RESEARCH BRIEF
How do we turn market hypotheses into testable systems?
AI-powered quantitative research for portfolio strategy, market analysis, and systematic decision-making.
This HARD Lab direction studies the problem as an end-to-end system rather than as a single model demo. Current related work provides useful building blocks and baselines [1] [2] [3] [4] [5]; the lab’s goal is to connect them into measurable, reproducible research artifacts.
Market-data pipelines and systematic strategy research
Backtesting, evaluation, and portfolio decision support
AI-assisted development and testing of new algorithms
PROPOSED WORKFLOW
Build, measure, iterate.
Research
Turn a market idea into a defined hypothesis, with a data source and an evaluation plan.
Evaluate
Test the strategy, examine its assumptions, and study sensitivity before using it for decisions.
Iterate
Compare outcomes with the original hypothesis and use the evidence to refine the system.
STUDENT ENTRY POINTS
Ways to start contributing.
- Market-data pipelines and systematic strategy research
- Backtesting, evaluation, and portfolio decision support
- AI-assisted development and testing of new algorithms
RELATED WORK / 2026-09-10
Selected papers & citations.
Five primary-source papers selected to frame this direction. Links point to the authors’ arXiv records; these are external works, not HARD Lab publications.
5 selected papers · newest first by initial submission
Agentic Quantitative Trading: A Survey of Workflows, Systems, and Evaluation
Organizes agentic quantitative trading across research, signals, portfolio construction, execution, and evaluation.
Why it matters here. A recent map of the full research workflow: promising prediction is not the same as a reproducible trading system.
Full citation & BibTeX
Fengrui Hua et al. (2026). Agentic Quantitative Trading: A Survey of Workflows, Systems, and Evaluation. arXiv:2608.31041. https://doi.org/10.48550/arXiv.2608.31041
@misc{hardlab_agentic_quant_survey,
title = {{Agentic Quantitative Trading: A Survey of Workflows, Systems, and Evaluation}},
author = {Fengrui Hua and Hengyi Yang and Xinlei Hao and Haohan Zhang and Bokai Cao and Yiyan Qi and Jia Li and Jian Guo},
year = {2026},
eprint = {2608.31041},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.2608.31041},
url = {https://arxiv.org/abs/2608.31041},
note = {External related work; metadata checked 2026-09-10}
}Agentic Trading: When LLM Agents Meet Financial Markets
Reviews LLM-agent trading systems and examines how their evaluations handle financial-market constraints.
Why it matters here. Motivates explicit point-in-time data, chronological testing, cost assumptions, and transparent experiment records.
Full citation & BibTeX
Yihan Xia et al. (2026). Agentic Trading: When LLM Agents Meet Financial Markets. arXiv:2605.19337. https://doi.org/10.48550/arXiv.2605.19337
@misc{hardlab_agentic_trading,
title = {{Agentic Trading: When LLM Agents Meet Financial Markets}},
author = {Yihan Xia and Panpan You and Taotao Wang and Fang Liu and Han Qi and Xiaoxiao Wu and Shengli Zhang},
year = {2026},
eprint = {2605.19337},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.2605.19337},
url = {https://arxiv.org/abs/2605.19337},
note = {External related work; metadata checked 2026-09-10}
}TradingAgents: Multi-Agents LLM Financial Trading Framework
Structures a trading research workflow around analyst, debate, risk, and decision-making agent roles.
Why it matters here. A role-based baseline for orchestration and ablation studies, not a guarantee of performance or a claim about WolfQuant returns.
Full citation & BibTeX
Yijia Xiao et al. (2024). TradingAgents: Multi-Agents LLM Financial Trading Framework. arXiv:2412.20138. https://doi.org/10.48550/arXiv.2412.20138
@misc{hardlab_tradingagents,
title = {{TradingAgents: Multi-Agents LLM Financial Trading Framework}},
author = {Yijia Xiao and Edward Sun and Di Luo and Wei Wang},
year = {2024},
eprint = {2412.20138},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.2412.20138},
url = {https://arxiv.org/abs/2412.20138},
note = {External related work; metadata checked 2026-09-10}
}FinRobot: An Open-Source AI Agent Platform for Financial Applications using Large Language Models
Proposes an open-source platform for coordinating financial applications with LLM-driven agents and tools.
Why it matters here. Useful background for separating reusable tools, data access, analysis roles, and recorded decisions.
Full citation & BibTeX
Hongyang Yang et al. (2024). FinRobot: An Open-Source AI Agent Platform for Financial Applications using Large Language Models. arXiv:2405.14767. https://doi.org/10.48550/arXiv.2405.14767
@misc{hardlab_finrobot,
title = {{FinRobot: An Open-Source AI Agent Platform for Financial Applications using Large Language Models}},
author = {Hongyang Yang and Boyu Zhang and Neng Wang and Cheng Guo and Xiaoli Zhang and Likun Lin and Junlin Wang and Tianyu Zhou and Mao Guan and Runjia Zhang and Christina Dan Wang},
year = {2024},
eprint = {2405.14767},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.2405.14767},
url = {https://arxiv.org/abs/2405.14767},
note = {External related work; metadata checked 2026-09-10}
}Qlib: An AI-oriented Quantitative Investment Platform
Presents an AI-oriented quantitative research platform spanning data processing, model development, and evaluation.
Why it matters here. A reproducible data/model/backtest foundation against which agent-driven strategy research can be compared.
Full citation & BibTeX
Xiao Yang et al. (2020). Qlib: An AI-oriented Quantitative Investment Platform. arXiv:2009.11189. https://doi.org/10.48550/arXiv.2009.11189
@misc{hardlab_qlib,
title = {{Qlib: An AI-oriented Quantitative Investment Platform}},
author = {Xiao Yang and Weiqing Liu and Dong Zhou and Jiang Bian and Tie-Yan Liu},
year = {2020},
eprint = {2009.11189},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.2009.11189},
url = {https://arxiv.org/abs/2009.11189},
note = {External related work; metadata checked 2026-09-10}
}No selected papers match this filter.