Stable Stock Market Prediction Using NARX Algorithm

Computational technologies have offered faster and efficient solutions to many diverse areas including the financial sector. In the financial market, the advancements in computational field have been mainly achieved through the use of neural networks and machine learning tools that delivered a number of financial applications. These applications include: stock market prediction, bankruptcy prediction, risk assessment etc. Thus, in this paper, we are developing a technique to predict the stock market index for the "Dow Jones" using deep learning algorithms. We propose a model based on an adaptive NARX neural network that can predict the closing price of a moderately stable market. In our model, non-linear auto regressive exogenous input model inserts delays into the input as well as the output acting as memory slots thereby raising the accuracy of the prediction. This model uses a time series analysis to improve the prediction accuracy. In addition, Levenberg-Marquardt algorithm has been used for training the network. The accuracy of the model is determined by the mean squared error between the predicted and the actual prices.