Deep technical research on LLMs, compliance systems, market microstructure, and the infrastructure questions that matter more than the hype.
Every forecasting model in quant finance assumes prediction is a single-pass problem. A February 2026 arXiv paper challenges that from first principles — and the implications go further than the paper admits.
Why compliance and risk are the most underinvested, highest-leverage AI problem in global markets — and what the stack should actually look like.
Alpha generation, earnings NLP, RAG on regulatory text, and where language models actually move the needle.
AML, surveillance, KYC, explainability — the unglamorous stack that has more AI upside than trading.
Time series forecasting, GNNs for transaction networks, reinforcement learning, and agentic pipelines.
Independent research at the intersection of AI and quantitative finance. Based on public papers and institutional reports — no employer affiliation, no proprietary data, no investment advice.
About →