Development of a multi-level system for forecasting stock prices of companies in the conditions of the highly volatile russian market

The purpose of the research is to develop and formalize a hybrid model for forecasting stock prices on the Moscow Exchange, which takes into account the specifics of linguistic analysis of Russian-language news and the integration of time series signals with semantic text analysis. The model is aimed at improving prediction accuracy in the conditions of high volatility of the Russian financial market and structural changes after 2022.

Methods. A three-level model for stock price prediction has been developed. Each level implies the use of a more suitable tool considering the forecasting horizon. Thus, the short-term level is based on a GRU recurrent network and uses a sliding window, which shows high efficiency during periods of low volatility. The medium-frequency level uses a regression model with a Huber loss function and modified CIS tone metrics to translate news influence into a monetary equivalent, which allows combining short-term and fundamental indicators; the use of the latter increases the model's adaptability during periods of high volatility. The low-frequency level combines the ARIMAX statistical model, the LSTM neural network, as well as medium-frequency level indicators to assess long-term trends and correct errors of the lower levels. The integration of the levels is implemented through a mechanism of dynamic weight redistribution and inter-level adjustment of confidence coefficients.

Results. A software complex for stock market forecasting was developed based on the Python programming language, implementing a hybrid architecture that integrates time series analysis, fundamental indicators, and news context. The simulation demonstrated sustained forecasting accuracy: under market turbulence conditions, levels L2 and L3 showed a mean absolute error not exceeding 0.5 rubles, which is significantly lower than the metrics of the technical level L1. The dynamic weight redistribution mechanism between levels ensured the model's adaptation to sharp changes in market conditions.

Conclusion. The proposed hybrid forecasting architecture combines machine learning, NLP, and econometric methods, ensuring a balance between sensitivity to short-term changes and the robustness of long-term forecasts. The integration of news analysis, time series, and macroeconomic factors allows for considering the specifics of the Russian market and reduces the impact of market shocks. The system is suitable for use in investment analytics and automated trading solutions, opening prospects for further adaptation to other emerging markets

1. Kingma D.P., Ba J.L. Adam: A Method for Stochastic Optimization. In: Proceedings of the 3rd International Conference for Learning Representations (ICLR). Available at: https://arxiv.org/abs/1412.6980 (accessed 17.12.2025).

2. Box G.E.P., Jenkins G.M., Reinsel G.C., Ljung G.M. Time Series Analysis: Forecasting and Control. 5th ed. Hoboken: John Wiley & Sons; 2015. 712 p.

3. Fama E.F., French K.R. A Five-Factor Asset Pricing Model. Journal of Financial Economics. 2015;116(1):1–22.

4. Hinton G.E., Srivastava N., Krizhevsky A., Sutskever I., Salakhutdinov R. Improving Neural Networks by Preventing Co-adaptation of Feature Detectors. In: Neural and Evolutionary Computing. Available at: https://arxiv.org/abs/1207.0580v1 (accessed 17.12.2025).

5. Belousov A.V. Methods of Time Series Analysis in Economics. Moscow: Finansy i statistika; 2018. 312 p. (In Russ.)

6. Shamraeva V.V. Mathematical Methods for Forecasting Stock Price Changes and Their Implementation Using Machine Learning. Fundamental'nye issledovaniya = Fundamental Research. 2024;(11):88-96. (In Russ.)

7. Vasiliev A., Smirnov I. Application of Machine Learning for Forecasting the Russian Stock Market. Computational Economics and Finance Journal. 2023;5(2):34–45. (In Russ.)

8. Huber P.J. Robust Estimation of a Location Parameter. The Annals of Mathematical Statistics. 1964;35(1):73–101.

9. Teplova T.V., Sokolova T.V., Tomtosov A.F., et al. Sentiment of Private Investors and Its Impact on the Returns of Russian Stocks. Zhurnal Novoi ekonomicheskoi assotsiatsii = Journal of the New Economic Association. 2022;(1):53–84. (In Russ.)

10. Hochreiter S., Schmidhuber J. Long Short-Term Memory. Neural Computation. 1997;9(8):1735–1780.

11. Cho K., van Merriënboer B., Gulcehre C., Bahdanau D., Bougares F., Schwenk H., Bengio Y. Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation. In: Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). Doha; 2014. P. 1724–1734.

12. Pyatibratov D.V., Loginov A.V., Bolotov T.A. Comparative Analysis of Machine Learning Algorithms in Stock Market Research Tasks. Vestnik nauki = Science Bulletin. 2025;3:644-650. (In Russ.)

13. Tetlock P. C. Giving Content to Investor Sentiment: The Role of Media in the Stock Market. The Journal of Finance. 2007;62(3):1139–1168.

14. Glebova A.G., Kovaleva A.A. Forecasting the Volatility of the Russian Stock Market under International Sanctions. Finansy: teoriya i praktika = Finance: Theory and Practice. 2024;(1):20–29. (In Russ.)

15. Damodaran A. Investment Valuation: Tools and Techniques for Determining the Value of Any Asset. 3rd ed. Hoboken: John Wiley & Sons; 2012. 992 p.

16. Paszke A., Gross S., Massa F., Lerer A., Bradbury J. et al. PyTorch: An Imperative Style, High Performance Deep Learning Library. In: Machine Learning. Available at: https://arxiv.org/abs/1207.0580v1 (accessed 17.12.2025).

17. Pedregosa F., Varoquaux G., Gramfort A., Michel V., Thirion B., et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12:2825–2830.

18. Minkin V.A., Kuleshov D.A. Application of Machine Learning Methods for Stock Market Forecasting. Vestnik Finansovogo universiteta = Bulletin of the Financial University. 2022;(3):55–66. (In Russ.)

19. Pershina A.A., Arutyunyan A.G., Kosnikov S.N. Application of neural networks in forecasting economic trends. Regional'naya i otraslevaya ekonomika = Regional and Sectoral Economics. 2023;(5):164-171. (In Russ.) https://doi.org/10.47576/2949-1916_2023_5_164