STOCK RETURNS FORECASTING FOR FINANCIAL, FOOD, INDUSTRIAL AND SERVICES COMPANIES USING NEURAL NETWORKS AND ARIMAGARCH MODELS

Authors

  • Mauri Aparecido de Oliveira Universidade de São Paulo
  • Alessandra de Ávila Montini Universidade de São Paulo
  • Daniel Reed Bergmann Universidade Presbiteriana Mackenzie

Abstract

The main purpose of this work is realize stock returns forecasting for financial, food, industrial and services companies using feedforward neural networks trained with Levenberg-Marquardt algorithm and Arima-Garch models. In each area two time series was selected from Economatica. To the financial area, Bradesco and Itaú was analyzed, Perdigão and Sadia in the food sector, Marcopolo and Gerdau in the industrial area, finally Pão de Açúcar and Lojas Americanas in the services. The forecasting generated by the two techniques had similar performance implying no significant differences between them.

KEYWORDS: Time series; Forecasting; Levenberg-Marquardt algorithm; Neural networks; Arima-Garch.

Downloads

Download data is not yet available.

Author Biographies

Mauri Aparecido de Oliveira, Universidade de São Paulo

Doutor em Administração de Empresas pela Faculdade de Economia e Administração de Empresas da Universidade de São Paulo (FEA-USP-SP).

Alessandra de Ávila Montini, Universidade de São Paulo

Doutora em Administração de Empresas pela Faculdade de Economia e Administração de Empresas da Universidade de São Paulo (FEA-USP-SP). Professora da Faculdade de Economia, Administração e Contabilidade da Universidade de São Paulo (FEA-USP-SP).

Daniel Reed Bergmann, Universidade Presbiteriana Mackenzie

Mestre em Contabilidade pela Faculdade de Economia e Administração de Empresas da Universidade de São Paulo (FEA-USP-SP). Professor da Universidade Presbiteriana Mackenzie (UPM).

Downloads

Published

2008-08-04

Issue

Vol. 9 No. 1 (2008)

Section

Presentation

License

Once the papers have been approved, the authors will assign their copyrights to this Journal.  The Copyright Assignment Conditions include:

1.  The Mackenzie Administration Journal holds the rights to all the papers published therein through assignment of copyright. 

2.  The author retains moral rights to the paper, including the right to identify the author whenever the article is published.

3. As of July 1, 2015 RAM adopted the CC-BY license standard (Creative Commons– BY). Authors are allowed to copy, distribute, display, transmit and adapt articles. Authors must attribute to RAM explicitly and clearly an article’s original publication (with reference to the journal’s name, edition, year and pages in which the article was originally published), yet without suggesting that RAM endorses the author or its use of the article. Contents are released by means of the CC-BY license to fully inter-operate with a variety of different systems and services, including for commercial purposes. In case of an article’s reuse or distribution, authors must make the article’s licensing terms clear to third parties. CC-BY criteria follow open access policies by major OA (Open Access) publishers and journals, such as PLoS, eLife, Biomed Central and Hindawi, among others.

4.  When formally requested by the author, this Journal may allow the paper to be published as a chapter or part of a book.  The only requirement is that prior publication in this Journal (Journal name, issue, year and pages) must be clearly and explicitly shown as a reference.