Market-oriented AI algorithmic stock prediction and analysis

• Main Author: Deng, Yibo
• Other Authors: Mohammed Yakoob Siyal
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Engineering
• Online Access: https://hdl.handle.net/10356/178222
• Institution: Nanyang Technological University

The aim of this report is to explore and evaluate the application of different machine learning algorithms in stock prediction. Machine learning, as a financial research method widely recognized by scholars nowadays, has a great advantage in the field of stock prediction due to its powerful self-learning and feature extraction capabilities. With the rapid development of information technology and the increasing attention of investors to the stock market, the accuracy and reliability of stock prediction becomes particularly important. In this study, various algorithms such as SVR, LSTM, MLP are selected to analyze and evaluate their performance in stock prediction by comparing the results. In the research process, we firstly selected the stock data of large listed companies in recent years as the research object, processed the data and implemented the models using Python and related libraries, evaluated each model through training and testing datasets, and finally drew conclusions by comparing the prediction accuracy and stability of different algorithms. The research results show that different algorithms will show different advantages and disadvantages in stock prediction. In this study, the MLP model becomes the optimal model by virtue of its high prediction accuracy; LSTM also achieves good prediction results benefiting from its excellent performance in handling time series data and its good long-term memory capability. The other two models are limited by their simple structure, which may be more suitable for quickly building a prediction model or some specific prediction scenarios. In summary, this study provides investors and researchers with a comparison and analysis of different stock prediction algorithms and offers some references and insights for future research and practice in the field of stock prediction.