Picking Industries According to Market State

Do U.S. industry portfolios perform predictably after similar U.S. equity market returns? In his May 2026 paper entitled "Industry Rotation Using Market-State Similarity",  Valeriy Zakamulin studies whether current market returns exploitably predict subsequent industry returns. Specifically, he:

1. Examines general market-to-industry monthly return predictability.
2. Assesses monthly market-to-industry predictability across ranges of market returns by:
   • Standardizing market excess return using a 120-month rolling window of past returns.
   • Identifying the 20% of market states most similar to the current state over a 480-month rolling window of past returns, excluding the latest 36 months.
   • Computing subsequent average industry returns for these similar market states.
3. Backtests a strategy that each month buys (sells) those of 30 industry portfolios with positive (negative) expected returns based on current market state. For robustness, he considers alternative industry classifications with 10, 12, 17, 38 and 48 portfolios and different market similarity parameters.

Using monthly returns for the U.S. stock market/industry portfolios and the monthly risk-free rate from the Kenneth French data library during July 1926 through December 2025, he finds that: