Predicting Stock Market Price Movement using Machine Learning Techniques

Abstract

Accurately predicting upcoming values of stock market index is a very difficult task due to instability of financial stock markets. In fact, an accurate prediction helps brokers to make adequate decision on buying or selling stock. Toward this aim, six Machine Learning (ML) techniques namely: Support Vector Regression (SVR), K-nearest Neighbor (Knn), Decision trees (DTs), Random Forest, Artificial Neural Networks (MLPs), Deep learning technique, were built to predict the future closing price for five companies that are part of the S&P500 index and the closing price of S&P500 index. Teen years of data and six new generated variables were used as inputs for our used models, which were assessed using two performance metrics and build using the grid search optimization technique. The results show that there is no best ML technique that may adopted to predict the trends of a given stock price. However, all the constructed techniques yield a very promising performance, and that the MLP and LSTM techniques, which belong to ANN family, may be considered as best techniques.