Intelligent Data Engineering and Automated Learning - IDEAL 2012

Volume 7435 of the series Lecture Notes in Computer Science pp 352-359

Hybrid Architecture to Predict Trends at Stock Exchange of São Paulo: Markowitz Model and a Multilayer Perceptron

  • Paulo Henrique KaupaAffiliated withIndustrial Engineering Post Graduation Program, Nove de Julho University
  • , Renato José SassiAffiliated withIndustrial Engineering Post Graduation Program, Nove de Julho University
  • , Edinalva Batista RamalhoAffiliated withFundação Getúlio Vargas CEAFE Master in Financial Economics

* Final gross prices may vary according to local VAT.

Get Access


The main challenge in Stock Exchange is choose stocks with uptrend in order to compose a profitable stock portfolio and also ensure security of investment, since the risk in stock investment is considered high. One tool that can help investors to identify the stocks behavior and help to select the right stocks becomes essential. The application of intelligent techniques, especially Artificial Neural Networks to forecast trends in stock prices generated good results. Thus, in this paper was created hybrid architecture, composed by the Markowitz Model and an Artificial Neural Network Multilayer Perceptron, in order to support the investor. For the experiments, the information of the ten most traded stocks on Stock Exchange of São Paulo was extracted. The hybrid architecture is given as follows: Processing Stock Information in Markowitz Model then result is presented as one of input variables of neural network. For analyze the results, was applied an investment simulator, where the investment return obtained point to the use of hybrid architecture in investments.


Hybrid architecture Multilayer Perceptron Markowitz