Cointegration Node
Tests if time series move together in equilibrium
Overview
Cointegration tests whether two or more time series that are individually non-stationary move together in a long-term equilibrium relationship. Unlike correlation which measures linear relationships, cointegration identifies stable relationships that persist over time, enabling profitable pairs trading even when price movements are random.
Two assets can be cointegrated if they're driven by similar fundamental forces and their prices deviate from a stable equilibrium only temporarily. When one asset moves significantly from the equilibrium, the other must follow to restore balance, creating predictable mean reversion opportunities for statistical arbitrage strategies.
Formula & Calculation
Parameters
| Parameter | Default | Description |
|---|---|---|
| series1, series2 | Required | Two price series to test for cointegration |
| lookback | 252 | Number of observations for test |
| confidence | 95% | Significance level for cointegration test |
Common Use Cases
1. Pairs Trading Setup
Find cointegrated stock pairs and trade deviations from equilibrium. When spread widens, short the outperformer and long the underperformer, betting on mean reversion. This creates market-neutral exposure suitable for any market regime.
2. Statistical Arbitrage
Use cointegration as foundation for algorithmic trading. Cointegrated pairs have mathematically guaranteed reversion (unlike correlation-based pairs), enabling profitable systematic strategies with quantifiable risks.
3. Portfolio Hedging
Use cointegrated relationships to hedge positions efficiently. If you own Stock A (which is cointegrated with Index), you can hedge with Stock B (also cointegrated with Index) rather than futures, avoiding carrying costs.
4. Cross-Asset Trading
Find cointegration between different asset classes (stock-bond, commodity-currency, etc.). These relationships are more stable than correlations and less likely to break down during crises.
Advantages & Limitations
Advantages
- •Identifies stable long-term relationships
- •Enables statistical arbitrage strategies
- •More reliable than correlation
- •Creates market-neutral opportunities
- •Hedging cost reduction
Limitations
- •Relationships break suddenly (structural breaks)
- •Requires multiple correlated assets
- •Complex calculation and interpretation
- •Statistically expensive to test
- •Regime-dependent behavior
Tips & Best Practices
📊 Monitor Spread Half-Life
Calculate how long it takes the spread to revert halfway. Shorter half-life = faster mean reversion = better trading opportunity.
🔄 Test Regularly
Cointegration relationships can break. Retest monthly and replace pairs that lose cointegration.
⚡ Use Multiple Tests
Combine Johansen test with ADF (Augmented Dickey-Fuller) for robustness. Different methods catch different relationships.
⚠️ Watch Structural Breaks
Cointegration can break suddenly due to economic changes. Set strict stop-losses and position size accordingly.
Example Strategy: Pairs Trading
1. Find cointegrated pairs using Johansen test (min 252-day period)
2. Calculate equilibrium price using hedging ratio from regression
3. Entry: Spread = (Price A - Beta × Price B) exceeds 2 standard deviations
4. Exit: Spread reverts to mean or 2-3X standard deviation exceeded
5. Position size: Scale based on spread volatility and account risk