Adaptive Monte Carlo methods to estimate financial risk models
We plan to study a new adaptive Monte Carlo importance sampling algorithm and to use it to estimate a novel model to predict volatility. The model will be cast both in a Bayesian and a classical framework and the predictive power of the two will be compared. A non-parametric version of the Bayesian model will be also proposed to gain higher flexibility.
Swiss National Science Foundation / Project Funding / Mathematics, Natural and Engineering Sciences (Division II)