Building the future profit and loss distribution of a portfolio holding highly nonlinear and path-dependent derivatives, among other assets, is a challenging task. Giacomo Bormetti, Flavio Cocco and Pietro Rossi provide a simple machinery where an increasing number of assets may be accounted for in a simple and semi-automatic fashion. They resort to a variation of the least squares Monte Carlo algorithm in which the continuation value of the portfolio is interpolated with a feed-forward neural network. They account for the profit and loss distribution of a whole portfolio even when the dependence structure between different assets is very strong, eg, for contingent claims written on the same underlying.
Deep learning profit and loss
Giacomo Bormetti;
2021-01-01
Abstract
Building the future profit and loss distribution of a portfolio holding highly nonlinear and path-dependent derivatives, among other assets, is a challenging task. Giacomo Bormetti, Flavio Cocco and Pietro Rossi provide a simple machinery where an increasing number of assets may be accounted for in a simple and semi-automatic fashion. They resort to a variation of the least squares Monte Carlo algorithm in which the continuation value of the portfolio is interpolated with a feed-forward neural network. They account for the profit and loss distribution of a whole portfolio even when the dependence structure between different assets is very strong, eg, for contingent claims written on the same underlying.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
