Backtesting Dual Momentum: Gary Antonaccis Strategy Put to the Test
# Backtesting Dual Momentum: Gary Antonaccis Strategy Put to the Test
Gary Antonaccis Dual Momentum strategy gained widespread popularity after his 2014 book demonstrated impressive backtested results. The approach combines relative momentum (which asset is strongest) with absolute momentum (is the trend positive at all). We replicated and extended his backtest to 2025.
The Original Strategy Rules
Each month, compare the trailing 12-month return of US stocks (SPY) versus international stocks (EFA). Invest in whichever has higher relative momentum. However, if the winner has negative absolute momentum (trailing 12-month return below T-bills), move to bonds (AGG) instead. Rebalance monthly.
Replication Results (1971-2025)
The full-period backtest produced 14.2% annualized returns with a maximum drawdown of 19.6% and a Sharpe ratio of 0.87. Buy-and-hold SPY over the same period returned 10.4% with 55% maximum drawdown and 0.52 Sharpe. The improvement in risk-adjusted returns is substantial and statistically significant.
Post-Publication Performance (2015-2025)
Since the strategy became widely known, performance has degraded somewhat. Returns from 2015-2025 were 9.1% annualized versus 11.8% for SPY. The strategy underperformed during the strong US-led bull market because international stocks rarely won the relative momentum comparison, and the few times they did proved to be false signals.
The US Dominance Problem
From 2010 to 2025, US stocks outperformed international stocks in 12 of 15 years. This made the relative momentum component essentially useless, as it almost always selected US stocks. The strategy occasionally whipsawed into international at exactly the wrong time, creating small but consistent drag.
Lookback Period Sensitivity
We tested 3, 6, 9, and 12-month lookback periods. The 12-month lookback (original specification) remained the best for avoiding major crashes. However, the 6-month lookback better captured intermediate trends and produced slightly higher returns post-2015. A blend averaging 6 and 12-month signals showed modest improvement.
Adding More Asset Classes
We expanded the universe to include US stocks, international developed, emerging markets, and REITs for the relative momentum comparison. This wider universe improved post-2015 performance to 10.8% annualized by occasionally capturing emerging market rallies that pure US/international comparison missed.
The Absolute Momentum Magic
Removing the absolute momentum filter (just using relative momentum alone) dramatically worsened drawdowns to 42%. The absolute momentum component, which moves to bonds during negative trends, is responsible for nearly all of the risk reduction. It caught the 2008, 2020, and 2022 drawdowns effectively.
Transaction Costs and Tax Drag
Monthly rebalancing with an average of 3-4 switches per year between assets creates minimal transaction costs in a commission-free environment. However, each switch is a taxable event. In taxable accounts, the after-tax returns dropped by approximately 1.5% annually due to short-term capital gains on positions held less than 12 months.
Combining with Other Signals
We tested adding a volatility filter (reduce allocation when VIX is above 25) and a breadth filter (require positive market breadth for equity allocation). The breadth filter improved the timeliness of crash protection by approximately 5 trading days without adding false signals, modestly improving risk-adjusted returns.
Final Assessment
Dual Momentum remains a valid and relatively robust approach to systematic asset allocation. The absolute momentum component is the true value-add, providing reliable crash protection. The relative momentum component has been less useful in the US-dominated post-2015 era but adds value over full cycles. For simplicity, robustness, and accessibility to non-professional investors, Dual Momentum continues to earn its reputation as one of the best simple quantitative strategies available.