What it is
Dual momentum combines two rules. Relative momentum picks the asset that has performed best over a lookback window (classically 12 months) from a small menu, such as US stocks, international stocks and bonds. Absolute momentum then checks whether that asset's return beat the return on cash over the same window; if not, the strategy holds bonds instead. One decision per month, one position at a time. It is a simple, widely copied framework, and it should be evaluated with the same skepticism as any other backtested idea.
The logic
Relative momentum exploits the persistence of asset-class leadership over 6 to 12-month horizons, the same effect as relative-strength-rotation. Absolute momentum exploits the persistence of bear markets: an asset whose 12-month return is below cash is more likely to keep falling than an asset above it. Together they aim to be in the strongest thing when something is strong and out of everything when nothing is.
The other side of this trade is the diversified buy-and-hold investor who holds all the assets all the time. Over most long histories dual momentum has shown lower drawdowns and comparable or better returns, but with long stretches of tracking error where it holds the wrong asset for a year. The counterparty is not losing so much as being smoother in different periods.
Setup rules
- Market: a small menu of liquid, low-cost index funds: a domestic equity fund, an international equity fund, an aggregate bond fund, and a cash proxy (short-term treasuries). Adding more assets makes the strategy noisier, not better, and increases the chance of overfitting.
- Timeframe: monthly data; one decision on the last trading day of the month.
- Lookback: 12-month total return. Alternative windows (6 or 9 months, or an average of several) exist and produce similar but not identical results.
- Decision: compare the equity funds' 12-month returns; take the higher. If that return exceeds the cash proxy's 12-month return, hold that equity fund for the next month. Otherwise hold the bond fund.
- Execution: trade at the next open after the signal; a one-day lag makes little difference and keeps the test honest.
Entry, stop, target
No chart entries and no stops; the monthly rule is the entry and the exit. The practical risk of any single holding is one month of that asset's return before the rule can react, which in an equity crash can be 10 to 20 percent.
| Item | Value | Notes |
|---|---|---|
| Menu | 2 equity funds, 1 bond fund, cash | Keep it small |
| Lookback | 12 months | Total return including dividends |
| Decision frequency | Monthly | 12 decisions per year |
| Trades per year | Typically 1 to 4 | Low turnover |
| Maximum exposure per position | 100 percent | Concentration is by design |
| Reaction lag | Up to one month | A crash inside a month is fully felt |
Position sizing and risk
Because the strategy is 100 percent in one asset at a time, its risk is the risk of that asset for one month. That is appropriate for a portion of a portfolio, not necessarily for all of it. Investors who want lower volatility hold dual momentum alongside a static allocation or run it on a fraction of capital. The framework at /learn/risk-management explains how concentration and time-to-react combine, and /tools/position-size can be used to size a volatility-scaled version. Do not use leverage or leveraged funds inside the rule; the monthly reaction lag makes leverage dangerous.
What breaks it
- Whipsaws around the cash test. Sideways equity markets where 12-month returns hover near the cash return produce several switches with small losses.
- Fast crashes and V recoveries. The rule exits after the decline and re-enters after the recovery; a one-month crash and rebound is the worst case.
- Concentration. Holding one asset means one country's bad year is fully absorbed.
- Edge decay. The published results rely heavily on a few bear markets; the number of independent events in even a 50-year backtest is small, and the effect has shown weaker results in the decade after it was popularised. That does not prove it stopped working, but it should temper expectations.
- Taxes and costs are low relative to most strategies, which is one of its genuine advantages.
- Behavioural failure. Holding bonds while stocks rally for a year, or holding international while domestic outperforms, is where investors abandon it.
How to test it
Assemble 40 or more years of monthly total-return data for each asset (index data before the funds existed). Run the rule with several lookbacks (6, 9, 12 months, and a blend) and report compound return, max-drawdown, worst 12-month period, number of switches and months in bonds. Compare against a static 60/40 portfolio and against each asset alone. Then run the test on a different menu (for example, sector funds, or another country's assets); a real effect survives menu changes. Treat any result as tentative: forty years of monthly data is 480 decisions but only a handful of regimes. See walk-forward-testing and expectancy-system-evaluation.
Variations
- Multi-lookback blend: average the 3, 6 and 12-month returns to reduce the sensitivity to a single window.
- Sector dual momentum: the same rules on a menu of sector funds; higher turnover and noise.
- Fractional version: hold the top two of a larger menu rather than the top one, which reduces concentration.
Further reading
max-drawdown, etf, index, diversification, correlation, sharpe-ratio, risk-on-risk-off, backtesting, survivorship-bias, recency-bias.
Related playbooks: trend-following-200-day, relative-strength-rotation, systematic-momentum-rules, rebalancing-bands