Deep Learning Vs. Machine Learning in Predicting the Future Trend of Stock Market Prices

The ability to predict the stock trend is one of the most challenging goals for today's traders. The successful prediction of a stock's future trend could yield significant profit. Various machine learning and deep learning have been introduced in the last decades. However, the trade-off between performance and computational complexity was not addressed. This paper aims to find a well-suited model to predict the stock market price trend, with increment in profit gain in Long and Short trading with comparable prediction performance and computational time. A state-of-art machine and deep learning methods have been investigated along with efficient feature engineering. Experimental results show that Feed Forward Neural Network (FFNN) has the best profitability performance (return) and a reasonable running time, among other tested models.