Dual Momentum Explained: GEM, ADM, and Beyond
Dual momentum is one of the most robust and well-documented concepts in systematic investing. The core idea is deceptively simple: combine two distinct types of momentum — relative and absolute — to determine both what to own and whether to own it at all. The result is a strategy that participates in strong uptrends while stepping aside during prolonged drawdowns.
This article examines the two most important implementations of dual momentum: Gary Antonacci's Global Equities Momentum (GEM) and the Accelerating Dual Momentum (ADM) variant that uses multi-period composite scoring. Both share the dual momentum framework but differ meaningfully in their signal construction, asset universes, and defensive behavior.
The Dual Momentum Framework
Relative Momentum
Relative momentum — also called cross-sectional momentum — ranks a set of assets by their recent performance and selects the strongest. The logic is straightforward: assets that have outperformed their peers tend to continue outperforming over intermediate horizons (typically 3 to 12 months). This persistence effect has been documented across nearly every asset class and geography since Jegadeesh and Titman's seminal 1993 paper.
In practical terms, relative momentum answers the question: which asset should I hold? It does not answer whether you should hold any risky asset at all.
Absolute Momentum
Absolute momentum — also called time-series momentum — compares an asset's return to a threshold, typically the risk-free rate or zero. If an asset's trailing return exceeds that threshold, it is said to have positive absolute momentum. If not, conditions are unfavorable and the asset should be avoided.
Absolute momentum answers the second question: should I be invested at all, or should I move to safety?
Why Combine Both
Used alone, relative momentum keeps you invested at all times — even during broad market declines when every option is losing money. Absolute momentum, on the other hand, can move you to safety but does not help you choose among risky assets when conditions are favorable. Dual momentum combines both: use relative momentum to pick the best asset, then apply an absolute momentum filter to confirm that conditions warrant holding any risk at all.
Global Equities Momentum (GEM)
Strategy Mechanics
GEM, developed by Gary Antonacci and published in his 2014 book "Dual Momentum Investing," uses a deliberately small universe to minimize complexity and maximize clarity of signal.
| Parameter | Value |
|---|---|
| Risk-On Assets | SPY (US large cap), VEU (all-world ex-US) |
| Risk-Off Asset | AGG (aggregate bonds) |
| Momentum Method | REL_STR_1P (single-period relative strength = 12-month return) |
| Lookback | 12 months |
| Top-N | 1 |
| Protection | ABSOLUTE (vs 1-year T-bill rate, DGS1) |
How GEM Works Step by Step
Each month, GEM executes a two-stage decision process:
- Relative momentum test: Compare the trailing 12-month returns of SPY and VEU. The asset with the higher return wins the relative race and becomes the candidate holding.
- Absolute momentum test: Has the US stock market (SPY) produced a 12-month return that exceeds the current 1-year Treasury bill rate? If yes, hold the relative winner at 100%. If no, the broad equity market lacks positive absolute momentum — move entirely to AGG.
A critical detail: the absolute momentum test uses SPY specifically, not the relative winner. This means that even if VEU wins the relative race, the absolute test still checks whether the US equity market is above the T-bill threshold. The rationale is that US equities serve as the best single indicator of global risk appetite — when SPY is underperforming T-bills, conditions are hostile for equities globally.
Why VEU, Not EFA
GEM uses VEU (FTSE All-World ex-US) rather than EFA (MSCI EAFE). The distinction matters: VEU includes emerging markets alongside developed international markets, providing broader global exposure. EFA covers only developed markets (Europe, Australasia, Far East), missing the growth dynamics of emerging economies. By using VEU, GEM captures the full non-US equity opportunity set in a single ETF.
GEM's Strengths and Limitations
GEM's greatest strength is its simplicity. With only three assets and a clear two-step decision rule, it is nearly impossible to over-optimize. The strategy makes one trade per month at most, and in many months makes no trade at all. This makes it practical for taxable accounts and easy to follow consistently.
The primary limitation is the single 12-month lookback. A pure R12 signal is slow to react to rapidly changing conditions. During the 2020 COVID crash, GEM's 12-month lookback meant it was still reflecting the strong 2019 returns when the market collapsed in March. The strategy eventually moved defensive, but the delay cost roughly 15-20% of drawdown that faster signals would have avoided.
