A quantitative trading engine toolkit — create strategies, run backtests, and analyze results using natural language.
create-strategy-demo.mp4
Global install with uv (recommended):
uv tool install "xqtrader[cli]"Or add as a project dependency:
uv add "xqtrader[cli]"Or with pip:
pip install "xqtrader[cli]"Note: zsh users must quote
"xqtrader[cli]"to avoid glob interpretation.
# uv global install
uv tool upgrade xqtrader
# uv project dependency
uv add --upgrade "xqtrader[cli]"
# pip
pip install --upgrade "xqtrader[cli]"XQTrader provides skill files (in the skills/ directory) that teach AI coding tools how to create strategies and run backtests.
Copy the contents of the skills/ directory to the appropriate location for your tool:
| Tool | Target path |
|---|---|
| Claude Code | Project: .claude/skills/, or personal: ~/.claude/skills/ |
| OpenCode | .opencode/skills/, .claude/skills/, or .agents/skills/ |
| OpenClaw | Configure in ~/.openclaw/openclaw.json |
For example, with Claude Code:
cp -r skills/* .claude/skills/Once configured, you can use natural language or slash commands (/create-strategy, /backtest) for all operations.
Describe your strategy idea and Claude will generate a complete strategy file:
Create an RSI overbought/oversold reversal strategy
Build a MACD crossover strategy for 1h timeframe with trend filtering
I want a Bollinger Band breakout strategy for ranging markets
Create an event contract strategy based on RSI and volume to predict short-term direction
Or use the slash command for quick generation:
/create-strategy rsi_reversal
The generated strategy file includes indicator declarations and signal logic, ready for backtesting.
Describe your backtest in natural language:
Backtest rsi_reversal.py on BTC/USDT for the last 3 months
Test this strategy on ETH/USDT from 2025-01-01 to 2025-03-01, 1h interval, 50k capital
Compare this strategy's performance on BTC and ETH
Backtest this strategy with event contracts on BTC/USDT for the last month
Or use the slash command:
/backtest rsi_reversal.py --symbol BTC/USDT --start 2025-01-01 --end 2025-03-01
Claude will validate the strategy, execute the backtest, and explain the results including returns, win rate, max drawdown, and Sharpe ratio.
After backtesting, continue refining with natural language:
Win rate is too low, try changing the RSI oversold threshold from 30 to 25
Add a stop-loss at 2% drawdown
Switch to 4h timeframe and re-run the backtest
1. Describe your strategy idea → Claude generates the strategy file
2. Review the generated code → Confirm or request changes
3. Say "run a backtest" → Claude executes and explains results
4. Suggest adjustments → Iterate and optimize
| Contract Type | Signals |
|---|---|
| Perpetual/Futures | LONG, SHORT, CLOSE_LONG, CLOSE_SHORT, CLOSE |
| Event Contracts | UP, DOWN, HOLD |
- Natural Language Workflow: Create and test strategies conversationally via Claude Code
- 60+ Technical Indicators: RSI, MACD, Bollinger Bands, ATR, KDJ, and more
- Multi-Exchange Support: 100+ exchanges via CCXT
- Futures Trading: Perpetual futures with hedge mode position management
- Risk Management: Multi-level risk control (daily loss 3.5%/5%, max drawdown 10%/15%)
- CLI Guide — Command-line reference
- Development Guide — Architecture and developer docs
MIT License - see LICENSE for details.