Elamur Research⌕
☰
home / papers / 2026 / QFinZero: A Unified Financial Toolchain for LLM-Based Trading Agents

QFinZero: A Unified Financial Toolchain for LLM-Based Trading Agents

Paper · 2026 · Haochen Luo, Yifan LI, Ho Tin Ko et al.

Summary

LLM trading agents are usually tested on pre-assembled market snapshots, where humans have already chosen prices and news. Real systems instead pull data from many providers with mismatched schemas, which makes integration costly and results hard to reproduce. QFinZero is an open-source environment that gives agents one consistent, JSON-based, time-aligned toolkit. It has three services:

  • Unified Price Query (UPQ) for minute-level and daily stock and option data, including option chains.
  • Event Stream Pipeline (ESP) for news plus earnings and economic calendars.
  • Paper Money Broker (PMB), a stateful simulated brokerage that tracks orders through their lifecycle and supports limit and stop-limit orders.

Logging and deterministic replay let the same runs be repeated exactly. An as-of timestamp cursor prevents agents from seeing future data.

The authors built an instruction-driven benchmark that scores tool use, not profit. Frontier models choose the correct tool almost every time. Their errors come mainly from filling in parameters and aligning time ranges. The authors present the system as a basis for live benchmarks and for agentic reinforcement learning.

Key points

  • Data sources and storage: By default, the NASDAQ API supplies calendars, MASSIVE supplies prices, and BENZINGA supplies news and earnings guidance. News is stored in MongoDB, calendars in SQLite, and prices in partitioned Parquet files.
  • Coverage: Crypto, stocks and options. MASSIVE price data arrives about 15 minutes late, while news and calendars update daily.
  • Metrics: Three measures: whether the agent selects the right tool (Tool Match Accuracy, TM), how correctly it fills in arguments (Parameter Accuracy, PA), and whether its time ranges line up (Time Alignment, TA).
  • Results: Gemini 2.5 Flash and Pro reach 91.00 overall on single-step tool calling, and DeepSeek Chat tops multi-step planning at 95.00. Llama 3.1 8B Instruct scores lowest on tool calling at 62.33.
  • Demo and code: Published as an ACL 2026 system demonstration, with a dashboard and playground. Code is at github.com/CityU-MLO/qfinzero.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of Science and Technology of ChinaHefei · China108 works · 112,698 citations🎓City University of Hong KongHong Kong · China45 works · 3,014 citations🎓Chinese University of Hong Kong, ShenzhenShenzhen · China47 works · 2,596 citations🎓Shanghai University of Finance and EconomicsShanghai · China13 works · 825 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Decision Sciences · 209 worksManagement Science and Operations Research · 108 worksBusiness, Management and Accounting · 32 worksEconomics, Econometrics and Finance · 17 worksFinance · 5 worksAccounting · 5 works

Cite

Cite as Haochen Luo, Yifan LI, Ho Tin Ko, An Binh Minh, Junjie Xu, Tang Pok Hin, Wang Chak Wong, Gao Yuan, Zhengzhao Lai, Yuan Zhang and Chen Liu. 2026. QFinZero: A Unified Financial Toolchain for LLM-Based Trading Agents. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-demo.7

@inproceedings{luo2026qfinzero,
  title = {{QFinZero: A Unified Financial Toolchain for LLM-Based Trading Agents}},
  author = {Haochen Luo, Yifan LI, Ho Tin Ko, An Binh Minh, Junjie Xu, Tang Pok Hin, Wang Chak Wong, Gao Yuan, Zhengzhao Lai, Yuan Zhang and Chen Liu},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-demo.7},
  url = {https://doi.org/10.18653/v1/2026.acl-demo.7}
}

