Freqtrade Backtester
/install freqtrade-backtester
Freqtrade Backtester
Run backtests on your trading strategies, interpret results, and iterate safely before going live.
What Backtesting Does
Backtesting runs your strategy against historical market data to see how it would have performed. It shows you win rates, losses, drawdowns, and exit reasons — everything you need to decide if a strategy is worth trading with real money.
Download Data
Historical data is cached, so download once and reuse it for many backtests.
⚠️ Download one pair at a time — passing multiple pairs to --pairs can cause errors. Run the command separately for each pair:
docker-compose run --rm freqtrade download-data \
--exchange kraken \
--pairs BTC/USDT \
--timeframe 5m \
--timerange 20240101-
Replace the pair, timeframe, and timerange as needed. The --timerange format is YYYYMMDD-YYYYMMDD; omit the end date to download through today.
Run a Backtest
docker-compose run --rm freqtrade backtesting \
--strategy YourStrategy \
--timerange 20240101-20260101 \
--export trades \
--export-filename user_data/backtest_results/my_backtest.json
Replace YourStrategy with your actual strategy class name. The --export trades flag creates a JSON file with detailed trade logs.
Key Metrics
- Win rate — Percentage of winning trades. >55% is decent; >60% is strong.
- Max drawdown — Largest peak-to-trough decline. Keep it \x3C20%; above 25% indicates excessive risk.
- Sharpe ratio — Risk-adjusted return. >1.0 is acceptable; >2.0 is excellent.
- Avg profit per trade — Average win/loss size. Wins should be larger than losses.
- Exit reasons — Breakdown of why trades closed (ROI, stop-loss, trailing stop, etc.).
The Key Insight
High win rate ≠ profitability. A 70% win rate with 5% average loss per trade loses money. A 40% win rate with 2% average wins and 1% average loss is profitable. Control your losses; the wins take care of themselves.
Example: A strategy with tight stops (-3%) outperforms one with loose stops (-7% or -8%), even if the loose version has more winners.
Iteration Pattern
- Run a backtest on a fixed time period (e.g., 6 months of data).
- Change one parameter (e.g., RSI threshold, stop-loss percentage).
- Backtest the same period again.
- Compare results — did it improve?
- If yes, keep it; if no, revert.
- Repeat until satisfied.
Always test across multiple market conditions (bull runs, bear markets, sideways consolidation) before going live. A strategy that works in one environment often fails in another.
Environment Variables
Freqtrade reads secrets from environment variables at runtime. Use the double-underscore format:
export FREQTRADE__EXCHANGE__KEY=your-api-key
export FREQTRADE__EXCHANGE__SECRET=your-api-secret
In your config.json, set these fields to empty strings:
{
"exchange": {
"key": "",
"secret": ""
}
}
Freqtrade will populate them from the environment at startup.
References
- Reading Results —
references/reading-results.md - Iteration Guide —
references/iteration-guide.md
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install freqtrade-backtester - After installation, invoke the skill by name or use
/freqtrade-backtester - Provide required inputs per the skill's parameter spec and get structured output
What is Freqtrade Backtester?
Run Freqtrade backtests for cryptocurrency trading strategies, interpret results, and iterate. Use when downloading historical data, running a backtest, read... It is an AI Agent Skill for Claude Code / OpenClaw, with 239 downloads so far.
How do I install Freqtrade Backtester?
Run "/install freqtrade-backtester" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Freqtrade Backtester free?
Yes, Freqtrade Backtester is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Freqtrade Backtester support?
Freqtrade Backtester is cross-platform and runs anywhere OpenClaw / Claude Code is available (linux, darwin, win32).
Who created Freqtrade Backtester?
It is built and maintained by Deonte Cooper (@djc00p); the current version is v1.0.3.