crypto_trader/docs/user_manual/backtesting.md

Backtesting Guide

This guide explains how to use the backtesting feature to evaluate trading strategies on historical data.

Running a Backtest

1. Navigate to the Backtesting page
2. Configure your backtest in the form:
   • Strategy: Select a strategy from the dropdown (required)
   • Symbol: Trading pair to test (e.g., BTC/USD)
   • Exchange: Data source exchange (e.g., coinbase)
   • Timeframe: Data timeframe (1m, 5m, 15m, 1h, 4h, 1d) - 1h recommended for most strategies
   • Start Date: Beginning of test period (required)
   • End Date: End of test period (required)
   • Initial Capital: Starting capital in USD (default: $100)
   • Slippage (%): Expected slippage percentage (default: 0.1%)
   • Fee Rate (%): Trading fee percentage (default: 0.1%)
3. Click Run Backtest
4. A progress overlay will appear showing the backtest is running
5. An operations panel will show the running backtest with status
6. Wait for completion (you'll receive a success notification)
7. Review results in the Backtest Results section below

Understanding Results

The backtest results include:

• Total Return: Overall percentage return
• Sharpe Ratio: Risk-adjusted return metric (higher is better)
• Sortino Ratio: Downside risk-adjusted return (higher is better)
• Max Drawdown: Largest peak-to-trough decline
• Win Rate: Percentage of profitable trades
• Total Trades: Number of trades executed
• Final Value: Portfolio value at end of backtest

Exporting Results

After a backtest completes, you can export the results:

1. In the backtest results section, find the export buttons
2. Export CSV:
   • Click Export CSV button
   • Downloads a CSV file with all trades from the backtest
   • File includes: timestamp, side, price, quantity, value
3. Export PDF:
   • Click Export PDF button
   • Generates a comprehensive PDF report
   • Includes charts, metrics, and trade analysis

Both exports are automatically named with the current date for easy organization.

Advanced Backtesting Features

Walk-Forward Analysis

Walk-forward analysis provides robust parameter optimization by using rolling windows:

1. Training Period: Strategy parameters are optimized on training data (e.g., 90 days)
2. Testing Period: Optimized parameters are tested on out-of-sample data (e.g., 30 days)
3. Rolling Window: Window advances by step size (e.g., 30 days) and process repeats

This method prevents overfitting and provides more realistic performance estimates than single-period optimization.

Benefits:

• Prevents overfitting to historical data
• Tests strategy robustness across different market conditions
• Provides confidence intervals for parameter estimates
• Validates strategy performance on unseen data

Monte Carlo Simulation

Monte Carlo simulation tests strategy robustness by running thousands of random scenarios:

• Random Parameter Variation: Tests strategy performance across parameter ranges
• Statistical Analysis: Provides distribution of returns, Sharpe ratios, and drawdowns
• Confidence Intervals: Shows expected performance ranges (e.g., 95% confidence)
• Risk Assessment: Identifies worst-case scenarios and tail risks

Use Monte Carlo simulation to:

• Validate strategy robustness
• Assess parameter sensitivity
• Understand potential downside risks
• Estimate performance under various market conditions

Parameter Optimization

Parameter optimization allows you to automatically find the best strategy parameters using multiple algorithms:

• Grid Search: Exhaustive search across parameter grid (best for small parameter spaces)
• Bayesian Optimization: Efficient exploration using Gaussian process (best for expensive evaluations)
• Genetic Algorithms: Evolutionary search that finds good solutions efficiently

Optimization metrics include:

• Sharpe Ratio (risk-adjusted returns)
• Total Return
• Maximum Drawdown
• Win Rate

When available via the backend API, you'll be able to:

• Select parameters to optimize
• Set parameter ranges
• Choose optimization method
• View optimization progress
• Compare optimization results

Interpreting Metrics

• Sharpe Ratio > 1: Good risk-adjusted returns
• Max Drawdown < 20%: Acceptable risk level
• Win Rate > 50%: More winning than losing trades

Backtesting Guide

Learn how to test your trading strategies on historical data.

What is Backtesting?

Backtesting is the process of testing a trading strategy on historical data to evaluate its performance before risking real money.

Running a Backtest

1. Navigate to Backtest view
2. Select a strategy
3. Configure backtest parameters:
   • Start Date: Beginning of test period
   • End Date: End of test period
   • Initial Capital: Starting capital
   • Symbol: Trading pair to test
   • Timeframe: Data timeframe
4. Click "Run Backtest"
5. Wait for completion
6. Review results

Backtest Parameters

Time Period

• Start Date: When to begin the backtest
• End Date: When to end the backtest
• Duration: Length of test period
• Longer periods provide more reliable results

Capital Settings

• Initial Capital: Starting amount (e.g., $10,000)
• Currency: Base currency (USD, EUR, etc.)

Market Settings

• Symbol: Trading pair (BTC/USD, ETH/USD, etc.)
• Timeframe: Data granularity (1m, 5m, 1h, 1d)
• Exchange: Historical data source

Understanding Results

Performance Metrics

• Total Return: Overall profit/loss percentage
• Final Capital: Ending portfolio value
• Sharpe Ratio: Risk-adjusted return measure
• Sortino Ratio: Downside risk-adjusted return
• Max Drawdown: Largest peak-to-trough decline
• Win Rate: Percentage of profitable trades

Trade Analysis

• Total Trades: Number of trades executed
• Winning Trades: Number of profitable trades
• Losing Trades: Number of unprofitable trades
• Average Win: Average profit per winning trade
• Average Loss: Average loss per losing trade
• Profit Factor: Ratio of gross profit to gross loss

Charts

• Equity Curve: Portfolio value over time
• Drawdown Chart: Drawdown periods
• Trade Distribution: Win/loss distribution
• Monthly Returns: Performance by month

Realistic Backtesting

Crypto Trader includes realistic backtesting features:

Slippage

Slippage simulates the difference between expected and actual execution prices.

• Default: 0.1% for market orders
• Configurable: Adjust based on market conditions
• Market Impact: Larger orders have more slippage

Fees

Trading fees are automatically included:

• Maker Fees: For limit orders (typically 0.1%)
• Taker Fees: For market orders (typically 0.2%)
• Exchange-Specific: Fees vary by exchange

Order Execution

• Market Orders: Execute at current price + slippage
• Limit Orders: Execute only if price reaches limit
• Partial Fills: Large orders may fill partially

Parameter Optimization

Optimize strategy parameters for better performance:

1. Select strategy
2. Choose parameters to optimize
3. Set parameter ranges
4. Select optimization method:
   • Grid Search: Test all combinations
   • Genetic Algorithm: Evolutionary optimization
   • Bayesian Optimization: Efficient parameter search
5. Run optimization
6. Review results and select best parameters

Best Practices

1. Use Sufficient Data: Test on at least 6-12 months of data
2. Avoid Overfitting: Don't optimize too aggressively
3. Test Multiple Periods: Verify performance across different market conditions
4. Consider Fees: Always include realistic fees
5. Check Slippage: Account for execution costs
6. Validate Results: Compare with paper trading

Limitations

Backtesting has limitations:

• Past Performance: Doesn't guarantee future results
• Market Conditions: Markets change over time
• Data Quality: Results depend on data accuracy
• Execution: Real trading may differ from simulation

Exporting Results

Export backtest results for analysis:

1. Click "Export Results"
2. Choose format:
   • CSV: For spreadsheet analysis
   • PDF: For reports
3. Save file

Troubleshooting

No results?

• Check date range has data
• Verify symbol is correct
• Check strategy parameters

Unrealistic results?

• Verify fees are enabled
• Check slippage settings
• Review data quality