Backtest Comparison & Portfolio Combination

Compare multiple backtests side-by-side and combine them into weighted portfolios to analyze diversification benefits and portfolio performance.

Overview

The comparison tool helps you:

• Evaluate variants - Compare different strategy configurations
• Find diversification - Analyze correlation between strategies
• Build portfolios - Combine backtests with custom weights
• Optimize allocation - Test different weight distributions

Starting a Comparison

From Backtests List

1. Select 2-10 backtests using checkboxes
2. Click "Compare & Combine" button
3. Comparison page opens with your selected backtests

Requirements

• Minimum: 2 backtests
• Maximum: 10 backtests
• Status: Only completed backtests can be compared

Comparison View

Metrics Table

The comparison table shows key performance metrics side-by-side for easy comparison:

Customizable Columns:

• Click "Configure Columns" to select which metrics to display
• Choose from all available backtest metrics
• Default view: Average P&L per day, Win rate, CVaR, Sortino ratio
• Customize to focus on metrics most relevant to your analysis

Visual Features:

• 🟢 Green background = Best performing value in that column
• 🔴 Red background = Worst performing value in that column
• Sortable columns - Click header to sort (asc → desc → reset)
• Sticky name column - Remains visible when scrolling horizontally

Equity Curves

Visualize all equity curves on a single chart:

• Overlay view - All curves on one chart
• Toggle visibility - Show/hide individual curves
• Zoom & pan - Examine specific periods
• Legend - Click to toggle curves on/off

Identify:

• Which strategies perform better in different market conditions
• Drawdown periods across strategies
• Recovery patterns and consistency

Correlation Matrix

Understand how strategies move together:

Interpretation:

• +1.0 = Perfect positive correlation (move together)
• 0.0 = No correlation (independent)
• -1.0 = Perfect negative correlation (move opposite)

Color Scale:

• 🔴 Red (high correlation) - Limited diversification
• 🟡 Yellow (moderate) - Some diversification benefit
• 🟢 Green (low/negative) - Excellent diversification

Minimum Data: Requires 30+ overlapping trading days between backtests.

Finding Good Diversification:

• Look for correlation < 0.5
• Negative correlation provides hedging benefits
• High correlation (>0.8) offers minimal diversification

Portfolio Combination

Combine backtests into a weighted portfolio to:

• Test diversification benefits
• Optimize risk-adjusted returns
• Evaluate multi-strategy approaches

Combination Modes

1. Equal Weight (Standard Portfolio)

• All weights sum to exactly 1.0
• Example: [0.5, 0.5] = 50% in each strategy
• Use for: Balanced diversification

2. Custom Weight (Flexible Allocation)

• Weights can sum to any value
• Example: [0.7, 0.8] = 150% total (1.5x leverage)
• Use for: Risk-adjusted or leveraged portfolios

3. Combination (Parallel Strategies)

• Strategies run simultaneously at full or reduced scale
• Example: [1.0, 1.0, 0] = Run first two at full scale
• Use for: Non-overlapping strategies (different times/conditions)

Creating a Portfolio

1. Select Backtests: Check the strategies to include
2. Choose Mode: Select combination mode
3. Adjust Weights: Set weight for each strategy
4. Combine: Click "Combine Portfolio"
5. View Results: See combined equity curve and metrics

Portfolio Results

The combined portfolio shows:

• Combined equity curve - Portfolio performance over time
• Recalculated metrics - All metrics computed from combined returns
• Portfolio row - Highlighted in blue at top of comparison table

Common Questions

Q: Why can't I compare more than 10 backtests? A: Performance and UI clarity. For more, run multiple comparisons.

Q: What if correlation matrix is empty? A: Need 30+ overlapping days. Check date ranges align.

Q: Can I save a portfolio as a strategy? A: Not directly. Clone individual strategies and note portfolio weights.

Q: Do weights have to sum to 1.0? A: Only in "Equal Weight" mode. "Custom Weight" allows any total.

Q: Why are combined metrics different from averaging? A: Metrics are recalculated from the combined equity curve for accuracy.

Q: How do I handle different date ranges? A: System uses overlapping dates only. Non-overlapping days show zero contribution.

Next Steps

• Learn about Workspace Organization to manage strategies and backtests
• Read Analyzing Results for detailed metrics explanations
• Explore Strategy Basics to optimize your configurations