Using Post-Lasso to Select Related Industry Returns for Rotation

Summary

This research summary studies whether lagged returns from related industries can help predict industry returns for sector rotation. It motivates the question by noting that such predictability would not be expected in an ideal frictionless market, while real market conditions may permit it. To handle many candidate industry predictors, the authors use Lasso for variable selection, then refit ordinary least squares on the retained variables. This Post-Lasso step is intended to reduce the shrinkage bias of using Lasso estimates directly.

Rather than selecting the penalty parameter with k-fold cross-validation, which the report says can be sensitive to fold count and sampling choices, the study uses AIC. It describes rolling tests with at least 60 months of prior data over a stated historical period. The reported average selection is a small number of related industries per target, with the power-equipment and new-energy sector noted as an exception. The supplied text gives no out-of-sample return, risk, or benchmark results, so it supports a modeling approach but not a conclusion that the signal is profitable.

Key ideas

• Lagged returns from related industries are evaluated as predictors for industry rotation.
• Lasso selects candidate predictors, followed by ordinary least squares refitting to reduce shrinkage effects.
• The penalty parameter is chosen with AIC rather than k-fold cross-validation due to sensitivity concerns.
• The method is assessed with rolling windows using a minimum history requirement.
• The excerpt reports predictor-selection counts but omits portfolio performance and risk results.

Tags

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