backtesting

backtesting

Purpose

This skill enables OpenClaw to perform backtesting on historical financial data, simulating trading strategies to measure metrics like returns, drawdowns, and Sharpe ratio. Use it to validate strategies before live trading, identifying potential flaws in logic or risk exposure.

When to Use

Apply this skill when developing or refining trading strategies, such as evaluating a moving average crossover on stock data. Use it for quantitative analysis in algorithmic trading, portfolio optimization, or risk assessment, especially when historical data is available and strategies need empirical validation.

Key Capabilities

• Simulate trades using historical price data from CSV, JSON, or API sources.
• Compute performance metrics including total return, volatility, maximum drawdown, and risk-adjusted returns.
• Support for common indicators like SMA, RSI, and Bollinger Bands via integrated libraries.
• Handle multiple assets or portfolios, with options for transaction costs, slippage, and position sizing.
• Output results in JSON format for easy parsing and visualization.

Usage Patterns

To backtest a strategy, provide a Python script defining the strategy logic, historical data source, and parameters. Always set the API key via environment variable $OPENCLAW_API_KEY before running. For CLI, use flags to specify inputs; for API, send a POST request with a JSON payload. Validate data integrity first by checking for missing values or incorrect timestamps. Run tests iteratively, adjusting parameters based on initial results.

Common Commands/API

Use the OpenClaw CLI for quick tests or the REST API for programmatic integration. Authentication requires setting $OPENCLAW_API_KEY in your environment.

• CLI Command: Run a backtest with a strategy file and data source.

  openclaw backtest --strategy strategy.py --data historical.csv --start-date 2020-01-01 --end-date 2023-01-01 --capital 10000
  

  Copy

  This executes the strategy on the specified date range with initial capital.

• API Endpoint: POST to /v1/backtest with JSON body.

  curl -X POST https://api.openclaw.ai/v1/backtest \
  -H "Authorization: Bearer $OPENCLAW_API_KEY" \
  -d '{"strategy": "def trade(data): return buy if data['close'] > data['sma'] else sell", "data_source": "historical.csv", "params": {"start_date": "2020-01-01"}}'
  

  Copy

  Expect a JSON response with keys like "returns" and "drawdown".

• Config Format: Use YAML for strategy configurations, e.g.,

  strategy:
    name: moving_average
    parameters:
      period: 20
      threshold: 0.05
  

  Copy

  Load this via CLI with --config strategy.yaml.

Integration Notes

Integrate with data providers like Yahoo Finance or Alpha Vantage by specifying URLs in the data source flag (e.g., --data https://example.com/data.csv). For Python workflows, import OpenClaw as a library: import openclaw; result = openclaw.backtest(strategy_file='strategy.py', data='historical.csv'). Ensure compatibility by using Python 3.8+ and handling dependencies via pip install openclaw-finance. If combining with other skills, chain outputs; for example, use results from a "data-fetch" skill as input here.

Error Handling

Anticipate errors like invalid data formats, missing API keys, or strategy failures. Check for $OPENCLAW_API_KEY before execution; if absent, the command will exit with code 401. Handle runtime errors by wrapping calls in try-except blocks, e.g.,

try:
    openclaw.backtest(strategy='strategy.py', data='historical.csv')
except ValueError as e:
    print(f"Data error: {e} - Verify CSV columns match expected format.")

Common issues: Invalid dates return code 400; use --verbose flag for detailed logs. Always validate strategy code for syntax errors before running.

Concrete Usage Examples

1. Backtest a simple SMA crossover strategy on Apple stock data:

   openclaw backtest --strategy examples/sma_crossover.py --data aapl_historical.csv --params '{"short_period": 10, "long_period": 30}' --capital 5000
   

   Copy

   This simulates buying when short SMA crosses above long SMA, outputting metrics like 15% annual return.

2. Evaluate a portfolio strategy with multiple assets and fees:

   curl -X POST https://api.openclaw.ai/v1/backtest -H "Authorization: Bearer $OPENCLAW_API_KEY" -d '{"strategy": "portfolio_alloc.py", "data_source": ["aapl.csv", "goog.csv"], "params": {"fee_per_trade": 5.0}}'
   

   Copy

   This assesses a diversified portfolio, calculating net returns after fees, e.g., 12% with 8% drawdown.

Graph Relationships

• Related to: trading (dependency for strategy execution)
• Related to: financial-analysis (provides input data and metrics)
• Connected to: risk-management (outputs risk metrics for further analysis)
• Links with: data-fetch (uses fetched historical data as input)