Template-Type: ReDIF-Article 1.0

Author-Name: Hugo Roberto Balacco 
Author-Workplace-Name: Facultad de Ciencias Económicas, Universidad Nacional de Cuyo 
Author-Name: Gustavo Germán Maradona  
Author-Workplace-Name: Facultad de Ciencias Económicas, Universidad Nacional de Cuyo 

Title: Modelización y predicción de series de tiempo financieras utilizando redes neuronales

Abstract: The purpose of this work is to model and predict Financials Time Series by using neural networks. In order to achieve this aim, a recurrent total neural network with two hidden layers has been chosen; one layer for the linear threshold function and the other for the arctangent function. The series used in this research paper are the MERVAL index (Argentina) and the DOW JONES (USA). These results are based on information obtained over a period that goes from 1995 to 2006. The presentation will deal with the comparison of alternative techniques and the results obtained by other research workers.

Classification-JEL: C40

Keywords: Neural Network, Forecast, Architecture Types, Transfer Functions, Mean Absolute Error.

Journal: Económica

Pages: 3-23
Volume: LVII
Year: 2011
Month: January-December
File-URL: https://revistas.unlp.edu.ar/Economica/article/view/5361/4393
File-Format: Application/pdf
Handle: RePEc:akh:journl:575