Stock Pairs Trading

Start with stocks in the same sector that have obvious economic relationships: competitors (BHP/RIO), similar business models (WOW/COL), or same industry (big 4 banks). Then test correlation (>0.7) and cointegration (ADF p-value <0.05). Good pairs have high correlation, confirmed cointegration, and reasonable half-life.

Minimum A$50,000 to trade a diversified portfolio of pairs. Each pair requires approximately A$10,000-20,000 per leg for meaningful position sizes. With less capital, you can trade a single pair but face concentration risk.

Use CFDs for the short leg through brokers like CMC Markets or IG. CFDs allow you to take short positions without borrowing physical shares. The trade-off is overnight financing costs. Some brokers also offer fractional short selling.

Monitor daily at minimum. Check the z-score to see if exit or stop levels are approaching. Also verify correlation remains stable. Weekly recalculation of hedge ratio is recommended. More frequent monitoring during volatile periods.

Small changes are normal - adjust at weekly recalculation. If the ratio changes more than 20%, the relationship may be breaking down. Consider exiting the trade early rather than adjusting into a deteriorating relationship.

If you are long the stock paying dividend, you receive it (plus franking credits). If you are short, you owe the dividend to the lender. Factor dividends into expected returns. Consider closing or adjusting positions before ex-dividend dates if the dividend is significant.

Yes, that is the beauty of market neutrality. In trending markets, both stocks tend to move together, so your spread trade is largely unaffected by market direction. You profit from relative value, not market direction.

Successful pairs traders typically achieve 60-70% win rates with modest average gains (1-3% per pair). The key is the combination of reasonable win rate and risk/reward. Losses should be cut at the stop (z-score 3.0) to maintain edge.

Use Python's pykalman or filterpy libraries. Model the hedge ratio as a hidden state that evolves according to process noise. Observations are the stock prices. Tune process noise (how much ratio can change) and observation noise (measurement error). Backtest to calibrate parameters.

Features include: current z-score, z-score momentum, half-life, correlation trend, cointegration p-value, fundamental ratio (P/E difference), sector sentiment, macro indicators (AUD, commodities, rates). Use cross-validation to select predictive features and avoid overfitting.

Simulate scenarios where correlation breaks down across multiple pairs simultaneously. Historical scenarios: March 2020 (COVID), 2011 (European debt crisis). Calculate maximum drawdown if all pairs hit stops together. Ensure portfolio can survive worst-case scenarios.

Avoid: when fundamental change explains divergence (M&A, spin-off, management change), when correlation has been declining for weeks, when the divergence is in the direction of fundamental difference (better stock getting more expensive), or when liquidity or borrow is problematic.