On the high-frequency financial forecasting problem with multilayer perceptron models

In this work we develop a study about the high-frequency financial forecasting problem. It presents some evidences suggesting that the generator phenomenon of this kind of time series cannot be a random walk process. In this context, we present an empirical analysis to investigate the forecasting performance of multilayer perceptron (MLP) model for three problems of high-frequency financial time series from the Brazilian stock market.