Creative LLM workflows - OpenClaw
Plus 13Fs analyzer, alternative data processor and more prediction arenas
Flat Circle tracks creative use cases of LLMs in hedge funds. If you haven’t already, join hundreds of PMs, analysts and engineers reading each week:
OpenClaw
OpenClaw (fka ClawdBot, Moltbot) allows a computer to run an LLM system in an “always-on” way and interact with almost anything - making it feel more like a proactive analyst that you can train and work with via slack, email, signal, etc.
How to build an OpenClaw investment research analyst (Saulius)
OpenClaw changes the equation. It is an open-source, self-hosted AI agent platform that runs persistently on your machine, connects to every messaging platform you use, and has access to a full suite of tools -- file operations, web search, browser control, code execution, and long-term memory. It does not just answer questions. It reads research reports, builds financial models, monitors markets around the clock, learns from its own experience, and proactively alerts you when something needs your attention.
Institutional Investor warns funds not to build their own OpenClaw (Institutional Investor)
Data scientist at real estate asset manager analyzes 120 data center projects by talking to OpenClaw via WhatsApp (Infrastructure Research)
Two funds already hiring engineers to build with OpenClaw (job post, job post)
Upwork post by small event driven fund requesting an OpenClaw screening system (job post)
We are actively exploring how to safely use OpenClaw for investor workflows, please reply to this email to discuss.
Creative LLM workflows
Founder of AI native hedge fund details how he builds with Claude Code (@thomasrice_au)
If it's front end or something we interact with, I start by describing what I need to do and initial ideas for interface. I'll then generate 30 mockups (10 each from GPT, Claude, Kimi), asking them to make each quite different, and to make each one an isolated html file with embedded JS and CSS. I'll then go through the 30, dismiss most, keep a few, then keep iterating until it feels right for what I want it to do.
LLM workflow to analyze new 13F holdings and generate a list of relevant positions and reason each fund likely owns them (@FundamentalEdge)
Bonus: New feature from Polymarket allows anyone to pledge rewards to encourage more research (@polymarket thanks to @adrien_nav). Currently the markets with the largest sponsored rewards are for whether the US strikes Iran, Fed decisions and the S&P.
Interesting jobs
Point72 hiring GenAI engineer focused on alternative data (job post)
Updates from trading arenas
In Prediction Arena, all models turned negative this week (predictionarena.ai)
My take: it appears the models made a few concentrated bets that were impacted by extreme weather and a surprising unemployment report. Like a lot of the other trading models, they often win for a period but are actually selling vol. Read more on AI trading arenas here.
Research paper outlines findings from nine-month LLM driven trading strategy. Findings include a 2.43 sharpe and skill at identifying longs but not shorts (arXiv.org)
Chamath Palihapitiya publishes paid post covering trading arenas (Substack)
Follow for more investor LLM workflows
If you would like to discuss incorporating LLMs or OpenClaw into your research process, reply to this email or reach out via X or LinkedIn.