Articial neural network, which were first opplied in the field of physics, have been used in experiments analysing the price of financial assets. We attempted to mode and predict changes intwo cross-rates (i.e the FRANC/MARK and the MARK/DOLLAR) and three stock-market idices (the KAX, the CAC40 and the S&PP500) using neural-base darchitecture with New-York at the entre of a hierarchy of finanical markets. Gnerally, the network produced good quality forecasts. Indeed, the network perfomed better than traditional statistical models.
展开▼