Australia)
Abstract: A Bayesian nonparametric method is proposed using a Dirichlet process mixture to model jumps. This new model encompasses existing parametric jump specification but allows for extreme jumps to occur in either tail of the return distribution via an infinite mixture model. The flexibility in the model specification allows for multi-modal jump size distribution. The robustness of the proposed model is investigated under different jump intensity and size specifications under simulations, and comparisons about the model-driven option implied volatility smiles are provided. The practical use of the model is verified with empirical data in jump detection, data-driven jump size density, and implied volatility smiles.
