O uso de redes neurais artificiais na previsÃo de tendÃncias no mercado de aÃÃes

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

Stock markets are considered a high return investment option, dominated by uncertainty and volatility. The forecast of the movement of that market is a difficult task, because is influenced by many economical, political and even psychological factors. The traditional statistical methods and the known analysis (technical and fundamental) are not capable to identify the non-linear relations between the variables that compose the price of a stock and the trends of the market, became necessary the use of more advanced techniques like Artificial Neural Networks (ANNs). Artificial Neural Networks (ANNs) are a tool that simulate the ability of learning of the human brain. Neural Networks have, among others, the capacity of model non-linear functions in complex environments and with noisy or partial information. Thus, its use in forecast is growing, including stock markets. In this work will be developed models of neural networks to predict futures values of stocks and its trends. Different forms of architecture were considered, always using a multilayer perceptron (MLP). The initial simulation had the objective of forecasting the daily and future value of the preferential stock of the Petrobras, and lately had been extended for the forecasting of the trends of one and two futures days of the former stock and of the index of the Bolsa de Valores de SÃo Paulo (Ibovespa). The studied models revealed a high level of accuracy in trends prediction of the assets analyzed, thus neural networks can be used by investors to assist them in the management of theirs investments portfolio

ASSUNTO(S)

engenharia de producao forecasting, stock market, neural networks previsÃo, mercado de aÃÃes, redes neurais

ACESSO AO ARTIGO

Documentos Relacionados