Resumo

CABRERA LLANOS, Agustín Ignacio  e  ORTIZ ARANGO, Francisco. Forecast of the IPC (Índice de Precios y Cotizaciones) return by means of differential neural networks. Contad. Adm [online]. 2012, vol.57, n.2, pp.63-81. ISSN 0186-1042.

Over the years the use of artificial neural networks as a tool for simulation, modeling and description of nonlinear dynamical systems has been consolidated as an effective and relatively fast technique thanks to the great development experienced in computer systems. This technique commonly used in some areas of Applied Engineering was frst used in financial applications since the mid-nineties. This paper uses one of the most recent and powerful techniques in the feld of neural networks: Differential Neural Networks Analysis (DNNA), a technique frequently used in analysis of biotechnology processes. This technique carries out the analysis and estimation of the evolution of behavior in the IPC (Stock Market Index) of the BMV (Mexican Stock Exchange) during the period from November 8, 1999 to January 27, 2011. The analysis also includes an intra-day forecast (6 values into a trading session) of the IPC return, the forecast extends during one day after the last data time series of the IPC. The predicted results showed a great similarity with actual data.

Palavras-chave : Artificial neural networks; differential neural networks; forecast; Mexican Stock Exchange; nonlinear functions; dynamic systems.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )