Forecasting Trends in Stock Prices Using Transformer Networks

Abstract

Financial time series forecasting is undoubtedly the top choice of computational intelligence for finance researchers in both academia and the finance industry due to its broad implementation areas and substantial impact.  The ever-changing tides of the stock market present a significant challenge for investors and financial institutions seeking to make informed decisions. Predicting future stock prices with accuracy remains a highly sought-after capability, offering the potential to optimize investment strategies and navigate market volatility. Traditional forecasting methods have limitations, often struggling to capture the complex dynamics and non-linear relationships inherent in stock price data. This project delves into the potential of deep learning techniques for stock price forecasting. We explore the application of a Transformer-based model, a powerful deep learning architecture known for its ability to learn long-term dependencies in sequential data. By leveraging this model's capabilities, we aim to develop a more robust and accurate approach to forecasting stock prices.