References

32 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. The same primitives that enable reproducible evaluation also make QFinZero a suitable infrastructure layer for two emerging research directions: live benchmarking of financial agents, and agentic reinforcement learning (agentic RL) of tool-using LLM policies. Static financial benchmarks increasingly… find
  2. Simone Balloccu, Patrícia Schmidtová, Mateusz Lango, and Ondřej Dušek. 2024. Leak, cheat, repeat: Data contamination and evaluation malpractices in closed-source llms. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Pape…
  3. Tianle Cai, Xuezhi Wang, Tengyu Ma, Xinyun Chen, and Denny Zhou. 2024. Large language models as tool makers. In The Twelfth International Conference on Learning Representations. find
  4. Zhiyu Chen, Wenhu Chen, Charese Smiley, Sameena Shah, Iana Borova, Dylan Langdon, Reema Moussa, Matt Beane, Ting-Hao Huang, Bryan R Routledge, and 1 others. 2021. Finqa: A dataset of numerical reasoning over financial data. In Proceedings of the 2021 Conference on Empirical Methods in Natural Langua…
  5. Jasper Dekoninck, Mark Müller, and Martin Vechev. 2024. Constat: Performance-based contamination detection in large language models. Advances in Neural Information Processing Systems, 37:92420– 92464. find
  6. Abul Ehtesham, Aditi Singh, Gaurav Kumar Gupta, and Saket Kumar. 2025. A survey of agent interoperability protocols: Model context protocol (mcp), agent communication protocol (acp), agent-to-agent protocol (a2a), and agent network protocol (anp). arXiv preprint arXiv:2505.02279. — [arXiv:2505.02279](https://arxiv.org/abs/2505.02279) find
  7. Conclusion and Future Work We present QFinZero, a unified and agent-oriented financial toolchain that integrates price delivery, event streaming, and stateful brokerage simulation into a coherent, callable framework. By standardizing multi-frequency market access, structured event retrieval, and rea… find
  8. Tianyu Fan, Yuhao Yang, Yangqin Jiang, Yifei Zhang, Yuxuan Chen, and Chao Huang. 2025. Ai-trader: Benchmarking autonomous agents in real-time financial markets. Preprint, arXiv:2512.10971. — [arXiv:2512.10971](https://arxiv.org/abs/2512.10971) find
  9. Jiazhan Feng, Shijue Huang, Xingwei Qu, Ge Zhang, Yujia Qin, Baoquan Zhong, Chengquan Jiang, Jinxin Chi, and Wanjun Zhong. 2025. Retool: Reinforcement learning for strategic tool use in llms. arXiv preprint arXiv:2504.11536. find
  10. Xinyi Hou, Yanjie Zhao, Shenao Wang, and Haoyu Wang. 2025. Model context protocol (mcp): Landscape, security threats, and future research directions. ACM Transactions on Software Engineering and Methodology. find
  11. Dongfu Jiang, Yi Lu, Zhuofeng Li, Zhiheng Lyu, Ping Nie, Haozhe Wang, Alex Su, Hui Chen, Kai Zou, Chao Du, and 1 others. 2025. Verltool: Towards holistic agentic reinforcement learning with tool use. arXiv preprint arXiv:2509.01055. — [arXiv:2509.01055](https://arxiv.org/abs/2509.01055) find
  12. Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. 2025. Search-r1: Training llms to reason and leverage search engines with reinforcement learning. arXiv preprint arXiv:2503.09516.
  13. Yaxuan Kong, Hoyoung Lee, Yoontae Hwang, Alejandro Lopez-Lira, Bradford Levy, Dhagash Mehta, Qingsong Wen, Chanyeol Choi, Yongjae Lee, and Stefan Zohren. 2026. Evaluating llms in finance requires explicit bias consideration. arXiv preprint arXiv:2602.14233. — [arXiv:2602.14233](https://arxiv.org/abs/2602.14233) find
  14. Timo Schick, Jane Dwivedi-Yu, Roberto Dessi, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. Toolformer: Language models can teach themselves to use tools. In Thirty-seventh Conference on Neural Information Processing Systems.
  15. Changlun Li, Yao SHI, Chen Wang, Qiqi Duan, Runke RUAN, Weijie Huang, Haonan Long, Lijun Huang, Nan Tang, and Yuyu Luo. 2025a. Time travel is cheating: Going live with deepfund for real-time fund investment benchmarking. In The Thirty-ninth Annual Conference on Neural Information Processing Systems… find
  16. Yu Shi, Yitong Duan, and Jian Li. 2025. Navigating the alpha jungle: an llm-powered mcts framework for formulaic factor mining. arXiv preprint arXiv:2505.11122. — [arXiv:2505.11122](https://arxiv.org/abs/2505.11122) find
  17. Zhengliang Shi, Shen Gao, Xiuyi Chen, Yue Feng, Lingyong Yan, Haibo Shi, Dawei Yin, Pengjie Ren, Suzan Verberne, and Zhaochun Ren. 2024. Learning to use tools via cooperative and interactive agents. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 10642–10657, Miami, F… find
  18. Haohang Li, Yupeng Cao, Yangyang Yu, Shashidhar Reddy Javaji, Zhiyang Deng, Yueru He, Yuechen Jiang, Zining Zhu, Kp Subbalakshmi, Jimin Huang, and 1 others. 2025b. Investorbench: A benchmark for financial decision-making tasks with llm-based agent. In Proceedings of the 63rd Annual Meeting of the As…
  19. Khanh-Tung Tran, Dung Dao, Minh-Duong Nguyen, Quoc-Viet Pham, Barry O’Sullivan, and Hoang D Nguyen. 2025. Multi-agent collaboration mechanisms: A survey of llms. arXiv preprint arXiv:2501.06322. find
  20. Yang Li, Yangyang Yu, Haohang Li, Zhi Chen, and Khaldoun Khashanah. 2023. Tradinggpt: Multi-agent system with layered memory and distinct characters for enhanced financial trading performance. arXiv preprint arXiv:2309.03736. — [arXiv:2309.03736](https://arxiv.org/abs/2309.03736) find
  21. Yijia Xiao, Edward Sun, Di Luo, and Wei Wang. 2024. Tradingagents: Multi-agents llm financial trading framework. arXiv preprint arXiv:2412.20138. — [arXiv:2412.20138](https://arxiv.org/abs/2412.20138) find
  22. Haochen Luo, Ho Tin Ko, Jiandong Chen, David Sun, Yuan Zhang, and Chen Liu. 2026. Alphabench: Benchmarking large language models in formulaic alpha factor mining. In International Conference on Learning Representations. find
  23. Qianqian Xie, Weiguang Han, Xiao Zhang, Yanzhao Lai, Min Peng, Alejandro Lopez-Lira, and Jimin Huang. 2023. Pixiu: A comprehensive benchmark, instruction dataset and large language model for finance. Advances in Neural Information Processing Systems, 36:33469–33484. find
  24. Shishir G Patil, Tianjun Zhang, Xin Wang, and Joseph E. Gonzalez. 2024. Gorilla: Large language model connected with massive APIs. In The Thirty-eighth Annual Conference on Neural Information Processing Systems.
  25. Fei Xiong, Xiang Zhang, Aosong Feng, Siqi Sun, and Chenyu You. 2025. Quantagent: Price-driven multi-agent llms for high-frequency trading. arXiv preprint arXiv:2509.09995. — [arXiv:2509.09995](https://arxiv.org/abs/2509.09995) find
  26. Cheng Qian, Emre Can Acikgoz, Qi He, Hongru Wang, Xiusi Chen, Dilek Hakkani-Tur, Gokhan Tur, and Heng Ji. 2026. Toolrl: Reward is all tool learning needs. Advances in Neural Information Processing Systems, 38:105523–105553. find
  27. Yuzhe YANG, Yifei Zhang, Minghao Wu, Kaidi Zhang, Yunmiao Zhang, Honghai Yu, Yan Hu, and Benyou Wang. 2025. Twinmarket: A scalable behavioral and social simulation for financial markets. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. find
  28. Mathieu Ravaut, Bosheng Ding, Fangkai Jiao, Hailin Chen, Xingxuan Li, Ruochen Zhao, Chengwei Qin, Caiming Xiong, and Shafiq Joty. 2024. A comprehensive survey of contamination detection methods in large language models. arXiv preprint arXiv:2404.00699. find
  29. Partha Pratim Ray. 2025. A survey on model context protocol: Architecture, state-of-the-art, challenges and future directions. Authorea Preprints. find
  30. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. In The eleventh international conference on learning representations.
  31. Varshini Reddy, Rik Koncel-Kedziorski, Viet Dac Lai, Michael Krumdick, Charles Lovering, and Chris Tanner. 2024. Docfinqa: A long-context financial reasoning dataset. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 445–458. find
  32. Yangyang Yu, Haohang Li, Zhi Chen, Yuechen Jiang, Yang Li, Jordan W Suchow, Denghui Zhang, and Khaldoun Khashanah. 2025. Finmem: A performance-enhanced llm trading agent with layered memory and character design. IEEE Transactions on Big Data. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/luo-2026-qfinzero-unified-financial-toolchain-llm/