Dual Momentum Strategy: Relative Plus Absolute Strength

The two tests

Relative momentum asks which of several assets has performed best recently, and holds that one. Absolute momentum, also called time-series momentum, asks whether the chosen asset has a positive return over the same lookback, typically compared against treasury bills. Dual momentum applies both.

Each month end:

1. Compute 12-month total return for:
      US equities,  International equities,  T-bills

2. Relative test:
      Which of US or International is higher?

3. Absolute test:
      Is that winner's 12-month return above the T-bill return?

4. Allocate:
      If yes  ->  100% into the winning equity index
      If no   ->  100% into aggregate bonds or T-bills

Hold until the next month end.  Typically 2 to 5 changes per year.

The elegance is that the absolute test is what removes exposure during sustained declines. Relative momentum alone would simply hold whichever equity market was falling more slowly, which in 2008 would still have meant a very large loss.

A complete implementation

Universe

A minimum of: a domestic equity ETF, an international equity ETF, an aggregate bond ETF, and a cash equivalent. Wider versions add small caps, real estate, gold, and emerging markets.

Lookback

12-month total return is the standard. Blending 6 and 12 months reduces sensitivity to a single month.

Relative test

Rank all risk assets by lookback return; select the top one, or the top two for a less concentrated version.

Absolute test

The selected asset must have a lookback return above the return on treasury bills over the same period.

Defensive allocation

If the absolute test fails, hold short-term treasuries or an aggregate bond fund. Which of these you choose materially affects results in rising-rate periods.

Rebalance

Monthly using month-end closes, executed on the first trading day of the new month.

Confirmation buffer

Optional: require two consecutive monthly signals before switching. Reduces whipsaw at the cost of extra lag.

Position sizing

The strategy is all-in by construction. For lower volatility, run it on a portion of capital alongside a static allocation.

Whipsaw: the cost of the defence

The absolute filter provides protection in sustained declines and costs money in choppy ones. A market that falls 12 percent, triggers an exit, then recovers within two months produces a realised loss plus a missed recovery. Sequences of two or three such events in a year are common and are the primary reason investors abandon the strategy.

None of these removes whipsaw, and any modification that removes it entirely has almost certainly also removed the protection. Accepting a defined number of false exits per year is part of the contract.

What to expect, and what the evidence supports

Published backtests of dual momentum over long histories generally show equity-like returns with substantially smaller maximum drawdowns, because the strategy exits during extended bear markets. Those results are robust in direction but sensitive in magnitude to the specific universe, the lookback, the defensive asset, and the exact rebalancing date.

• Date sensitivity. Rebalancing on the last day of the month versus mid-month can change results meaningfully over a long sample. Test several dates; if the result depends on one, it is fragile.
• Defensive asset choice. Long-duration bonds performed exceptionally during the multi-decade decline in interest rates, which flatters backtests that used them. Short-term treasuries are the more conservative assumption.
• Limited independent observations. With 2 to 5 switches per year, a 50-year backtest contains perhaps 150 decisions, which is a small sample for a strategy with this much variation.
• Underperformance in strong bull markets. The strategy will lag a simple buy-and-hold approach during long uptrends, sometimes for years, because of switching costs and lag.