DOI: 10.21276/sjet.2016.4.10.7

Pages: 512-519

Publication of bond and stock through stock market, is a way in order to prepare the capital for investment. According to this fact that stock price is the first and important issue for an investor, evaluating and forecasting of the future price will be propounded. Artificial Neural Network (NN) is a way of stock price forecasting. High efficiency of NNs in forecasting and importance of forecasting in different fields, have caused researchers to search the ways of improving the strategies used for increase of the accuracy of forecasting by neural networks. Accuracy level of a NN greatly depends on its weight and bias values. In a NN, weight and bias values depend on the kind of training algorithm. In this article, the NN using for stock price forecasting is trained by a new metaheuristic training algorithm, named bird mating optimizer (BMO). Effectiveness and efficiency of this algorithmis compared with NN trained by Genetic Algorithm (GA) algorithm based on the average, median and standard deviation of the Mean Square Error (MSE).The experimental results show that the BMO lends itself very well to forecast of stock price.