Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Forecasting turning points in stock market prices by applying a neuro – fuzzy model

Zopounidis Konstantinos, Atsalakis Georgios

Full record

---

Summary

Many researchers have tried to model the human behavior in stock markets. Because stock markets are influenced by many economical, political and psychological factors, it is very difficult to forecast the future stock prices. However, it is impossible to develop a precise mathematical model that can predict crisp numerical values of the stock prices. This is due to uncertainties involved in these parameters, the uncertain behavior of the factors, and tolerances. The potential of fuzzy logic in developing a model for characterization by approximate reasoning really lies here. This paper presents a modest attempt to catch rather the turning points of stock prices by using the Fuzzy Inference System (FIS) rather than the absolute prices. The methodology consists of two steps: (1) developing the basic model using special data to train the model and (2) validating the basic model by using the special and raw stock exchange data. The results have indicated that the fuzzy inference system provides a prudent way to capture uncertainty (non-statistical) in relationships among parameters that control the turning points of the stock prices. The novelty of this article is the selected data used for training the model. Each sample of data includes four values that correspond to the stock prices of four continued sessions. These data samples are then divided into the training and checking data. In the evaluation phase, unseen data are presented in the model in order to forecast the turning point of the stock.

Services