Stock Price Prediction Using Machine Learning Strategies

Abstract

Being able to foresee the potential opportunities or crisis in stock market has always been desirable among investors. Especially during the Covid-19 global pandemic, the skill of risk management is of great importance to sustain in such an unstable environment. Apart from various kinds of strategies in traditional business analysis, a robust intelligent system that can correctly predict stock price is desired to determine investment strategies. At present, much related research involve with predicting the stock price trend, and most of them use deep learning methods. Although these research managed to achieve an ideal result of their tasks, seldom surveys focus on the summary of deep learning methods employed in stock price prediction. As a result, the aim of this paper is to summarize the machine learning methods used in forecasting stock price, the development context of the task, and, finally, analyze the development trend of the task based on previous published papers. These papers were classified by deep learning methods, which included Long Short-Term Memory (LSTM); Gated Recurrent Units (GRU); Recurrent Neural Network (RNN); and other hybrid deep learning methods. Furthermore, this paper identifies some of the dataset, variable, model, and results of each article. The survey adopted presents the results through the most used performance metrics: Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE) and Mean Square Error (MSE).