Structural Break Tests for Detecting Market Regime Changes
Summary
The article explains structural break tests as features for identifying changes in market behavior. It covers the Chu–Stinchcombe–White test for departures from a no-trend baseline, Chow-type Dickey–Fuller testing for a single unknown break, and SADF and sub- or super-martingale tests for explosive behavior. It also describes QADF and CADF as ways to reduce the influence of an extreme ADF window, and advises using log prices for ADF-based tests rather than raw prices.
The tests produce time-indexed statistics that can help a downstream model select or condition trading strategies. The article discusses how expanding versus rolling windows affect computation and the duration of regimes detected. It reports that optimized inner loops were 32–50 times faster than the book snippets, and describes a two-year daily lookback as one practical choice. These statistics do not establish profitable trading signals: thresholds, window choices, and strategy use still require validation, and some implementation details are absent from the supplied text.
Key ideas
- CUSUM and explosiveness tests measure different forms of departure from stable market behavior.
- SADF searches over nested sample windows to detect explosive episodes that a single-break test can miss.
- QADF and CADF help assess whether an extreme supremum reflects a broad pattern or one outlier window.
- Log-price inputs make ADF results more interpretable in terms of return behavior.
- Window selection trades computational cost against sensitivity to regimes of different durations.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.