A production agent that reads market structure, reasons about risk, and executes trades continuously on Hyperliquid.
Ingests funding, open interest, liquidation maps, and market state snapshots in near real time.
Builds hypotheses from structured context, then selects actions with explicit risk framing.
Combines model context, technical state, and market microstructure into actionable trade plans.
Places, manages, and exits positions with hard constraints on size, stop logic, and exposure.
Our agent design assumes market structure is contextual and adaptive. Pattern interpretation matters more than static formulas.
Read the thesis →
Postmortem of the $1.30/day LLM trading agent pitch. What actually happened: simple Python rules replaced the AI decision layer, the bot went live on Hyperliquid, and the LLM's real job turned out to be research coordination — not trade execution.
A real-time dashboard that watches 14 coins for panic, maps liquidation zones, scores cascade intensity, and wakes an AI agent when things get interesting. FastAPI + matplotlib, zero JavaScript charts.
We built a cascade evaluator, calibrated 710+ events, and found cascade-fade alive on SOL (PF 1.44–2.68 walk-forward) using a simpler proxy detector — with a fill assumption that IS the edge. BTC is dead. Funding rates were never backtested standalone. The "nobody uses this data" claim was cope.
Why treating markets as language problems yields better models than physics. A thesis we believe in, with honest notes on what we tested and what remains theory.