Intelligent Portfolio Rebalancing with Reinforcement Learning and Dynamic Resource Allocation Under Market Uncertainty

Authors

  • Karol Dziadowicz Department of Computer Science, Colorado State University, Fort Collins, CO, USA.
  • Margot Benoit School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR, USA.
  • Natasza Cich Department of Computer Science and Engineering, University of Nevada, Reno, Reno, NV, USA.
  • Jeremi Lupina Department of Computer Science, University of Houston, Houston, TX, USA.

Keywords:

constrained reinforcement learning; portfolio rebalancing; transaction costs; conditional value-at-risk; dynamic risk budgeting; nonstationary markets

Abstract

Portfolio rebalancing is a sequential decision problem in which the benefit of moving toward an estimated target must be weighed against trading costs, the risk of breaching a loss limit, and the possibility that the estimate is itself wrong. Calendar rules, threshold rules, and most learning-based portfolio agents treat these considerations in sequence: allocation is optimized first, filtered for turnover second, and examined for risk third. This paper develops a constrained reinforcement learning framework in which friction, risk, and estimation uncertainty enter the decision problem directly. Three components are coupled. The trade vector rather than the target weight is the action, so the self-financing condition and the cost function are structural rather than penalized. An analytically derived no-trade band, obtained from a first-order condition that equates marginal risk reduction to marginal trading cost, is embedded as a feasible-action constraint that the policy cannot violate. A dynamic resource allocation layer maps a scalar uncertainty index and a regime posterior to a risk budget, a trading propensity, and a re-optimization frequency, so that capital and computational attention contract when forecasts are least reliable. The loss limit is imposed as a conditional value-at-risk constraint and handled by Lagrangian dual ascent with ensemble-based tail estimation. The paper derives comparative statics for the band, specifies the conditions under which a learned policy can improve on the analytic rule, and sets out the failure modes and diagnostic protocol that such a system must satisfy before it is allowed to trade. The contribution is a coherent decision architecture whose components are individually interpretable and jointly constrained, rather than a black-box allocation model.

References

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

2. Perold, A. F. (1988). The implementation shortfall: Paper versus reality. The Journal of Portfolio Management, 14(3), 4–9.

3. Almgren, R., & Chriss, N. (2001). Optimal execution of portfolio transactions. Journal of Risk, 3(2), 5–39.

4. DeMiguel, V., Garlappi, L., & Uppal, R. (2009). Optimal versus naive diversification: How inefficient is the 1/N portfolio strategy? The Review of Financial Studies, 22(5), 1915–1953.

5. Moody, J., & Saffell, M. (2001). Learning to trade via direct reinforcement. IEEE Transactions on Neural Networks, 12(4), 875–889.

6. Luo, D., Gattas, J. M., & Tan, P. S. S. (2020, July). Real-time defect recognition and optimized decision making for structural timber jointing. In The International Conference on Computational Design and Robotic Fabrication (pp. 36-45). Singapore: Springer Singapore.

7. Jiang, Z., Xu, D., & Liang, J. (2017). A deep reinforcement learning framework for the financial portfolio management problem. arXiv preprint arXiv:1706.10059.

8. Deng, Y., Bao, F., Kong, Y., Ren, Z., & Dai, Q. (2017). Deep direct reinforcement learning for financial signal representation and trading. IEEE Transactions on Neural Networks and Learning Systems, 28(3), 653–664.

9. Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction (2nd ed.). MIT Press.

10. Kaelbling, L. P., Littman, M. L., & Cassandra, A. R. (1998). Planning and acting in partially observable stochastic domains. Artificial Intelligence, 101(1–2), 99–134.

11. Hamilton, J. D. (1989). A new approach to the economic analysis of nonstationary time series and the business cycle. Econometrica, 57(2), 357–384.

12. Ang, A., & Bekaert, G. (2002). International asset allocation with regime shifts. The Review of Financial Studies, 15(4), 1137–1187.

13. Ledoit, O., & Wolf, M. (2004). A well-conditioned estimator for large-dimensional covariance matrices. Journal of Multivariate Analysis, 88(2), 365–411.

14. Lakshminarayanan, B., Pritzel, A., & Blundell, C. (2017). Simple and scalable predictive uncertainty estimation using deep ensembles. In Advances in Neural Information Processing Systems 30 (pp. 6402–6413).

15. Artzner, P., Delbaen, F., Eber, J.-M., & Heath, D. (1999). Coherent measures of risk. Mathematical Finance, 9(3), 203–228.

16. Rockafellar, R. T., & Uryasev, S. (2000). Optimization of conditional value-at-risk. Journal of Risk, 2(3), 21–41.

17. Altman, E. (1999). Constrained Markov decision processes. Chapman & Hall/CRC.

18. Achiam, J., Held, D., Tamar, A., & Abbeel, P. (2017). Constrained policy optimization. In Proceedings of the 34th International Conference on Machine Learning (pp. 22–31). PMLR.

19. Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., Petersen, S., Beattie, C., Sadik, A., Antonoglou, I., King, H., Kumaran, D., Wierstra, D., Legg, S., & Hassabis, D. (2015). Human-level control through deep reinforcement learning. Nature, 518(7540), 529–533.

20. Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017). Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347.

21. Haarnoja, T., Zhou, A., Abbeel, P., & Levine, S. (2018). Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. In Proceedings of the 35th International Conference on Machine Learning (pp. 1861–1870). PMLR.

22. Gârleanu, N., & Pedersen, L. H. (2013). Dynamic trading with predictable returns and transaction costs. The Journal of Finance, 68(6), 2309–2340.

23. Davis, M. H. A., & Norman, A. R. (1990). Portfolio selection with transaction costs. Mathematics of Operations Research, 15(4), 676–713.

24. Henderson, P., Islam, R., Bachman, P., Pineau, J., Precup, D., & Meger, D. (2018). Deep reinforcement learning that matters. In Proceedings of the AAAI Conference on Artificial Intelligence, 32(1), 3207–3214.

Downloads

Published

2026-10-26

How to Cite

Karol Dziadowicz, Margot Benoit, Natasza Cich, & Jeremi Lupina. (2026). Intelligent Portfolio Rebalancing with Reinforcement Learning and Dynamic Resource Allocation Under Market Uncertainty. Global Financial Analytics Research Review, 3(4). Retrieved from https://www.gfarr.org/index.php/home/article/view/170