Portfolio Risk Management: An Empirical Study Based on ARIMA and Random Forest

Yushan Guo

doi:10.62177/apemr.v2i6.940

Authors

Yushan Guo Boston University

DOI:

https://doi.org/10.62177/apemr.v2i6.940

Keywords:

Portfolio Risk Management, ARIMA, Random Forest, Value-at-Risk, Chinese A-share Market, Machine Learning, Time Series Forecasting

Abstract

This study proposes a hybrid framework for portfolio risk management in the Chinese A-share market, combining diagnostics-driven ARIMA identification with Random Forest based feature integration under a Value-at-Risk (VaR) optimization scheme. Unlike conventional parametric VaR models that depend on restrictive distributional assumptions, the framework separates mean dynamics from residual volatility and incorporates nonlinear predictors, including momentum, realized volatility, and higher-order moments. By extending prior ARIMA-machine learning hybrids, which have primarily focused on return forecasting and mean-variance allocation, this study advances the methodology through direct quantile estimation and its integration into a VaR-constrained portfolio decision process. Empirical evidence indicates that the proposed framework generates more accurate VaR forecasts, stabilizes portfolio volatility, and enhances backtesting performance relative to equal-weighted and benchmark strategies.

Downloads

Download data is not yet available.

References

Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77–91.

Allen, D. E., McAleer, M., & Singh, A. K. (2017). Risk measurement and risk modeling using VaR and expectiles. Journal of Econometrics, 198(2), 363–372.

Mandelbrot, B. (1963). The variation of certain speculative prices. Journal of Business, 36(4), 394–419.

Engle, R. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of U.K. inflation. Econometrica, 50(4), 987–1007.

Du, S., Tang, G., & Li, S. (2019). Risk measurement of Chinese stock market based on GARCH model and Extreme Value Theory. Emerging Markets Finance & Trade, 55(10), 2159–2175.

Li, J., & Zhang, Y. (2024). Value-at-Risk forecasting for the Chinese new energy stock market via quantile regression neural network. Procedia Computer Science, 232, 1145–1152.

Chen, K., & Xu, L. (2025). Hybrid models for financial forecasting: Combining ARIMA-type and machine learning methods. arXiv preprint arXiv:2505.19617.

Wang, Y., Xiang, Y., Lei, X., & Zhou, Y. (2022). Volatility analysis based on GARCH-type models: Evidence from the Chinese stock market. Economic Research – Ekonomska Istraživanja, 35(1), 2530–2554.

Song, S., & Li, H. (2022). Predicting VaR for China’s stock market: A score-driven model based on normal inverse Gaussian distribution. Finance Research Letters, 45, 102165.

Zhang, Z. (2023). Research on risk measurement in Chinese stock market — Based on GARCH-VaR modeling. Journal of Risk and Financial Management, 16(3), 145.

Zhao, L. (2023). Comparing ARIMA and Random Forest for stock return forecasting in A-shares. Computational Economics, 61(4), 923–940.

Cai, X. (2025). ARIMA and Random Forest hybrid models for stock price prediction in Chinese A-shares. Working Paper.

Deng, L. (2021). ARIMA-based forecasts and Monte Carlo simulation in portfolio optimization. Quantitative Finance, 21(2), 145–162.

Zheng, Y., Liu, H., & Chen, M. (2022). LSTM-based forecasting and portfolio optimization in CSI300. Economic Modelling, 114, 105873.

Nguyen, T., & Tran, Q. (2023). Application of ARIMA and mean–variance models in financial market forecasting. Heliyon, 9(11), e105664.

Poon, S.-H., & Granger, C. W. J. (2003). Forecasting volatility in financial markets: A review. Journal of Economic Literature, 41(2), 478–539.

Campisi, T., et al. (2024). Random Forest approaches for volatility forecasting in financial markets. Applied Economics Letters, 31(6), 513–520.

Mutinda, J., et al. (2024). Machine learning for financial risk prediction: A Random Forest perspective. Expert Systems with Applications, 233, 120045.

Buse, A., et al. (2025). Quantile Random Forests for Value-at-Risk forecasting. Journal of Forecasting, forthcoming.

Shiraishi, Y., & Watanabe, T. (2024). Quantile forests for time series applications. Statistical Modelling, 24(3), 321–345.

Candila, V., et al. (2025). MIDAS quantile Random Forest for mixed-frequency VaR prediction. International Journal of Forecasting, forthcoming.

Qiu, L. (2024). Deep quantile LSTM for tail-risk prediction in financial markets. Financial Innovation, 10(2), 45.

Aprea, G., et al. (2024). Quantile recurrent neural networks for VaR and ES. Procedia Computer Science, 227, 189–197.

Jiang, X., & Ren, J. (2024). Machine learning signals for crash risk prediction. Journal of Financial Data Science, 6(1), 56–74. Diversification / VaR benchmark

Jorion, P. (2007). Value at Risk: The New Benchmark for Managing Financial Risk (3rd ed.). McGraw-Hill.