Risk-calibrated LLM portfolio recommendation with reinforcement learning under investor and market constraints

Authors

  • Moritz Hartmann Mathematics, Engineering Sciences, and Natural Sciences Faculty of Mathematics and Computer Science, Heidelberg University, Heidelberg, Baden-Wurttemberg, Germany

DOI:

https://doi.org/10.66421/int.j.datasci.intell.syst69

Keywords:

Portfolio Management, Reinforcement Learning, Large Language Models, Risk Calibration, Market Constraints

Abstract

The integration of artificial intelligence into financial decision-making has fundamentally transformed portfolio management. Traditional quantitative models often struggle to capture qualitative market nuances, while standalone machine learning algorithms frequently fail to adhere to strict financial constraints and risk parameters. This paper introduces a novel framework that synergizes Large Language Models with Deep Reinforcement Learning to optimize portfolio recommendations under rigorous risk calibration. By leveraging the advanced natural language processing capabilities of Large Language Models, the proposed system extracts high-fidelity market sentiment and macroeconomic indicators from unstructured textual data, including financial news, earnings call transcripts, and central bank policy statements. This qualitative state representation is then fused with quantitative price and volume data to form a comprehensive state space for a reinforcement learning agent. To ensure practical viability, the agent is trained under a constrained Markov Decision Process that explicitly incorporates both investor-specific constraints, such as risk aversion and sector exposure limits, and market-specific constraints, including transaction costs and liquidity bounds. Empirical evaluations conducted on a multi-asset universe demonstrate that the risk-calibrated reinforcement learning framework significantly outperforms traditional mean-variance optimization and unconstrained reinforcement learning benchmarks in terms of risk-adjusted returns. The methodology provides a robust, interpretable, and adaptable approach to automated portfolio management, bridging the gap between qualitative market analysis and quantitative execution.

References

1. Liu, R., McKay, D., & Buchanan, G. (2024). Sociotechnical factors affecting interoperability of digital humanities. Proceedings of the Association for Information Science and Technology, 61(1), 218-231.

2. Yao, Z., Li, R., Chen, X., Cheng, J., Le, C., Yue, M., ... & Dong, J. (2026). PreDiff-LM: Pretrained Discrete Masked Diffusion Language Modeling with Hybrid Attention. arXiv preprint arXiv:2607.25157.

3. Yu, Y., & Hu, Z. (2025). Comparative Explanations via Counterfactual Reasoning in Recommendations. arXiv preprint arXiv:2510.10920.

4. Liu, X., Yu, Z., Yao, Y., Yang, K., Zeng, H., & Chen, C. P. (2026). A Comprehensive Survey of Cross-Modal Retrieval: Breaking Through Modal Barriers. ACM Computing Surveys.

5. Xu, Y., Trine, M. Q., & Wu, H. (2026). DARC-Sched: A Data-Aware Multi-Agent Large Language Model Framework for Scalable Software Vulnerability Detection and Repair with Optimized Code Data Scheduling. Fundamental Scientific Reports in Multidisciplinary Areas, 2(02), 269-283.

6. Wang, Y., Fan, S., Wang, M., Gao, S., & Wang, C. (2026, April). DAMR: Efficient and adaptive context-aware knowledge graph question answering with LLM-guided MCTS. In International Conference on Learning Representations (Vol. 2026, pp. 25548-25570).

7. Zhang, X. (2026, June). Workforce Reskilling and Digital Transformation: Empirical Analysis of Human Capital Agility in Global E-Commerce Platforms. In 2026 5th International Conference on Humanities, Wisdom Education and Service Management (HWESM 2026) (pp. 692-703). Atlantis Press.

8. Fama, E. F., & French, K. R. (1992). The cross-section of expected stock returns. Journal of Finance, 47(2), 427–465.

9. Rui, L., & Jingzhu, W. (2020). Research on public willingness to use the national public culture cloud APP. Information and documentation services, 41(4), 39-48.

10. Mou, Y., Yuan, Z., Sun, L., & Huang, Y. (2026). Finance-structured temporal transformers for realized volatility forecasting from implied volatility surfaces.

11. Sun, F., He, S., Wang, R., Ke, L., Shen, H., & Liao, Q. (2026). Modeling Structural Deviation in 10-K Risk Factors: A Semantic Anomaly Detection and Explainable AI Approach. Risks, 14(4), 87.

12. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review, 76(2), 323–329.

