Momentum: Cross-Sectional vs Time-Series

Momentum is the most documented anomaly in the academic factor literature. Two implementations dominate, and they are not interchangeable.

Cross-sectional momentum

Within a universe (e.g., S&P 500 stocks), rank by trailing 3–12 month returns. Long the top decile, short the bottom decile. Rebalance monthly.

• Edge driver: the persistence of relative outperformance across stocks.
• Holding period: 1–3 months.
• Failure mode: momentum crashes — sharp reversals in market-wide sell-offs (March 2009, April 2020) erase years of gains in weeks. The Daniel & Moskowitz paper documents this rigorously.

Time-series momentum

For each instrument independently, go long if trailing 12-month return > 0; short if < 0. No relative ranking — each market votes for itself.

• Edge driver: the persistence of absolute trends within an instrument.
• Holding period: weeks to months.
• Failure mode: range-bound markets generate whipsaws. Behavior closer to trend following than to factor momentum.

When to use which

• Equities, factor portfolio: cross-sectional. The universe is large, stable, and rebalancing-friendly.
• Multi-asset futures portfolio: time-series. Each market has its own regime; relative ranking is less meaningful when comparing rates to commodities.
• Single-asset systematic: time-series. Cross-sectional needs a universe.

Implementation discipline

• Skip the most recent 1 month. Both forms of momentum are documented to work better with a 1-month skip — buying based on month t−12 to t−1 returns rather than t−12 to t. Short-term reversal contaminates the signal otherwise.
• Vol-target each leg. Vol targeting prevents the highest-vol names from dominating exposure.
• Manage tax drag for cross-sectional. Monthly rebalancing in taxable accounts has serious after-tax cost. Often better suited to tax-deferred allocation.

Related

• Trend following
• Mean reversion
• Single strategy vs ensemble