Stock Price Prediction Using an Improved Transformer Model: Capturing Temporal Dependencies and Multi-Dimensional Features

Abstract

This paper proposes a stock price prediction model based on an improved Transformer and verifies its effectiveness in predicting stock market fluctuations through experiments. Traditional stock price forecasting methods mostly rely on simple linear regression or traditional machine learning models, but these methods have certain limitations in capturing the complex nonlinearities and temporal dependencies in stock market data. To address this problem, this paper combines the self-attention mechanism of the Transformer model and proposes an improved method to improve the accuracy of stock price prediction by better modeling the time dependencies and multi-dimensional features in stock market data. Experimental results show that the improved Transformer model is significantly better than traditional benchmark models such as XGBoost, CNN, and LSTM in terms of RMSE, MSE, and MAE. Our model effectively enhances the ability to predict short-term fluctuations and long-term trends by introducing techniques such as feature fusion and adaptive time windows. Future research can further explore the fusion of multi-modal data and combine cutting-edge technologies such as graph neural networks and reinforcement learning to promote the intelligence and accuracy of stock market prediction models and provide more effective solutions for decision support systems in the financial industry.