Factor Decay

The factor decay problem describes a striking pattern: after academics publish research documenting a return anomaly, the anomaly often weakens or disappears as capital flows into the strategy, arbitrage tightens pricing, and the signal’s half-life compresses. Publication, paradoxically, reduces future returns. This dynamic raises a fundamental question for factor investing: can a factor premium survive its own discovery?

The observation and its pattern

The factor decay story begins with a simple observation. Researchers identify a strategy—say, buying high-quality accruals stocks—and test it on historical data, finding a 5% annualized excess return. They publish. Practitioners read the paper, build funds around it, and deploy capital. Ten years later, the realized return has fallen to 1–2% per year. Did the researchers cherry-pick data? Or does discovery itself kill the profit?

Empirical evidence supports the latter. A meta-analysis across dozens of published anomalies reveals a consistent pattern: returns post-publication are typically 20–50% lower than returns pre-publication. Some factors persist at diminished levels; others fade to zero. Accruals anomaly, momentum, and value factors have all shown measurable decay since their academic codification.

The decay is not universal or immediate. Factors driven by deep transaction costs (e.g., liquidity factor) or structural market frictions (e.g., carry factor) show slower decay than behaviorally-driven anomalies. But decay is visible in most cases, starting within 3–7 years of publication.

Why capital inflows compress premiums

The mechanism is mechanical. When a factor premium exists, it arises because mispricing persists. An underexploited mispricing is by definition exploitable—there is room to buy cheap and sell dear without moving the price too much. But as money flows into the arbitrage, prices shift.

Consider accruals anomaly concretely. Before publication, a handful of practitioners are aware that low-accrual stocks outperform. They buy them, but not so much that prices fully correct. After publication, thousands of managers add low-accrual screens to their portfolios. The price of low-accrual stocks rises, the price of high-accrual stocks falls, and the return differential compresses. The factor still exists—low-accrual stocks still have more sustainable earnings—but the mispricing that drove the premium is gone.

The speed of this adjustment depends on factor capacity: how much capital can be deployed before prices move substantially. Small factors with narrow applicability (e.g., microstock illiquidity premium) have low capacity and may show rapid decay. Broad factors (e.g., value) have high capacity and decay more slowly.

Publication bias in factor research

A subtle wrinkle complicates the picture: publication bias. Researchers submit only successful strategies to academic journals. A strategy that works on historical data but yields mediocre real-world returns is less likely to be published. This selection bias inflates reported premiums.

When practitioners then implement the published strategies, they are implementing strategies that have already been culled for out-of-sample noise. Real-world returns cannot be as high as in-sample returns by definition. Add in transaction costs, management fees, and slippage, and the decay relative to published numbers is guaranteed.

However, publication bias alone does not fully explain the decay. Even after adjusting for reasonable levels of multiple testing and overfitting, significant decay remains.

The half-life of a factor

Researchers have attempted to quantify decay as a “half-life”—the time in years after publication at which a factor’s return premium falls to half its historical level. Estimates vary, but a consensus range is 5–15 years, with wide variation.

Factors with strong theoretical or structural underpinnings decay more slowly. Carry and liquidity premiums persist because they reflect genuine economic frictions—borrowing costs, transaction costs, and agency risks that cannot be arbitraged away entirely. Behavioural factors like momentum or specific valuation anomalies decay faster because the cognitive biases that drive them can be addressed through awareness and discipline.

Implications for factor investors

For practitioners, factor decay creates several challenges and opportunities.

The live-factor trap: A successful historical factor (high Sharpe ratio, strong information coefficient) becomes a crowded live factor (low returns, high correlation to similar strategies). The investor who believes a factor should work continues to hold even as returns evaporate. Discipline to exit or rotate is difficult.

Factor turnover and discovery: Some managers attempt to stay ahead of decay by continuously hunting for new, unpublished factors. The idea is to identify anomalies before they are arbitraged or published. This is the “factor alpha” game—finding the next accruals anomaly before it is commoditised. But it is a hard game, and returns to factor discovery are uncertain.

Multi-factor diversification: Rather than betting on a single factor with uncertain future returns, investors combine multiple factors—value, momentum, carry, quality, and liquidity—and rebalance regularly. This approach spreads the decay risk and captures newer, smaller factors alongside the decaying incumbents. The realized return is lower than a single pre-decay factor would suggest, but more stable and more reliable.

Timing and capacity awareness: Practitioners with deep operational capacity may enter factor trades before publication, exit near peak crowding, and redeploy to emerging anomalies. This requires speed, agility, and access to unpublished research—advantages reserved for well-funded firms.

Decay versus true structural change

Not all return decay is factor decay in the classical sense. Structural market changes—increased index inclusion, regulatory shifts, technology that reduces trading frictions—can erode premiums independently of publication. For example, the size factor decay of the 1990s–2000s partly reflected the rise of passive index funds and improved equity market technology that benefited smaller stocks.

Distinguishing pure decay (driven by discovery) from structural changes is important but difficult. A factor’s decline may reflect both.

The evolution of factor investing in the post-decay era

Awareness of factor decay has matured the factor investing industry. Modern factor-based ETFs and managers do not promise historical factor returns. Instead, they frame factors as persistent sources of return with time-varying premia, complementary to each other, and subject to market and economic cycles.

The best-run factor investing programs combine:

1. Multiple uncorrelated factors to diversify decay risk
2. Regular rebalancing to avoid overweighting decaying factors
3. Monitoring of forward-looking indicators (valuation, crowding, capacity) rather than relying on historical returns
4. Openness to alpha research and emerging anomalies

See also