Using a Gaussian Mixture Model to Classify Market Regimes

Summary

The article describes an unsupervised approach to identifying market regimes from the returns of 17 US factors beginning in 1970. A Gaussian mixture model clusters observations by their return distributions; cross-validation is used to choose four clusters. The authors interpret them as crisis, steady, inflation, and fragile regimes by comparing factor returns, volatility, and correlations within each cluster.

The historical classification assigns each period to its most probable regime. The article reports that crisis labels coincide with the 1987, 2008, and early 2020 shocks, while inflation labels are concentrated in the 1970s and 1980s. These patterns suggest potential use in portfolio risk monitoring and allocation decisions. However, regime names are interpretations of statistical clusters, and the article offers no detailed out-of-sample performance metrics. It also stresses that the model classifies the current state rather than forecasting regime changes, so it should not be treated as a predictive trading signal.

Key ideas

• The model fits a Gaussian mixture to returns from 17 US market factors to identify regimes.
• Cross-validation selects four clusters, interpreted as crisis, steady, inflation, and fragile conditions.
• The article describes distinct factor return, volatility, and correlation patterns across the four regimes.
• Historical labels align with several major market disruptions, though detailed validation metrics are not provided.
• The model estimates the current regime but does not forecast future regime transitions.

Tags

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