Pearson-Correlation Pairs Trading with Return-Divergence Signals

Summary

This implementation describes an equity pairs strategy that selects stocks with highly correlated historical returns, then compares each stock’s return with a portfolio of its selected peers. It estimates a regression coefficient during a formation period and uses that coefficient, peer returns, and a risk-free rate to calculate return divergence. In the trading period, it ranks stocks by the prior month’s divergence and takes long positions in the highest-ranked group and short positions in the lowest-ranked group. The portfolio can use equal weighting or a correlation-based weighting option.

The document explains the signal logic and exposes the selected pairs, regression coefficients, and monthly position signals. It provides code but no backtest results or evidence that the strategy is profitable. Its usefulness is therefore methodological rather than empirical. The implementation also appears to calculate pair portfolio returns differently in signal generation than in formation, and its weighting calculations warrant review before use. Results may depend on the formation window, universe, transaction costs, and data quality.

Key ideas

• The formation stage ranks other stocks by Pearson correlation with each stock’s historical returns.
• A regression coefficient links each stock’s returns to the returns of its selected peer portfolio.
• Return divergence is ranked monthly to assign long positions to the highest group and short positions to the lowest group.
• The implementation offers equal-weighted and correlation-based portfolio weighting choices.
• The document reports no performance tests, and implementation details should be validated before relying on its signals.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.