Value investing with machine learning: the South American market

• Main Author: Chen, Ye
• Other Authors: Wang Lipo
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; Machine learning; Value investing; Layers; Neurons; Training speed; Prediction accuracy
• Online Access: https://hdl.handle.net/10356/180455
• Institution: Nanyang Technological University

Machine learning has been a highly popular research topic in recent years. This study aims to apply machine learning to value investing, with the goal of predicting future stock price trends of various companies. It assists investors in making informed decisions to achieve high returns on investments. The primary machine learning method employed in this study is LSTM (Long Short-Term Memory), which performs well with time series data such as financial data of companies. This paper compares the training speed and prediction accuracy of models using different numbers of layers and neurons. The conclusion drawn is that a model with two layers, where the first layer has 200 neurons and the second layer has 100 neurons, exhibits the best performance. Such a model demonstrates satisfactory accuracy in predicting stock price trends for both large and small companies.