Orateur
Emmanuel Gobet
(Po)
Description
We develop the reversible shaking transformation methods of Gobet and Liu (2014) to estimate the rare event probability arising in different financial risk settings driven by general Gaussian noise. The underlying Markov chains introduced in our approaches take values directly in the path space. We provide theoretical justification for few key properties of these Markov chains which are required for their ergodicity. Further, using these properties, we prove consistency results for the simulation estimator. The examples in our work cover usual semi-martingale stochastic models (not necessarily Markovian) driven by Brownian motion, and, also fractional Brownian motion based models to address various financial risks. Our approach also handles the important problem of sensi- tivities of rare event probability. We compare our numerical studies to the already existing results and demonstrate improved computational performance.
(Joint work with A. Agarwal, S. De Marco, G. Liu.)
Summary
see abstract
Auteur principal
Emmanuel Gobet
(Po)
