Research on Artificial Intelligence Sector Stock Price Prediction based on Event Driven and CNN-LSTM-Attention

Ziye Jin

doi:10.6981/FEM.202508_6(8).0016

Authors

Ziye Jin

DOI:

https://doi.org/10.6981/FEM.202508_6(8).0016

Keywords:

Artificial Intelligence; Event-driven; Financial Time Series; CNN-LSTM; Attention Mechanism.

Abstract

This study proposes a stock price prediction model that integrates an event-driven mechanism with CNN-LSTM-Attention for the highly volatile and event-driven nature of the artificial intelligence (AI) sector. By adopting a progressive architecture that includes BERT for semantic feature extraction, multi-scale CNN for capturing local features, bidirectional LSTM for modeling temporal dependencies, and an Attention Mechanism for weighting key events, the model conducts multi-dimensional analysis of AI sector stock prices. Taking the CSI Artificial Intelligence Theme Index (930713) as the research object, the model integrates time series data from January 2021 to December 2024 and professional stock commentaries from Eastmoney.com. After polarity quantification annotation using a financial sentiment dictionary and alignment with trading day-level timestamps, the model's effectiveness in trend tracking is verified. The results show that the model has an average absolute percentage error (MAPE) of 3.08%, effectively capturing policy-driven market movements, but it has limitations in predicting minute-level fluctuations caused by sudden technological "black swan" events and the resonance of retail investor sentiment. This study provides an interpretable framework for quantitative investment in the AI sector and proposes optimization paths, such as incorporating real-time technical indicators and improving the event embedding module.

Downloads

Download data is not yet available.

References

[1] Wang, H., & Ma, L. (2009). Stock price time series analysis and prediction based on fractal interpolation model. Statistics and Decision, (16), 29–30.

[2] Wu, Y., & Wen, X. (2016). Short-term stock price prediction based on ARIMA model. Statistics and Decision, (23), 83–86.

[3] Li, K., & Tan, M. (2014). Stock prediction based on wavelet support vector machine regression. Statistics and Decision, (06), 32–36.

[4] Ci, B., & Zhang, P. Y. (2022). Financial time series prediction based on ARIMA-LSTM model. Statistics and Decision, 38(11), 145–149.

[5] Wang, Z., Chen, Y., & Qiu, F. (2023). A Study on Financial Time Series Prediction Modeling Based on CEEMDAN-GAN. Journal of Huainan Normal University, 25(06), 48–55.

[6] Li, W. (2024). The Relationship between the Persistence of Financial Time Series Fluctuations and the Optimization of Investment Portfolio -- A Perspective Based on the Moments of GARCH (1,1) Model. Modern Business, (04), 105–108.

[7] Ding, X., Zhang, Y., Liu, T., et al. (2015). Deep learning for event-driven stock prediction. In Twenty-fourth international joint conference on artificial intelligence.

[8] Dong, Y., Li, S., & Gong, X. (2017). Time series analysis: An application of arima model in stock price forecasting. In 2017 International Conference on Innovations in Economic Management and Social Science (IEMSS 2017) pp. 703–710.

[9] Han, T., Peng, Q., Zhu, Z., et al. (2020). A pattern representation of stock time series based on DTW. Physica A: Statistical Mechanics and its Applications, 550, 124161.

[10] Kesavan, M., Karthiraman, J., Ebenezer, R. T., et al. (2020). Stock market prediction with historical time series data and sentimental analysis of social media data. In 2020 4th international conference on intelligent computing and control systems (ICICCS) pp. 477–482.

[11] Liu, S., Zhang, C., & Ma, J. (2017). CNN-LSTM neural network model for quantitative strategy analysis in stock markets. In Neural Information Processing: 24th International Conference, ICONIP 2017, Guangzhou, pp. 198–206.

[12] Luo, S. S., Weng, Y., Wang, W. W., et al. (2017). L1-regularized logistic regression for event-driven stock market prediction. In 2017 12th International Conference on Computer Science and Education (ICCSE) pp. 536–541.

[13] Sharma, V., Khemnar, R., Kumari, R., et al. (2019). Time series with sentiment analysis for stock price prediction. In 2019 2nd International Conference on Intelligent Communication and Computational Techniques (ICCT) pp. 178–181.

[14] Tan, J., Wang, J., Rinprasertmeechai, D., et al. (2019). A tensor-based elstm model to predict stock price using financial news.

[15] Xie, Y., Wulamu, A., Wang, Y., et al. (2014). Implementation of time series data clustering based on SVD for stock data analysis on hadoop platform. In 2014 9th IEEE Conference on Industrial Electronics and Applications. pp. 2007–2010.

[16] Yao, J., Feng, X., Wang, Z., et al. (2021). Tone, Emotion and Market Impact: Based on the Financial Sentiment Dictionary. Journal of Management Science, 24(05), 26–46.