The investment industry spends a great deal of time and resources searching for alpha. Asset managers often speak of “harvesting alpha” — the idea that excess returns can be identified, captured, and compounded through time.
In practice, alpha is more fragile than that language suggests.
Not all excess returns are created equal, and not all persist in the same way. Broadly speaking, most sources of alpha fall into a few distinct categories — each with different strengths, limitations, and failure modes.
Manager alpha exists, but identifying it in real time is difficult. Distinguishing skill from luck ex ante remains one of the hardest problems in finance, particularly over short horizons. Persistence is elusive, and confidence often arrives only after the opportunity has passed.
Systematic alpha, including factor-based strategies, represents a more durable and repeatable source of excess return. These approaches are grounded in long-standing economic and behavioral relationships rather than individual discretion. While factors have demonstrated persistence over long horizons, their performance is not constant through time. Their effectiveness depends critically on the environment in which they operate.
Tactical alpha seeks to exploit shorter-term signals, trends, or market inefficiencies. When well designed, such strategies can be robust. Over time, though, tactical approaches often face decay as signals become crowded and excess returns are gradually competed away. Markets adapt.
There is, however, a fourth and more structural source of excess return that is often overlooked.
We call it regime alpha.
Regimes are not forecasts. They are not narratives. They are simply the recognition that markets operate in a small number of persistent risk states—states in which the behavior of returns, volatility, and correlations changes in meaningful ways.
When risk behaves differently across these states, and when those states persist long enough to matter, portfolio alignment becomes more important than prediction. Excess return emerges not from being clever, but from being correctly positioned for the environment that actually exists.
The challenge is that many regime frameworks fail before they start.
Regimes defined after the fact are often interesting and explanatory, but they are rarely useful for decision-making. Once labels such as “recession,” “inflation,” or “tightening” are imposed, interpretation enters the model. Bias follows quickly behind, which begs the question:
What if we allow the data to identify regimes, without imposing labels or narratives in arrears?
When statistically significant and economically meaningful regimes are extracted directly from the data — using a disciplined, walk-forward process — we preserve their original structure. In doing so, we create a regime framework that is descriptive without being retrospective, and systematic without being fragile. What matters most in this framework is not the level of returns, but the shape of outcomes.
Across asset classes and factors, regimes may not materially change average monthly returns. They do, though, reliably change dispersion, asymmetry, and tail behavior. Durable environments tend to compress outcomes and reward consistency. Fragile environments widen distributions, amplify downside risk, and subtly erode the assumptions embedded in single-state portfolios.
This is why portfolios can feel “fine” as risk is increasing.
Equity returns may remain plausible even as downside asymmetry grows. Factors can appear to “work” even as dispersion widens and persistence deteriorates. Bonds may hedge risk in some fragile environments, while offering little protection in others.
These are not anomalies—they are the predictable fingerprints of regime shifts. Risk changes form before it changes level.
The implication is straightforward: when regimes reshape the structure of risk, portfolio construction must adapt. Static averages are blunt instruments in a world where risk is clustered, conditional, and state-dependent.
To be clear, regimes do not need to be “timed” to add value. They are powerful even when treated as structural information rather than as trading signals. Simply recognizing that different behaviors dominate in different environments — and designing portfolios that align with those realities — creates a more resilient foundation than any single-state, averaged solution.
In the next post, we’ll explore what this means for portfolio design — and why recognizing regimes naturally leads to a different way of structuring equity portfolios.
About CrestCast™