Using AI and big data to predict stock market

• Main Author: Li, Jing
• Other Authors: Mohammed Yakoob Siyal
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/168380
• Institution: Nanyang Technological University

Stock market prediction has gradually become an intriguing topic to research due to its significant potential of making huge profits. However, the fact that stock market is a chaotic and volatile system makes it challenging for people to establish a reliable method to predict the stock price accurately. In previous studies, solutions such as buy-and-hold strategy, random selecting strategy, and other traditional statistical analysis methods were provided, but their performance was not good enough to be financially applied. Recently, stock prediction is experiencing a disruptive revolution due to the tremendous growth of “Big data” and advancements in Artificial Intelligence (AI), which has a significant impact. Nowadays, it is easier to access stock and financial information of public companies. Meanwhile, Machine Learning techniques, especially Deep Learning models are being applied for making stock price prediction for both classification task, which aims at predicting the price trend (upward or downward), and regression task trying to predict the exact price value. In this paper, three of the most active stocks including Sembcorp Marine Ltd (S51.SI), Marco Polo Marine Ltd. (5LY.SI), and Thai Beverage Public Company Limited (Y92.SI), are selected and analyzed to predict stock market information based on a large number of historical statics from Yahoo Finance. The prediction experiments are conducted by building and training three machine learning models: Decision Tree, Support Vector Machine, and deep learning model Long Short-Term Memory. After comparing experiment results, LSTM shows the best performance in studying stock movement patterns and predicting stock price value, achieving an accuracy of 91%.