EmidLabs is backtesting and live-execution infrastructure you plug into instead of building yourself. Define strategies as a JSON DSL small enough for an AI to generate safely, run them server-side in ~260ms for 5 years of data, then subscribe validated ones to live markets. Three MCP servers connect Claude, ChatGPT, or any agent directly — no SDK, no self-hosted bot, no environment to stand up.
FREE BETA ACCESS FOR PRODUCT HUNT 🎉
Sign up at console.emidlabs.com, then email founder@emidlabs.com with code PRODUCTHUNT and the address you signed up with — I'll activate it myself (still a one-person show here, so it's manual for now).
Hey Product Hunt! 👋
I'm a solo founder, and I built EmidLabs because every existing option for backtesting and running a trading strategy either meant standing up your own environment (self-hosting a bot, writing everything in raw Python) or committing to a full platform with its own IDE and workflow. I wanted something you could just plug into — like a lab that's already built, not one you have to assemble first.
The core bet: strategies as a small, structured JSON DSL — restricted enough that an LLM can generate one safely, but expressive enough for real indicators, conditions, and risk rules. That same DSL runs through both the Backtesting API (server-side, ~260ms for 5 years of data) and the Live Execution API, so a strategy that backtests well can go live with zero rewriting.
The part I'm most excited about is the MCP layer: three scoped servers (backtest, live, strategy) that let you connect Claude, ChatGPT, or any MCP-compatible agent straight to the infra — describe a strategy in plain language, backtest it, and put it live, no code required. That opened the door to traders who aren't developers, which wasn't the original plan but turned out to be the most interesting audience.
Would love feedback, especially from anyone who's fought with self-hosted bots or overfit backtests before — that pain is exactly what this is built to remove.