Momentum Trading Strategy: Buying Strength With Rules

Two kinds of momentum, often confused

Combining both is known as dual momentum: rank the universe relatively, then require the selected asset to also have a positive absolute return, otherwise hold cash or bonds. This preserves the selection benefit while adding the bear-market protection that pure cross-sectional momentum lacks.

Why momentum persists

• Under-reaction to news. Analysts and investors revise estimates gradually, so good news is priced in over weeks rather than instantly.
• Disposition effect. Investors sell winners too early and hold losers too long, which slows the adjustment to new information in both directions.
• Flows follow performance. Capital chases recent returns, mechanically reinforcing the existing direction, particularly at month and quarter boundaries.
• Career and benchmark risk. Professionals are reluctant to hold assets that have underperformed, which sustains demand for recent leaders.
• Slow diffusion of information. Effects spread across related firms and supply chains over weeks, producing predictable follow-through.

The effect has been measured in equities since at least the 19th century, across dozens of countries, and in currencies, commodities, and bonds. That breadth is what distinguishes it from a curve-fitted pattern: the same behaviour appears in markets that share no participants and no structure except human decision-making.

A complete cross-sectional momentum strategy

Universe

A defined, liquid set: for example the S&P 500 constituents, or 12 to 20 sector and country ETFs. Must be defined point-in-time to avoid survivorship bias.

Ranking signal

Total return over the past 12 months, excluding the most recent month (often written as 12-1 momentum).

Selection

Hold the top decile, or the top 5 of 20 ETFs. Fewer holdings means higher return dispersion and higher risk.

Absolute filter

Only hold a selected asset if its 12-1 return is positive and its price is above its 200-day moving average. Otherwise hold cash or short-term treasuries.

Weighting

Equal weight, or inverse-volatility weight so that each holding contributes similar risk.

Rebalance

Monthly, on the first trading day. Quarterly reduces costs and turnover at a modest cost in responsiveness.

Exit

On rebalance, sell anything that has left the top selection or that fails the absolute filter. There is no intramonth stop in the classic version.

Position size

Equity divided by the number of holdings, scaled by the inverse of each asset’s recent volatility if using risk weighting.

Worked example: a monthly sector rotation

Universe: 11 US sector ETFs. Capital 60,000 USD. Rules: rank by 12-1 return, hold the top 3 equally weighted, require each to be above its 200-day moving average, otherwise hold short-term treasuries in that slot.

The mechanics are deliberately dull: one calculation per month, three orders, no intraday decisions. Turnover is typically 40 to 80 percent per year, which for liquid ETFs costs a fraction of a percent annually. That low cost relative to expected return is a large part of why this approach is practical for individuals.

The momentum crash, and how to blunt it

Momentum’s characteristic disaster occurs at market bottoms. During a severe decline, the momentum portfolio holds defensive assets and is short or absent from the most beaten-down ones. When the market turns violently, the previously worst assets rally hardest, and a long-short momentum portfolio can lose a large fraction of its value in weeks. Long-only versions suffer a milder version of the same effect: they lag badly in the first months of a recovery.

• Volatility scaling. Reduce gross exposure when the strategy’s own recent volatility is elevated. This is the most consistently documented mitigation.
• Market-state filter. Reduce or pause after severe market declines, when crash risk is highest, rather than filtering on momentum values themselves.
• Longer rebalancing. Quarterly rebalancing reduces the whipsaw of rotating into leadership just as it changes.
• Avoid concentrated long-short. The most severe historical crashes affected long-short implementations. Long-only with an absolute filter is considerably gentler.

Variations worth testing

Testing momentum without deceiving yourself

1. 1

   Use point-in-time universe membership

   Ranking today’s index members over the past decade embeds enormous survivorship bias. You must know which stocks were in the index at each rebalance date.

2. 2

   Include delisted and merged securities

   Momentum portfolios hold assets that occasionally collapse or get acquired. Omitting them removes real losses and real gains.

3. 3

   Model realistic rebalance execution

   Trading at the month-end close is competitive and crowded. Test execution on the next open and with a day’s delay to confirm the edge is not an artefact of the exact timing.

4. 4

   Charge real costs on turnover

   Compute annual turnover and multiply by a realistic round-trip cost. Momentum survives this comfortably in liquid ETFs and much less comfortably in small-cap stocks.

5. 5

   Examine the worst 12 months specifically

   Report performance around 2009, 2020, and any sharp reversal in your sample. The average obscures the behaviour that determines whether you can hold the strategy.