Factor Cyclicality
Every documented factor has spent long periods underperforming. The horizon required to evaluate one is longer than most people's patience, which is part of why the premiums may persist.
MadStockAlerts Research · Updated August 29, 2026
What to take away
- Factor premiums are long-run averages, not reliable annual returns.
- Every factor has had drawdowns lasting years and sometimes more than a decade.
- Factors are imperfectly correlated, which is the argument for combining them.
- Timing factors has proved as difficult as timing markets.
- The difficulty of holding through a drawdown is a candidate explanation for the premium.
MAD Academy Training Video · 0:46
Every Factor Has a Bad Decade
Factors take turns working, and the periods of underperformance are long enough that almost nobody sits through them.
This lesson is part of a Stock Alerts + Tools plan.
How long the bad periods last
| Factor | Notable drawdown |
|---|---|
| Value | More than a decade of underperformance from around 2007 |
| Momentum | Sharp crashes at turning points, including 1932, 2009 and 2020 |
| Size | Extended underperformance across multiple decades |
| Low volatility | Lags substantially in strong rising markets |
| Quality | Underperforms in sharp recoveries led by weaker companies |
The pattern is general rather than particular to any one factor. An effect that produced a premium over ninety years produced it in a subset of those years, and the periods in between were long enough to make abandonment the norm.
The evaluation problem
The same sample size problem that runs through this library applies with force here. A premium of a few percent a year against annual volatility of fifteen percent requires decades of data before it can be distinguished from zero with any confidence.
That means a factor cannot be evaluated over the horizon on which any individual is likely to judge it. Three years of underperformance is entirely consistent with a real premium, and it is also entirely consistent with the premium having disappeared.
This is the practical form of the arbitrage limit argument. If the premium requires holding through a decade of underperformance, very few participants will, and that is a plausible reason the premium is not competed away.
Why combining them is the standard answer
- Factors are imperfectly correlated with each other, so their bad periods do not coincide fully.
- Value and momentum in particular have been negatively correlated over long periods, which is unusual and useful.
- A combination reduces the variability of the combined premium without proportionally reducing it.
- It also reduces the chance of any single factor's drawdown ending the investor's patience.
The second item is the most cited relationship in factor construction. Value buys what has fallen and momentum buys what has risen, so the two are structurally opposed and their combination has historically been smoother than either.
Scroll the chart sideways to see all of it.
- Value
- Momentum
- Equal combination
Timing them
The natural response to cyclicality is to hold a factor when conditions favour it. Attempts to do so have generally not succeeded, and the reasons mirror those for market timing.
- Valuation spreads within a factor have some predictive power and not enough to time on.
- Macro conditioning variables work in sample and generalise poorly.
- The turning points are identifiable afterwards and not at the time.
- Timing adds turnover, which adds cost, which the strategy has to overcome.
The conventional conclusion in the literature is that a diversified, patiently held combination has been more reliable than any attempt to rotate between them. That is a description of what the research has found rather than a recommendation.
Factor crowding
A factor that has performed well attracts capital, and capital changes the factor. This is the mechanism by which a documented effect becomes a crowded position.
- 1The factor performsOver a period long enough to be noticed.
- 2Products are launchedAnd allocations are made to them.
- 3Valuations within the factor riseThe securities it selects become more expensive relative to the rest.
- 4The forward premium fallsBecause more was paid for the same expected cash flows.
- 5A reversal is amplifiedBecause the same holders exit together.
The fifth step is the risk that crowding creates and that the factor's history does not contain. A factor documented on decades when few people traded it does not tell you how it behaves when many do.
Valuation spreads within a factor are the usual measure of this, and they have some predictive power over long horizons and not enough to time on, which is the same conclusion the timing section reaches.