Authors
Abstract
Appropriate methods for prediction of future trends in capital markets lead to a better decision making for market participants. Classic methods don not perform well in prediction of financial markets due to the nonlinear and chaotic nature of these markets. Moreover, information extracted from data disappear quickly, so these method are not workable in the long run.
The goal of this paper is using ant colony optimization algorithm for prediction of Tehran Stock Exchange's total return index (TEDPIX) data. First, we used the largest Lyapunov exponent to the consider chaotic nature of TEDPIX and then the ant colony optimization paradigm we employed to analyze topological structure of the attractor behind the given time series and to single out the typical sequences corresponding to the different parts of the attractor. The typical sequences were used to predict the time series values.
Eventually with respect to MSE , RMSE and MAE, ACO has lower error than GARCH and EGARCH models; however, Diebold Marino test shows that there is no difference if we use ACO or GARCH models for prediction; this represents that differences of error for different models in this article are very little. This article with detachment of typical sequences allows a structural method for prediction of chaotic data. So in prediction of data with many fluctuations and in long term, it can result to a better predictions. The algorithm of this paper is able to provide robust prognosis to the periods comparable with the horizon of prediction.
Keywords
Keywords
)