Effective utilization of neural networks for constructing an intelligent decision support system for dealing stocks

Abstract

In this paper, we propose a new decision support system for dealing stocks which utilizes the predictions (obtained by NNs) concerning the occurrence of the "Golden Cross (GC) and Dead Cross (DC)", those (also obtained by NNs) concerning the rate of change of the future stock price several weeks ahead, and that (also obtained by NNs) concerning the relative position of the stock price versus "GC" and "DC". Computer simulation results concerning the dealings of the TOPIX for the last 15 years confirm the effectiveness of our approach. © 2009 Springer Berlin Heidelberg.