Deep-learning-based stock market prediction incorporating ESG sentiment and technical indicators

Area of Science:

• Financial Markets
• Sustainable Finance
• Data Science

Background:

• Sustainability and ESG factors are increasingly critical for modern enterprises.
• ESG indicators influence investor trust, company growth, and stock prices.
• Integrating ESG into financial assessments is essential for evaluating sustainable practices.

Purpose of the Study:

• To propose an innovative approach combining ESG sentiment index and technical indicators for S&P 500 prediction.
• To explore the optimal deep learning model and window sizes for predictive accuracy.
• To clarify the influence and causality of ESG on the S&P 500 index through ablation tests.

Main Methods:

• Extraction of ESG sentiment index from news data.
• Utilization of a deep learning model for stock price prediction.
• Application of mean absolute percentage error (MAPE) for model evaluation.
• Conducting ablation tests to assess ESG's impact and causality.

Main Results:

• Improved predictive accuracy for the S&P 500 index when ESG sentiment is incorporated.
• Demonstrated superior performance compared to models using only technical indicators or historical data.
• Validated the significant influence of ESG sentiment on stock price prediction.

Conclusions:

• Combining technical indicators (short-term) with ESG information (long-term) enhances stock price prediction.
• ESG considerations are necessary for financial assets, offering new investment strategy perspectives.
• Provides valuable insights for investors and financial market experts on ESG integration.