Modelling stock return sensitivities to economic factors with the Kalman filter and neural networks

摘要: Sensitivity analysis of asset returns to various economic variables provides investors with a useful tool build portfolios and manage their risk. However, there are strong reasons believe that stock exposures evolve through time factor models involving them only pertinent if they use reliable estimates future sensitivities. Both Kalman filtering neural networks may be used provide such estimates. While the filter is good at modelling structure sensitivities, capable relating exogeneous in non linear way. Furthermore, because two approaches perform complementary tasks sensitivity forecasting, combined achieve better performances. These procedures evaluated controlled simulation experiment real exposure analysis. Stock sensitivities interest exchange rates forecasted for 90 French shares built accordingly.