13. Li, B., Gu, B., & Ding, Z. (2025). LLM-based Personalized Portfolio Recommender: Integrating Large Language Models and Reinforcement Learning for Intelligent Investment Strategy Optimization. arXiv preprint arXiv:2512.12922.

14. Yang, Z., & Tao, Z. CARE-Fin: A Causal-Aware and Robust Meta-Learning Framework for Safe and Reliable SME Financial Decision-Making. In Second Workshop on Safe AI.

15. Jin, H., Tsai, F. S., & Martinez-Vazquez, J. (2025). Optimal expenditure decentralization for sustainable development: evidence from a 52-country panel analysis. Financial Innovation, 11(1), 111.

16. Zhao, Z., & Welsch, R. E. (2024). Hierarchical reinforced trader (hrt): A bi-level approach for optimizing stock selection and execution. arXiv preprint arXiv:2410.14927.

17. Jin, C., Luo, J., Lyu, S., Huang, C., Zhang, Y., Li, X., & Hu, Q. (2026, July). RoR-CV: A Reasoning-Enhanced Large Language Model Recommendation System for Financial Scenarios with Integrated Computer Vision. In International Conference on Intelligent Computing (pp. 33-44). Singapore: Springer Nature Singapore.

18. Hu, X., Wan, Z., & Yin, Q. (2025). Token Design for Favor Trading with Dynamic Membership. Available at SSRN 6175079.

19. Fan, D.; Liu, M.; Shao, Y.; Yang, L.; Liu, Y.; Zhang, Y.; Ren, Y.; Wang, Z. Domain-Specific Large Language Model for Maintenance Decision-Making on Wind Farms by Labeled-Data-Supervised Fine-Tuning. Engineering, 2025; in press.

20. Hu, Y., Yang, Z., Zhao, C., & Ji, W. (2026, May). Adaptive guidance semantically enhanced via multimodal llm for edge-cloud object detection. In ICASSP 2026-2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 12962-12966). IEEE.

21. Zhu, R., Ma, Z., Wu, J., Gao, J., Wang, J., Lin, D., & He, C. (2025, April). Utilize the flow before stepping into the same river twice: Certainty represented knowledge flow for refusal-aware instruction tuning. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 39, No. 24, pp. 26157-26165).

22. Zhao, J., Zhang, C., Qin, M., & Yang, P. (2025). QuantFactor REINFORCE: mining steady formulaic alpha factors with variance-bounded REINFORCE. IEEE Transactions on Signal Processing, 73, 2448-2463.

23. Cakmak, S., Wang, Y., Gao, S., & Zhou, E. (2024). Contextual ranking and selection with gaussian processes and optimal computing budget allocation. ACM Transactions on Modeling and Computer Simulation, 34(2), 1-24.

24. Sharpe, W. F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance, 19(3), 425–442.

25. Deng, H. When Work Went Remote: Impacts of Telework on Inter-Metro Migration and Urban Housing Markets. Available at SSRN 6588333.

26. Hu, X., Wan, Z., & Zhou, Q. (2025). The profitability puzzle of on-demand carpooling: Pricing, matching, and route compatibility. HKU Jockey Club Enterprise Sustainability Global Research Institute Paper, (2025/001).

27. Meng, X., & Lin, J. (2026). PRISM-Twin: Data-Minimizing and Interpretable Multimodal Campus IoT Student-State Modeling for Risk Support. IEEE Internet of Things Journal.

28. Guo, F., Zhang, Z., Song, W., Tian, H., Zhang, J., & Mo, H. (2026). A TEE-Protected Framework for Explainable LLM-Assisted Microgrid Decision Support.

29. Mo, Z. (2025, May). Development of an Intelligent Retrieval and Recommendation System for Chinese Language Educational Resources Based on DNN Technology. In 2025 2nd International Conference on Intelligent Computing and Robotics (ICICR) (pp. 170-175). IEEE.

30. Hu, X., & Caldentey, R. (2023). Trust and reciprocity in firms’ capacity sharing. Manufacturing & Service Operations Management, 25(4), 1436-1450.

31. Yu, H., Zhang, J., Chen, C., Xiang, T., Fang, Y., Niebles, J. C., & Adeli, E. (2026, March). Socialgen: Modeling multi-human social interaction with language models. In 2026 International Conference on 3D Vision (3DV) (pp. 1-17). IEEE.

Downloads

Published

2026-09-23