USING HYBRID SYSTEMS TO MODEL AND PREDICT THE BEHAVIOR OF FINANCIAL TIME SERIES
USING HYBRID SYSTEMS TO MODEL AND PREDICT THE BEHAVIOR OF FINANCIAL TIME SERIES

Summary/Abstract: In the last time, more and more investors are attracted by easy earnings, so one of solutions is to invest in stock, even if the risks that they have to assume are proportional with expected returns. But the risks are not the last problem. The real problem for investors is to find the right stock price trend and, for that, in a simplicity manner to say, investors have to build forecasting models, to do some assumptions for variables implied, which are not easy to understand. To refine the prediction process, researchers developed hybrid models for forecasting. A well-known class of those techniques is from soft computing area. The aim of this paper is to test and present results of a hybrid model applied on stock transactions listed on Bucharest Stock Exchange. Also, another objective is to reveal some possibilities, form soft computing side, to use other known hybrid systems, like hybridization of neural networks with fuzzy inference systems, to model and predict the stock price trend.

• Details
• Contents