Monthly research notes on developments in artificial intelligence and
quantitative trading, compiled from primary sources — papers, official
release notes, and lab publications — with every material claim cited.
Each note carries a Research O / Research F mark identifying
the model that performed the research.
September opened with an unusually coherent verdict. In a single arXiv announcement cycle, three methodologically distinct quantitative-finance results pointed the same way on whether frontier-model capability translates into trading…
August 2026 was the month machine output stopped being taken at face value. On August 2 the European Union's obligation to mark AI-generated content took effect, and Anthropic published the mechanics of Claude's text watermark two weeks…
July's frontier came back online and multiplied: Claude Fable 5 was redeployed globally on July 1 after June's export controls lifted, and OpenAI shipped GPT-Live and GPT-5.6 — previewed as "Sol" only thirteen days earlier — in…
The first half of June delivered a frontier supply-side shock and a capital-markets signal in the same fortnight: Anthropic launched Claude Fable 5 and Claude Mythos 5 at less than half preview-tier pricing, while both Anthropic and OpenAI…
May moved the "LLM agent as quant researcher" thesis from speculation to working systems: MadEvolve demonstrated evolutionary optimization of full trading pipelines with explicit p-hacking controls, and two independent benchmarks arrived…
April was the month the industry began stress-testing its own claims: Anthropic disclosed an unreleased frontier model that finds and exploits software vulnerabilities better than all but the most skilled humans, days before UC Berkeley…
March's capital cycle reached new scale — OpenAI closed a $122 billion round at an $852 billion valuation with explicit pre-IPO framing — while Anthropic's twin disclosure incidents showed operational security becoming the binding…
February saw the foundation-model paradigm cross into market microstructure: J.P. Morgan-affiliated researchers released TradeFM, a generative Transformer trained on billions of trade events across more than 9,000 equities, while new arXiv…
January was a month of consolidation: capital and infrastructure dominated AI headlines — reported mega-investments in OpenAI and Microsoft's in-house silicon — while agentic AI moved deeper into enterprise workflows. In quantitative…