Probe - Accounting, Auditing and Taxation

Research on stock price prediction based on LSTM model integrated with SHAP interpretation: A case study of CSI 300 index

Chenglong Cao

School of Statistics and Applied Mathematics, Anhui University of Finance and Economics

DOI: https://doi.org/10.59429/paat.v8i2.14477

Keywords: stock price prediction; LSTM model; SHAP interpretation; CSI 300 index

---

Abstract

Aiming at the nonlinear noise feature and poor interpretability of financial time series prediction, this paper constructs an interpretable LSTM prediction framework combined with SHAP, taking CSI 300 daily data from 2015 to 2025 as samples. The dataset is divided at 70%:10%:20%, and comparative tests verify that the 5-day time window has optimal prediction accuracy. The model can fit medium-long term index trends but shows obvious lag under drastic market fluctuations. SHAP analysis indicates that opening, high, low and closing prices are core predictive features, while trading and volatility indicators play auxiliary roles. This research balances prediction precision and model transparency, providing quantitative decision support for index investment and risk control.

---

References

[1] Agrawal J G,Chourasia V,Mittra A.State-of-the-artinstockpredictiontechniques[J]. International Journal of Advanced Researchin Electrical,Electronics and Instrumentation Engineering,2013,2(4):1360-1366.

[2] Box G,Jenkins G M.Analysis:Forecasting and control[J].Sanfrancisco,1976,10.

[3] Bollerslev T. Generalized autoregressive conditional heteroskedasticity[J].Journal of econometrics, 1986,31(3):307-327.

[4] Cortes C,Vapnik V.Support-vector networks[J].Machine learning,1995,20(3):273-297.

[5] Rumelhart D E,Hinton G E,Williams R J.Learning internal representations by error propagation [R].1985.

[6] Bengio Y,Le Cun Y. Scaling learning algorithms towards AI[J].Large-scale kerne lmachines, 2007,34(5):1-41.

[7] Singh R,Srivastava S. Stock prediction using deep learning[J].Multimedia Tools and Applications,2017,76(18):18569-18584.

[8] Yadav A K,Vishwakarma V P. An integrated blockchain based real time stock price prediction model by CNN,BiLSTM and AM[J].Procedia Computer Science,2024,235:2630-2640.[9] Lin Y,Yan Y,Xu J,et al. Forecasting stock index price using the CEEMDAN-LSTM model[J].The North American Journal of Economics and Finance,2021,57:101421.

[10] Elman J L. Finding structure in time[J].Cognitive science,1990,14(2):179-211.

[11] Hochreiter S,Schmidhuber J. Longshort-termmemory[J].Neural computation,1997,9(8):1735-1780.

[12] Lundberg S M,Lee S I.A unified approach to interpreting model predictions[J].Advances in neural information processing systems,2017,30.