Strategy lab

The strategy lab tests a trading strategy against Hacc’s recorded market history and shows how it would have done. It is for research: it never trades, and it cannot promote a strategy to live.

What you get

• Plan: Max. Free and Pro keys do not reach it.
• Access: the Hacc API at api.hacc.fun with your Candle agent key as the bearer token. Candle does not proxy it.
• Replay: run a strategy definition over a window you choose, with nothing saved. 20 replays a day, one at a time, each stopping after 10 minutes.
• Saved strategies: each saved draft is backtested over the last 14 days. A backtest passes with at least 5 trades and at most 20% drawdown.
• Research reports: which signal patterns paid afterwards, recent runners, and first signals after launch.

Routes, request shapes and the report list are in Data and strategy backtesting.

How far to trust a backtest

A backtest fills at recorded prices with an assumed stop slippage and no competition for the entry. It is a model of the past. A pass is a reason to try a strategy with small size, not proof it will pay.

From backtest to live

The path Candle teaches is: backtest, then deploy live through your own agent. The lab does not deploy anything. To go live:

1. Have your agent backtest the strategy in the lab with your Max key, and read the result.
2. Give your agent the strategy as plain instructions: entry and exit rules, position sizing, the limits you want, and the backtest result it came from.
3. Your agent writes the script using the Candle skills (start with candle-strategy-to-bot, candle-risk-gate and candle-first-live-trade) and the SDK or CLI.
4. You review it and deploy it with a key whose spend limits you set.

Quant does not run backtests. Paste the result your agent got from the lab into Quant (the strategy definition, the window, the trades, the win rate and the drawdown) and ask it to turn the backtest into instructions for your agent. It explains and links; it does not start trades from chat.