Accelerating Dual Momentum (ADM)

Jul 2008 – Sep 2026	Strategy	SPY
Annual return (CAGR)	11.5%	12.7%
Worst drawdown	-38.5%	-47.2%
Sharpe ratio	0.74	0.70
Volatility	16.6%	19.7%
Annual return since publication (Jun 2018)	10.4%	15.0%

Accelerating Dual Momentum (ADM) was published on the Engineered Portfolio blog in 2018 as a faster version of Gary Antonacci's Dual Momentum. Each month it holds a single fund: the S&P 500 or international small-cap stocks, whichever has the stronger momentum over the last one to six months, or long-term Treasuries when even the stronger one is losing ground.

The idea#

Antonacci's Global Equities Momentum looks back twelve months, so it is slow to leave a falling market and slow to return. The ADM authors scored each fund by adding its 1-, 3-, and 6-month returns. Because the most recent month counts in all three terms, the score reacts faster to a change in trend, which is the "accelerating" part.

The dual momentum structure stays. Relative momentum picks the stronger of the two stock funds. Absolute momentum asks whether that fund's score is above zero, and if not, the strategy moves to bonds. The authors also changed the assets: international small caps instead of broad non-US stocks, and long-term Treasuries instead of aggregate bonds as the safe asset.

How it works#

At the close of the last trading day of each month:

Score the S&P 500 (SPY) and international small-cap stocks (SCZ) by recent momentum.
If the higher of the two scores is positive, hold 100% of that fund.
If the higher score is negative, hold 100% long-term Treasuries (TLT).

The original score is the sum of each fund's 1-, 3-, and 6-month returns. Tactfolio's builder cannot add returns together, so this version measures momentum as a fund's price relative to its seven-month (147-session) moving average. That measure counts the latest month most and older months progressively less, in nearly the same proportions as the 1/3/6 sum, and it picks the same fund in most months. SPY and SCZ are compared with the same measure applied to the ratio of their prices.

The original used Vanguard mutual funds (VFINX, VINEX, and VUSTX); SPY, SCZ, and TLT are the ETF equivalents. The authors also suggested an optional rule, holding whichever of long Treasuries and TIPS had the higher one-month return. This version uses their baseline, long Treasuries only.

What the backtest shows#

The ETF test starts in mid-2008, because SCZ launched at the end of 2007. Since then ADM compounded a little over 11% a year, slightly less than the S&P 500, with a smaller worst drawdown and a slightly better risk-adjusted return. It trailed a 60/40 stock and bond mix on risk-adjusted return, because it always holds a single, volatile fund.

The start date caught its best moment. It held long Treasuries through the crash of late 2008 and gained about a third while the S&P 500 fell by a quarter. Its fast signal also cost it. In 2016 it lost 5% while the S&P 500 gained 12%: it moved to Treasuries after the market's early-year drop, returned only after the rebound, and switched between US and international stocks at poor moments.

Its worst year was 2022. Stocks and long Treasuries fell together, so the bond fallback lost money too, and ADM finished the year down 35%, nearly twice the S&P 500's loss. The drawdown from January to October 2022 reached about 38% and was not recovered until June 2025.

Since publication in mid-2018 it has compounded about 10% a year. Simulated history cannot take the test further back, because no reference history exists for SCZ.

When it struggles#

Stocks and bonds falling together. Long Treasuries are the only defense, and in a rate shock like 2022 they fall with stocks.
Choppy markets. A fast signal flips more often, and each false switch sells after a drop and buys after a rebound.
One fund at a time. The portfolio carries the full risk of whichever fund it holds.

Using it on Tactfolio#

The live strategy above runs these rules with the seven-month moving-average score in place of the 1/3/6-month sum, so its results can differ from the original rule in months when the two scores disagree. Copy it to try TIPS as a second defensive fund, intermediate Treasuries instead of long ones, or a different average length.

Year by year

Year	Strategy	SPY
2026*	9.8%	14.0%
2025	20.2%	17.7%
2024	19.7%	24.9%
2023	18.9%	26.2%
2022	-35.1%	-18.2%
2021	25.3%	28.7%
2020	20.9%	18.3%
2019	14.7%	31.2%
2018	6.3%	-4.6%
2017	23.4%	21.7%
2016	-5.0%	12.0%
2015	1.1%	1.2%
2014	14.1%	13.5%
2013	18.5%	32.3%
2012	14.5%	16.0%
2011	2.9%	1.9%
2010	7.7%	15.1%
2009	18.4%	26.4%
2008*	32.9%	-25.4%

* Partial year.

The rules as implemented

This is the exact tree Tactfolio runs, rebalanced monthly with signals and trades at the close. Open it to inspect or copy it.

Sources and caveats

The source scores each fund by the sum of its 1-, 3-, and 6-month returns; the builder cannot add returns together. This version scores momentum as the price relative to its 147-session (seven-month) moving average, which weights the latest month most and older months less in nearly the same proportions: on random daily returns the two scores correlate about 0.98, against 0.93 for 13612U and 0.95 for 13612W. At month ends from 1997 to 2026, on VFINX and VINEX, it picked the same holding as the exact sum in 93% of months (13612U: 84%).
The US-versus-international comparison uses the same measure on the SPY/SCZ price ratio, so it asks which fund has the stronger score; the winner is held only if its own price is above its 147-session average, the stand-in for a positive score.
SPY, SCZ, and TLT stand in for the source's Vanguard funds VFINX (S&P 500), VINEX (international small caps), and VUSTX (long-term Treasuries). SCZ began trading in late 2007, which limits the backtest.
Uses the source's baseline bond choice, long-term Treasuries. The source's optional rule, holding whichever of long Treasuries and TIPS had the higher one-month return, is not applied.
Signals and trades use the close of the last trading day of each month, as in the source.

Common questions#

What is Accelerating Dual Momentum?#

It is a monthly rotation between US large-cap stocks, international small-cap stocks, and long-term Treasuries, published by Engineered Portfolio in 2018. It holds whichever stock fund has the stronger short-term momentum, and switches to Treasuries when that momentum turns negative.

What ETFs does Accelerating Dual Momentum use?#

SPY for the S&P 500, SCZ for international small caps, and TLT for long-term Treasuries. The original backtest used the Vanguard funds VFINX, VINEX, and VUSTX. Some versions, including Allocate Smartly's, add TIP and hold whichever of TLT and TIP did better over the last month.

How is Accelerating Dual Momentum different from Dual Momentum?#

Dual Momentum (GEM) uses 12-month returns, broad non-US stocks, a Treasury bill hurdle, and aggregate bonds as the safe asset. ADM uses shorter 1-, 3-, and 6-month returns, international small caps, a hurdle of zero, and long Treasuries. It trades more often and reacts faster.

Does Accelerating Dual Momentum still work?#

Since its publication in 2018 it has compounded about 10% a year, but it lost 35% in 2022 and needed until mid-2025 to recover. The live record on Tactfolio shows how it is doing now.