PREDICCIÓN NO-LINEAL DE TIPOS DE CAMBIO: ALGORITMOS GENÉTICOS, REDES NEURONALES Y FUSIÓN DE DATOS

It is widely proved the existence of non-linear deterministic structures in the exchange rates dynamic. In this work we intend to exploit these non-linear structures using forecasting methods such as Genetic Algorithm and Neural Networks in the specific case of the Yen/$ and British Pound/$ exchange rates. We also employ a novel perspective, called Data Fusion, based on the combination of the obtained results by the non-linear methods to verify if it exists a synergic effect which permits a predictive improvement. The analysis is performed considering both the point prediction and the devaluation or appreciation anticipation.