Strategy Lab

Build a strategy in plain English or code, then backtest it against real historical order books — real fills, real fees, real settlements — and deploy it to paper.

Real order-book backtests Monte Carlo + FinSight Score Deploy to paper

Every dataset is available to your backtest— read it with ctx["data"]:weatherfundamentalstechnicalscalibrationnewsfull catalog →

1Build a strategy

Describe it in plain English

The AI writes the strategy code for you, ready to backtest. Ask again to refine it.

…or start from a template:

2Backtest it

Strategy code

strategy(ts, ticker, book, ctx) → [{outcome, side, size_usd}]. ctx.data carries the datasets known at that moment (no lookahead) — weather, fundamentals, technicals, calibration, news (catalog). No imports; math available.

Validate the edge

Infer datasets → backtest → lead-lag the thesis → verdict. An honest check — “no edge” is a valid answer.

Create an account on Get started first.