Neuro-evolutionary modeling of the Milan stock exchange for intraday trading

We investigate the correlations among the intraday prices of the major stocks of the Milan Stock Exchange by means of a neuro-evolutionary modeling method. In particular, the method used to approach such problem is to apply a very powerful natural computing analysis tool, namely evolutionary neural networks, based on the joint evolution of the topology and the connection weights together with a novel similarity-based crossover, to the analysis of a nancial intraday time series expressing the stock quote variations of the FTSE MIB components. We show that it is possible to obtain extremely accurate models of the variations of the price of one stock based on the price variations of the other components of the stock list, which may be used for statistical arbitrage. The approach has been also validated by implementing a simple trading system that performs decisions by considering the results computed by the algorithm.