Evaluation of Machine Learning Techniques for Stock Market Movement Prediction

Financial market prediction is a classic problem in both academia and industry. This problem has attracted not only researchers in the financial community, but also many researchers in the computer science field. In recent years, many machine learning and deep learning models have been applied to stock market movement prediction with different horizons. This paper evaluates 18 machine learning and deep learning models for 3 stock indexes, predicts the movement of close price after 1, 5 or 10 days, and presents the detailed prediction results. This paper finds that the fully convolutional neural network has no obvious advantages over other machine learning models. And the 10-day movement is easier to predict, which would be useful for designing a long-term investment strategy