FINANCIAL DATA ANALYSIS

In this framework, our research efforts involved some issues related to the financial markets dynamics, such as for example modelling and forecasting the behaviour of financial indexes in the context of the active management of a portfolio of financial stocks. For this problem, we considered a particular formulation of the Generalized Additive Model (GAM), based on the use of recursive partitioning algorithms for the selection of non constant binwidth for smoothing functions as well as for the ordering of the predictors entering in the model. We illustrated the advantages of the proposed methodology, based on an integration of the GAM procedure with regression trees, in comparison with the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models usually used for predicting stochastic volatility.