Cointegration-Based Pairs Trade

Entry Logic

• Long the underperforming asset and short the outperforming asset of a cointegrated pair when the spread deviates by 2 standard deviations from the equilibrium level.
• Confirmation requires the spread to show signs of reverting to the mean, such as a bullish or bearish candlestick pattern on the spread chart.
• Timeframe is 1-hour chart.
• Location context is a significant deviation from the long-term cointegrating relationship.
• Market condition should be stable, without major news impacting the pair.

Exit Logic

• Profit target is when the spread reverts to its equilibrium level (mean).
• No scaling out.
• Trailing stop is not used.
• Exit on signal failure if the spread widens to 3 standard deviations.
• Exit on opposite signal is not applicable.
• Exit on time expiration after 10 trading days.
• Exit on momentum loss if the spread moves sideways for more than 2 days.

Stop Loss Structure

• Hard stop is a 3 standard deviation move against the position.
• Soft stop is a breakdown in the cointegrating relationship, confirmed by an Engle-Granger test.
• Maximum dollar loss is $250 per trade.
• Maximum percent loss is 2% of account equity.
• Structural stop is not applicable.

Risk Management Framework

• Risk per trade is 1% of account equity.
• Maximum daily loss limit is 3% of account equity.
• Maximum weekly loss limit is 6% of account equity.
• Maximum drawdown is 20%.
• Risk-reward ratio requirement is a minimum of 1:1.5.

Position Sizing Model

• Sizing is based on the hedge ratio determined by the cointegration vector.
• Volatility adjustment is not used.
• Conviction sizing is not used.
• No scaling in.
• No scaling out.

Trade Filtering

• Only trade pairs with a p-value of less than 0.05 in the Engle-Granger cointegration test.
• Avoid pairs with a half-life of mean reversion that is too long (e.g., more than 30 days).
• Avoid holding positions through earnings or major company announcements.
• Time of day restrictions are not a primary concern.
• Avoid pairs in different sectors.

Context Framework

• Trend direction of the broader market is not a primary factor.
• VWAP and moving average relationships are not used for this strategy.
• Range location is determined by the standard deviation of the spread.
• Higher timeframe alignment is not applicable.

Trade Management Rules

• The trade is held until the spread reverts to the mean or the stop loss is hit.
• No adjustments are made to the position.
• No scaling in or out.
• The trade is managed as a single unit.

Time Rules

• Optimal trading window is during regular market hours.
• Avoid entering trades in the first and last 30 minutes of the trading day.
• Session notes are not applicable.

Setup Classification

• A+ setup: 2.5 standard deviation divergence with a strong cointegrating relationship (p-value < 0.01).
• A setup: 2.0 standard deviation divergence with a p-value < 0.05.
• B setup: 1.5 standard deviation divergence.
• C setup: Less than 1.5 standard deviation divergence or a weak cointegrating relationship.

Market Selection Criteria

• Instruments are pairs of assets that have a long-term economic relationship, such as two companies in the same industry or a commodity and a related currency.
• Minimum daily volume and liquidity are required for both assets.
• The spread should exhibit stationary properties.

Statistical Edge Metrics

• Expected win rate is 60%.
• Average win is 1.5R.
• Average loss is 1R.
• Profit factor is 1.8.
• Expectancy per trade is +0.4R.

Failure Conditions

• The cointegrating relationship breaks down due to a fundamental change in one of the assets.
• The market experiences a systemic shock that affects all assets.
• Avoid if the pair's historical relationship is not stable.

Psychological Rules

• Must have the patience to wait for the statistical signal to trigger.
• Trust the cointegration model and avoid emotional decision-making.
• Cut losses immediately if the relationship breaks down.

Advanced Components

• The Johansen test can be used to test for cointegration in a multivariate setting.
• A Kalman filter can be used to dynamically estimate the hedge ratio.
• Correlation filters are not the primary tool; cointegration is the key.
• Multi-timeframe alignment is not applicable.

Location

• Strongest in markets where assets have strong economic links.
• Weakest when these links are disrupted by external events.
• The strategy's success depends on the persistence of the cointegrating relationship.