Research on stock price prediction based on LSTM-Transformer fusion model in financial management

Abstract

Stock price prediction is a challenging task due to its non-linearity, volatility and temporal dependence that traditional models are hard to fully capture. This study proposes an LSTM-Transformer fusion model, combining LSTM’s strength in long-term temporal feature extraction and Transformer’s advantage in global correlation capture. Comparative experiments on multi-source financial data show the model outper-forms single deep learning and traditional models, providing reliable support for investment decision-making.