Abstract:

This article presents an intelligent system based on artificial neural techniques for predicting the evolution of Romanian stock exchange. It is developed a hybrid neural network with supervised learning algorithm able to learn to predict the values of a stock for a period of time. Learning model proposed for the hybrid network is based on a First Input First Output (FIFO) queue with input values taken from the values obtained by prediction by the neural network at previous time. Experimental study highlights the effectiveness of the proposed learning model for hybrid neural, predictive system properties and its usefulness in a modern economy.