Quantitative Research + Financial AI Engineering

I build research systems around cross-sectional factors, realistic backtesting, and evidence-constrained financial agents. My focus is on making assumptions, data boundaries, validation results, and failure modes explicit.

a-share-multifactor-research

• Cross-sectional research using value, 12-1 momentum, and 60-day low-volatility signals.
• Includes point-in-time data handling, realistic execution constraints, walk-forward/OOS evaluation, and auditable outputs.
• Public demos use synthetic data; formal private-data results are kept distinct from the reproducible demo.

fund-evidence-agent

• Evidence-constrained fund research workflow with deterministic retrieval, citation checks, numerical recomputation, and refusal states.
• A typed tool layer and mock single-agent evaluation demonstrate routing, guardrails, audit traces, and offline reproducibility.
• The default public implementation is MOCK_ONLY; it does not claim real-LLM quality or real-fund performance.

• Separate public reproducible demos from private or restricted research data.
• Treat out-of-sample evaluation, transaction costs, and data leakage controls as first-class requirements.
• Prefer traceable evidence and explicit refusal over unsupported financial conclusions.

Quantitative Research + Financial AI Engineering | Cross-sectional factors, realistic backtesting, evidence-constrained financial agents

1. a-share-multifactor-research
2. fund-evidence-agent