Backtrack the
trade before
you risk it
Describe a strategy in plain words. Backtrack turns it into rules, tests them on three years of real prices, and only lets your AI agent trade it after paper trading and your approval.

From a sentence
to a tested strategy
- English or Indonesian
- Unclear text is flagged, never guessed
- Signals at the close, fills at the next open
- Stop loss before take profit
- Slippage on every fill
- Against buy & hold and the S&P 500
/ Honest backtesting /
Most strategies don't pass the gate
A strategy moves draft → backtested → paper → live. It only reaches paper if it beats simply holding, risk for risk. Only you can approve live.
Don't just guess.
Backtrack.
These are real runs on real prices, Nov 2023 to Nov 2026. Most simple rules lose to buy & hold in a market that went up. Backtrack says so before your money finds out.
Try your own {dashboard}A backtest should be
allowed to say no
Strategy tools usually flatter you: perfect fills, no slippage, results measured against nothing.
Backtrack fills every signal at the next open, assumes the stop hits before the target, and puts every result next to simply holding the same stocks.
Most ideas don't survive that. That is the point: you find out on paper, not with your savings.
And when one does survive, it still has to earn its way through paper trading. Live is your decision, never the agent's.

Put your strategy on stage
Runs on your computer. Your strategies and positions never leave it.
Traders
Open the strategy lab: write a rule, see the curve, save it.
npx backtrack-mcp dashboard # then open localhost:6200/app
Agent builders
Give Claude or Cursor eleven tools, from parse to check exits.
claude mcp add backtrack -- npx -y backtrack-mcp claude mcp add brakeline -- npx -y brakeline