Factor Timing and Cycles

Every factor goes through cycles. Value underperformed for a decade. Momentum crashed in 2020. Quality lagged in the 2020 to 2021 bull market. Small-caps had a brutal 3-year drawdown from 2018. Anyone who invested in factors and watched their portfolio underperform the simple index year after year has asked the question: is the factor broken? Should I switch?

This is the most important behavioural challenge in systematic factor investing. And the answer | while nuanced | is mostly: no, you should not time factors, and yes, the cycle will likely end.

Why factors cycle

Factor cycles have two distinct causes:

1. Valuation cycles (mean reversion)

Factors become expensive and cheap relative to their history. When everyone crowds into a factor | say, momentum in a prolonged bull market | the stocks in that factor portfolio get bid up above their fair value. Expected future returns compress. Eventually the factor "mean-reverts" | either through a crash or a long period of muted returns while the valuation premium is digested.

2. Economic regime cycles

Different economic environments favour different factors:

The case FOR factor timing

There's a reasonable argument for adjusting factor exposures based on valuation spreads. When value stocks are trading at historically wide discounts to growth stocks, expected returns to value are higher than normal. This is documented | factor valuation spreads have some predictive power for future factor returns over 3 to 5 year horizons.

Similarly, when momentum is at historically high valuations (momentum stocks trading at extreme premiums to their own history), the risk of a momentum crash is elevated. Some sophisticated factor investors reduce momentum exposure at these times.

The case AGAINST factor timing | and why it's stronger

Despite the theoretical appeal, the evidence for successfully timing factors in practice is weak. Here's why:

• Factor cycles are long and irregular | Value underperformed for 10 years in the US before the 2022 recovery. If you exited value after 2 years of underperformance, you missed the eventual payoff. The signal-to-noise ratio for timing is terrible.
• Valuation spreads are imprecise predictors | even if wide value spreads predict better value returns over 5 years, they're terrible at predicting returns over 1 year | which is when most investors would actually change their allocation.
• You add another layer of decision-making error | every active decision you add is another opportunity to make a behavioural mistake. The primary reason systematic investing works is that it removes these decisions. Factor timing adds them back in.
• The market impact of switching | switching from a momentum portfolio to a value portfolio generates transactions, costs, and tax events. The switching cost must be overcome by the timing benefit. In practice, it rarely is.

What to do instead: strategic factor allocation

Rather than tactical timing, the evidence supports strategic, long-horizon factor allocation with these principles:

• Diversify across factors from the start | a multi-factor portfolio from day one means you're never fully dependent on a single factor's cycle
• Set a rebalancing rule and stick to it | annual rebalancing back to target factor weights, regardless of recent performance, captures some of the mean reversion effect without active timing
• Use drawdown events to add, not subtract | when a factor you believe in has a significant drawdown (value down 20%, momentum down 30%), that's typically a better entry point, not an exit point
• Measure in 5-year windows | short-term underperformance should be expected and budgeted for. Evaluate factor performance only over full cycles (5+ years)

Glossary