Grafting Information in Scenario Trees Application to Option Prices

Many mathematical models used in management science do not impose any complex restrictions on the parameter values arising in a given problem. As a consequence, when the instances arise from a specific real-world application, the models under consideration are usually sufficiently robust to remain meaningful even if their numerical parameters are not estimated with very high accuracy or, alternatively, if small perturbations are applied to the parameter values. Things can be quite different, however, with models developed for financial applications. Indeed, in such applications, many models turn out to be utterly meaningless unless one can ensure that the data sets are realistic and internally consistent with the assumptions underlying the model.