Stock forecasting using transformers

• Main Author: Peng, Xiaoqi
• Other Authors: Wang Lipo
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/169344
• Institution: Nanyang Technological University

This paper develops a model for predicting stock investment value using past stock market information based on the transformer deep learning method. To obtain the investment value of stock more accurately, the technical analysis method of predicting the future stock price based on the past stock price and the fundamental analysis method based on the past intrinsic value of stocks (Composite Index, earnings per share, Market capitalisation in circulation) are adopted. To predict stock investment value as accurately as possible, transformer as a new deep learning method is tried to solve this problem. The main research work and contributions of this paper are as follows: (1) Build a model of stock prediction method in a Python environment based on technical analysis and fundamental analysis of the actual stock market; (2) Based on the PyTorch platform and Python language, a stock prediction method based on the transformer algorithm is developed. By debugging parameters and comparing the predicted value with the actual value, the effectiveness of this method in stock investment value prediction is demonstrated. Based on the above facts, this paper successfully realises the use of the transformer method to predict the future investment value of stocks.