Statistics Seminar: Finlay -- Option Pricing with VG-like Models

In this talk I relax separately two assumptions regarding the Variance
Gamma (VG) process and price options accordingly. In the case of the
Difference of Gammas model a better fit to market data is achieved
than that achieved by other comparable models. In the case of the long
range dependent VG model, the current `skew-correcting’ approach to
pricing options is found to have shortcomings, and a number of model
characteristics are identified (flexible skewness, dependence of
squared returns, accommodation of the leverage effect) which appear to
be important in achieving a good fit to market data.