Composite Dual Momentum (CDM)
Developed by Gary Antonacci · Dual Momentum · Med Risk
Composite Dual Momentum was developed by Gary Antonacci as an evolution of his foundational Global Equity Momentum strategy. Published on his Optimal Momentum research site, CDM addressed the primary criticism of GEM: that concentrating the entire portfolio in a single asset creates unnecessary concentration risk that diversification could mitigate without sacrificing the core dual momentum signal.
Antonacci's insight was that the dual momentum framework — relative strength for selection, absolute momentum for protection — could be applied independently across multiple asset class pairs rather than just one. By dividing the portfolio into four self-contained modules, each running its own dual momentum process, the strategy maintains the decisive binary switching within each pair while achieving portfolio-level diversification across fundamentally different market segments. This modular architecture represents the natural evolution of dual momentum from a single-market concept to a complete portfolio construction system.
The motivation for CDM emerged from a practical observation about GEM's behavior during choppy markets. When GEM's single momentum signal produces a timing error — entering equities a month too early or exiting a month too late — the entire portfolio absorbs that error. With CDM's four independent modules, a timing error in the equity module affects only twenty-five percent of the portfolio while the other three modules continue operating on their own, potentially correct, signals. This error diversification is the primary source of CDM's improved risk-adjusted returns relative to GEM.
Each module pairs two assets from the same broad category that tend to exhibit alternating leadership cycles: US versus international equities, investment-grade versus high-yield bonds, equity REITs versus mortgage REITs, and gold versus long-term Treasuries. These pairings are deliberate — within each module, the two assets share enough economic exposure to make a relative comparison meaningful, but differ enough in their sensitivity to specific macro factors that one typically outperforms the other at any given time.
How It Works
The Four Paired Modules
The portfolio is divided into four independent modules, each receiving twenty-five percent of capital. The equity module compares US stocks (SPY) against international stocks (EFA), capturing the multi-year US-versus-international rotation cycle driven by relative monetary policy, currency strength, and earnings growth. The credit module matches investment-grade bonds (LQD) against high-yield bonds (HYG), reflecting the market's risk appetite in fixed income — high yield outperforms during credit expansions while investment-grade leads during tightening. The real estate module pits US REITs (VNQ) against mortgage REITs (REM), distinguishing equity-like property exposure from interest-rate-sensitive mortgage instruments. The commodity-duration module sets gold (GLD) against long-term Treasuries (TLT), capturing the inflation-deflation dynamic that drives relative performance between these traditional safe havens.
Relative Momentum Within Each Module
Each month, the two assets in every module are compared by their trailing twelve-month total return. The asset with higher momentum wins the relative comparison and becomes the candidate for that module's allocation. This is the same relative momentum logic used in GEM, but applied four times independently rather than once.
The paired structure ensures that each module always has a meaningful comparison to make. Unlike strategies that rank assets from a large universe — where the spread between top and bottom can be enormous or negligible depending on market conditions — the paired approach guarantees a direct one-on-one contest between closely related alternatives. This tends to produce more stable and interpretable signals than broad-universe ranking.
Trend Health Filter
After selecting each module's relative winner, a trend health check is applied. By default, the winner's twelve-month return must exceed the three-month Treasury bill annualized rate. This is the absolute momentum gate — ensuring the selected asset is not merely less bad than its pair competitor, but genuinely in an uptrend relative to risk-free rates.
If the winner passes the health check, it receives that module's full twenty-five percent allocation. If it fails, the entire module shifts to short-term Treasuries (BIL). This per-module defense means the portfolio can be partially defensive — perhaps two modules in their winners and two in BIL — providing a graduated risk posture that GEM's binary all-or-nothing approach cannot achieve. During broad market weakness, all four modules may shift to BIL independently, producing full defensive positioning through the aggregation of four separate signals rather than a single binary decision.
Portfolio-Level Diversification
The four modules operate entirely independently — the equity module's decision has no influence on the commodity-duration module's decision. This independence means the portfolio always holds four positions across fundamentally different asset categories, providing structural diversification that persists regardless of market conditions.
When one module experiences a whipsaw or timing error, the other three continue operating on their own signals, dampening the impact at the portfolio level. This error diversification is CDM's primary advantage over GEM. The probability of all four modules simultaneously making the wrong call is substantially lower than the probability of GEM's single signal being wrong, producing smoother returns and smaller drawdowns over time.
Explore Composite Dual Momentum (CDM)
See the full backtest across 18 years of market data, or run your own what-if scenarios by adjusting all research parameters.