Who it is forA developer with a crypto strategy idea who wants to see how it would have done before running it.
The problemWithout a simulation you only learn a strategy is bad after it loses money.
The outcomeBacktest results for your strategy on downloaded exchange data, plus a check for look ahead bias.
What you need first
- Python 3.12 or newer, pip, git and TA-Lib, or Docker
- An accurate system clock synced to NTP
- About 2GB RAM, 1GB disk and 2 vCPU recommended
- Exchange API keys only when you move past dry-run
Steps
- Follow the Docker Quickstart in the docs, or install natively with Python 3.12 or newer
- Run Freqtrade with
create-userdir, thennew-configandnew-strategyto scaffold your setup - Download candles with
download-datafor your pairs and timeframe - Run
backtestingand review past results withbacktesting-show - Check the strategy for look ahead bias with
lookahead-analysis
Prompt for your agent
Using freqtrade/freqtrade, scaffold a user directory and a strategy for <idea> on <pairs> and <timeframe>, download data, backtest it over <period>, and run lookahead-analysis. Report profit, drawdown and trade count, and any bias it found.
Caveat. Backtests on past candles do not guarantee future results. Not financial advice: paper trade first, and remember you can lose money.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for freqtrade/freqtrade and "Backtest a crypto strategy on real exchange data"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer