Rethinking Trend Following: Optimal Regime-Dependent Allocation

Most trend-following research focuses on signal construction: how to detect trends better, faster, or earlier. The paper asks a different question, and arguably a more important one for investors: once a market regime has been identified, what is the optimal portfolio exposure in that regime?

That is the central novelty of the paper which is available here.

Traditional time-series momentum strategies typically impose exposures mechanically. In the standard two-regime version, the investor is fully long in an uptrend and fully short in a downtrend. More recent approaches enrich the signal by introducing more regimes, but they still place strong restrictions on the portfolio weights. In other words, literature has spent substantial effort refining how to detect regimes, while paying much less attention to how to position optimally once those regimes are detected.

This paper separates these two decisions. It takes the detected market regime as given and derives the Sharpe-optimal portfolio weight for each regime from first principles. The result is a simple and tractable framework for optimal regime-dependent allocation that can be applied to any finite number of regimes. Standard trend-following rules then appear as special cases of the framework—and, in general, as suboptimal ones.

This shift in perspective leads to a clear practical message: better trend following does not necessarily come from inventing yet another signal. It can come from allocating more intelligently across the regimes we already know how to detect.

The empirical results strongly support this idea. Across the U.S. equity market, U.S. style portfolios, international equity indices, and diversified portfolio implementations, the optimal regime-dependent strategy delivers consistently higher out-of-sample Sharpe ratios than both standard time-series momentum and dynamic-speed momentum. One especially important finding is that the full short exposure commonly used in bear regimes is often far from optimal. In many cases, the optimal bear-market exposure is close to zero and sometimes even mildly positive. That is a striking result, because it challenges one of the most common assumptions embedded in trend-following strategies.

The table below illustrates the practical importance of the paper’s main idea. Across 18 diversified portfolio datasets, the optimal regime-dependent strategy beats the benchmark in every single case. In the two-regime setting—bull and bear markets—the strategy is compared with the Time-Series Momentum rule of Moskowitz, Ooi, and Pedersen (2012), and the average out-of-sample Sharpe ratio rises from 0.208 to 0.506. In the four-regime setting—bull, bear, correction, and rebound—it is compared with the Dynamic Speed Momentum strategy of Goulding, Harvey, and Mazzoleni (2023), and the average Sharpe ratio increases from 0.496 to 0.628. The implication is both simple and powerful: better trend following does not always require a better signal. It can also come from better positioning within the regimes that current signals already detect.

Out-of-sample performance of diversified trend-following portfolio strategies

Notes: This table reports annualized out-of-sample Sharpe ratios for diversified Time-Series Momentum (TSM), Dynamic Speed Momentum (DSM), and regime-optimal (OPT) strategies implemented across 18 portfolio datasets from Kenneth French's data library. All strategies are estimated using a training period from July 1963 to December 1997 and evaluated out of sample from January 1998 to December 2025. For the two-regime specification, DIF denotes the difference between the Sharpe ratios of the OPT and TSM strategies. For the four-regime specification, DIF denotes the difference between the Sharpe ratios of the OPT and DSM strategies.

For investment professionals, the usefulness of the paper is straightforward. It provides a transparent portfolio-design framework for translating regime signals into economically justified position sizes. Instead of relying on arbitrary rules such as “+1 in bull markets, -1 in bear markets,” investors can estimate regime-specific expected returns and risks and map them directly into optimal exposures. This has immediate relevance for tactical asset allocation, managed futures, CTA-style investing, and any systematic strategy that conditions exposure on market states.

References

Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). “Time Series Momentum.” Journal of Financial Economics, 104(2), 228-250.

Goulding, C. L., Harvey, C. R., & Mazzoleni, M. G. (2023). “Momentum Turning Points.” Journal of Financial Economics, 149(3), 378-406.