EFFICIENT NEURAL STOCK MARKET PREDICTION SYSTEM

Arosha Senanayake, J. Janaththanan

ABSTRACT

Colombo stock market trading varies depending on the quantitative and qualitative factors. The intelligent system introduced in this research predicts the stock market trading in an efficient manner. The analysis of quantitative factors is primarily based on how effective data mining does from the parameters extracted in the Colombo stock market exchange. Initially, it was done based on the extraction of noisy data taking into consideration the principal component analysis and secondly, the feed-forward neural network is applied to optimize the same. The analysis of qualitative factors is based on how frequently the stock market trading is disturbed by the political situation in the country. Fuzzy logic approach is proposed to decide the degree of influence of political situation for the prediction. Finally, another feed-forward neural network is introduced to determine whether the stock market is fully, partially or not based on the quantitative factors. Based on the degree of influence obtained from the fuzzy analysis, the output is projected. The results show that the system can be effectively applied for Colombo Stock Market Exchange with less than 20% error.