Predicting Factor Decay: ML Models for Cross-Factor Predictability Erosion
Abstract
We develop ML models that predict when cross-factor predictability will erode, enabling prospective risk management. A decay prediction model achieves C-index = 0.93 on held-out factor pairs, providing early warning of signal erosion---the key predictor is initial signal strength: stronger signals decay faster (``regression to the mean''). An online regime detector enables real-time classification with 6.6-day average detection delay. Prior work explains why factors decay (arbitrage, publication effects); we predict when decay will occur. Limitation: The primary HML→SMB signal does not replicate in post-break OOS (2013--2024); this null result is consistent with genuine decay but limits confirmatory claims. Motivation: Among 30 directed Fama-French factor pairs (1990--2024), predictability targeting the Size factor (SMB) decays fastest. Of 7 pairs showing exponential decay (R2 > 0.3), 4 target SMB (Mkt-RF→SMB, CMA→SMB, HML→SMB, RMW→SMB; half-lives 2.4--4.1 years). This SMB-concentration pattern motivates the predictive models. Methodology: Using a regime-conditional Granger framework with Student-t HMMs, we document structural breaks clustering around June 1998 (p < 10−10). A pre-break OOS test (HMM trained 1990--1995, tested 1996--2007) confirms signals existed: HML→SMB shows p = 0.013 (Elevated) and p < 0.001 (Crisis). The primary HML→SMB signal does not replicate in post-break OOS (2013--2024), consistent with genuine decay rather than overfitting---after ∼6 half-lives, the expected signal is <1% of its 1990s peak. International analysis finds 2/4 non-US markets with Bonferroni-surviving OOS effects; correction for full specification search was not applied, so results remain exploratory.
Related
- Predicting resistance to fluoroquinolones among patients with rifampicin-resistant tuberculosis: A cross-country validation study
- BER-PEF: Unified Human Mobility Predictability Evaluation via Bayes Error Rate Estimation
- Fitness, predictability and equilibrium in biological dynamics *
- Quantifying volumetric erosion from symbolic expression of landslide morphology
- The erosion of proof in paradigm transitions: the rise of narrative rationality in health technology assessment
BibTeX
@misc{lee2026predicting,
title = {{Predicting Factor Decay: ML Models for Cross-Factor Predictability Erosion}},
author = {Chorok Lee},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.03668},
eprint = {2611.03668},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.03668},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Predicting Factor Decay: ML Models for Cross-Factor Predictability Erosion
AU - Chorok Lee
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.03668
UR - https://arxiv.org/abs/2611.03668
AB - We develop ML models that predict when cross-factor predictability will erode, enabling prospective risk management. A decay prediction model achieves C-index = 0.93 on held-out factor pairs, providing early warning of signal erosion---the key predictor is initial signal strength: stronger signals decay faster (``regression to the mean''). An online regime detector enables real-time classification with 6.6-day average detection delay. Prior work explains why factors decay (arbitrage, publication effects); we predict when decay will occur. Limitation: The primary HML→SMB signal does not replicate in post-break OOS (2013--2024); this null result is consistent with genuine decay but limits confirmatory claims. Motivation: Among 30 directed Fama-French factor pairs (1990--2024), predictability targeting the Size factor (SMB) decays fastest. Of 7 pairs showing exponential decay (R^2 > 0.3), 4 target SMB (Mkt-RF→SMB, CMA→SMB, HML→SMB, RMW→SMB; half-lives 2.4--4.1 years). This SMB-concentration pattern motivates the predictive models. Methodology: Using a regime-conditional Granger framework with Student-t HMMs, we document structural breaks clustering around June 1998 (p < 10^−10). A pre-break OOS test (HMM trained 1990--1995, tested 1996--2007) confirms signals existed: HML→SMB shows p = 0.013 (Elevated) and p < 0.001 (Crisis). The primary HML→SMB signal does not replicate in post-break OOS (2013--2024), consistent with genuine decay rather than overfitting---after ∼6 half-lives, the expected signal is <1% of its 1990s peak. International analysis finds 2/4 non-US markets with Bonferroni-surviving OOS effects; correction for full specification search was not applied, so results remain exploratory.
ER -