Accelerating Dual Momentum (ADM)
The Multi-Period Innovation
ADM addresses GEM's speed limitation by replacing the single 12-month lookback with a composite score that blends multiple timeframes. Instead of relying solely on R12, ADM computes the average of 1-month, 3-month, and 6-month returns.
| Parameter | Value |
|---|---|
| Risk-On Assets | SPY (US large cap), SCZ (international small cap) |
| Risk-Off Asset | TLT (long-term Treasuries) |
| Momentum Method | ADM = avg(R1M, R3M, R6M) |
| Lookback | 6 months (longest component) |
| Top-N | 1 |
| Protection | ABSOLUTE (score must be > 0) |
How ADM Works
Each month, ADM scores SPY and SCZ using the composite formula: avg(R1M, R3M, R6M). The asset with the higher composite score wins the relative test — it is the stronger momentum candidate. Then the absolute test: if the winner's composite score is positive (above zero), hold it at 100%. If the winner's score is negative, momentum has turned unfavorable across all timeframes, and the strategy moves entirely to TLT.
The multi-period composite is the key innovation. By blending short, medium, and intermediate lookbacks, ADM detects trend changes faster than GEM's pure R12 signal. The 1-month component reacts quickly to sudden shifts; the 3-month component confirms the trend; the 6-month component provides stability. All three must align for a strongly positive score, creating a natural consensus mechanism.
Why SPY vs SCZ
ADM pairs US large caps (SPY) against international small caps (SCZ). This is a deliberate choice: SCZ has low correlation with SPY and tends to lead during global growth cycles when capital flows to smaller international companies. The wide behavioral gap between these two assets means the relative momentum signal is more meaningful — when one is winning, it is typically winning decisively.
GEM vs ADM: Key Differences
| Dimension | GEM | ADM |
|---|---|---|
| Risk-On Universe | SPY, VEU | SPY, SCZ |
| Risk-Off | AGG | TLT |
| Scoring | R12 (single period) | avg(R1, R3, R6) |
| Absolute Test | SPY R12 vs T-bill rate | Winner score > 0 |
| Signal Speed | Slower (12-month lag) | Faster (1-month component) |
| Defensive Posture | Aggregate bonds | Long-term Treasuries |
The defensive asset choice also differs meaningfully. GEM uses AGG (aggregate bonds), which provides moderate duration and credit diversification. ADM uses TLT (20+ year Treasuries), which has much higher duration and therefore benefits more from flight-to-quality rallies during equity crashes. TLT can gain 20-30% during severe recessions, providing a powerful offset to equity losses — but it also carries more interest rate risk during rising-rate environments like 2022.
The Broader Dual Momentum Family
GEM and ADM represent two points on a spectrum of dual momentum implementations. The framework is flexible enough to accommodate many variations, and several are worth understanding.
Variations in Momentum Scoring
The 13612W formula — a weighted average of 1, 3, 6, and 12-month returns with heavier weight on recent periods — is used by strategies like DAA to produce a momentum score that balances responsiveness with stability. Other approaches use SMA ratios (price divided by a moving average) as a trend-following proxy that behaves differently from pure return-based momentum.
Expanding the Universe
While GEM and ADM use only two risk-on assets, other dual momentum implementations expand the universe significantly. BAA applies momentum ranking across 12 offensive assets and 7 defensive assets, selecting the top 6 and top 3 respectively. The principle remains the same — rank by momentum, filter by absolute conditions — but the larger universe provides more diversification and more opportunities for the momentum signal to add value.
Canary-Based Protection
Some strategies replace the absolute momentum filter with a canary system. Instead of checking whether the winner's own momentum is positive, they monitor separate indicator assets whose behavior predicts broad market stress. DAA uses VWO and BND as canary assets — when these begin to deteriorate, the strategy moves defensive regardless of what the offensive assets are doing. This approach can detect trouble earlier than self-referential absolute momentum tests.
Practical Considerations
Tax Efficiency
Both GEM and ADM are inherently tax-efficient for tactical strategies. With only two risk-on assets and a single defensive position, turnover is low — typically 4-8 trades per year. Holdings that persist for more than 12 months qualify for long-term capital gains treatment. The concentrated nature of these strategies (always holding a single position at 100%) means there is no rebalancing drag from maintaining multiple positions.
Behavioral Challenges
Dual momentum strategies demand conviction during their weakest periods. When the strategy is in defensive mode — holding AGG or TLT while equities rally — the tracking error against a buy-and-hold benchmark can be substantial. During the 2019 bull run, for example, periods where the absolute filter was on could mean missing several months of strong equity gains. The investor must trust the process and understand that the protective mechanism will eventually justify itself during the next downturn.
Implementation Discipline
The simplicity of dual momentum is both its greatest asset and its greatest risk. Because the rules are so clear and the trades so infrequent, it is tempting to second-guess the signal. "The market just had a bad month — surely it will bounce back." This kind of discretionary override defeats the purpose of a systematic approach. The strength of dual momentum lies precisely in its mechanical discipline: follow the signal every month, without exception.