Combining Mean Reversion and Momentum Across Market Regimes

Summary

The article reviews a proposed arbitrage portfolio that combines equity mean reversion with momentum across stock market indices. Its study separates data into an in-sample period from November 2005 to October 2007 and an out-of-sample period from November 2007 to October 2009, spanning the Quant Quake and the 2008 financial crisis. The mean-reversion component uses S&P 500 closing prices, while the momentum component uses prices for ten indices; quadratic programming is used for portfolio rebalancing.

The reported discussion finds strong in-sample mean-reversion results but cautions that optimizing to historical data can impair generalization. The mean-reversion model also failed to maintain its intended market-neutral exposure, so its value as a hedge is uncertain. Momentum appeared effective near the end of 2008, though the article suggests that recovery conditions may explain some of that performance. It raises regularization and cross-validation as possible ways to improve robustness, but supplies no evidence that these remedies solve the problem. Out-of-sample results and live trading remain essential checks.

Key ideas

• The study combines a mean-reversion strategy with a momentum strategy to diversify arbitrage exposures.
• The mean-reversion component did not consistently achieve its intended market neutrality.
• Strong in-sample performance can reflect parameter fitting and may not carry over to new market conditions.
• Momentum performed well near the end of the 2008 crisis, potentially benefiting from a market recovery.
• Regularization and cross-validation are proposed as possible tools for improving generalization, not proven solutions.

Tags

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