US Cross-Asset Momentum (stocks and Treasuries time each other)

May 2008 – Sep 2026	Strategy	SPY
Annual return (CAGR)	7.5%	11.8%
Worst drawdown	-33.7%	-50.7%
Sharpe ratio	0.63	0.67
Volatility	12.8%	19.7%
Annual return since publication (Jan 2017)	6.8%	15.3%

Jan 1963 – Sep 2026	Strategy	SPY
Annual return (CAGR)	9.9%	10.8%
Worst drawdown	-33.7%	-55.2%
Sharpe ratio	0.98	0.71
Volatility	10.2%	16.5%
Annual return since publication (Jan 2017)	6.8%	15.3%

US Cross-Asset Momentum is a long-only version of the cross-asset time-series momentum that Aleksi Pitkäjärvi, Matti Suominen, and Lauri Vaittinen described in a 2016 working paper, published in the Journal of Financial Economics in 2020. Each month it holds US stocks when both stocks and Treasuries have beaten Treasury bills over the past year, Treasuries when only Treasuries have, and bills otherwise.

The idea#

Ordinary time-series momentum says an asset that has risen over the past year tends to keep rising. The authors found that stocks and bonds also carry information about each other. Strong bond returns have been followed by better stock returns, and strong stock returns by weaker bond returns. A common reading is that falling yields support stocks with a delay, while a strong economy that lifts stocks eventually pushes yields up.

Combining an asset's own momentum with the other market's momentum gave a clearly better risk-adjusted return than own momentum alone in their 20 developed markets. The paper trades long and short in every country; this version, the one Allocate Smartly tests, keeps the long side in the US.

How it works#

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

Measure the 12-month returns of the S&P 500 (SPY), 7–10 year Treasuries (IEF), and Treasury bills (BIL).
If both SPY and IEF beat BIL, hold 100% SPY.
If only IEF beats BIL, hold 100% IEF.
Otherwise hold 100% BIL.

What the backtest shows#

Over the ETF era, from May 2008, the strategy returned about 7.5% a year, against about 12% for the S&P 500 and 8.4% for a 60/40 stock and bond mix. It trailed the 60/40 on return, on worst drawdown, and on risk-adjusted return.

It sidestepped the 2008 crash and gained 16% from late May that year while the S&P 500 lost 34%, and it made a small gain in 2022 when stocks fell 18%. Its weak years came in strong markets: in 2020 and 2021 it made 1% and 6% while the S&P 500 made 18% and 29%. Its worst drawdown, about a third, came from February to March 2020, when it was holding stocks through the crash; it recovered by February 2021.

Since the working paper appeared at the end of 2016 it has returned about 7% a year with a risk-adjusted return close to its full-period figure.

With simulated history the test starts in 1963 and returns about 10% a year, just below the S&P 500, with a far better risk-adjusted return. It made money in 1973, 1974, 2001, and 2002, years when US stocks fell. The 2020 crash remains its worst drawdown.

When it struggles#

Sudden crashes. With both 12-month signals positive going into early 2020, it held stocks through a fast decline.
Strong bull markets with falling bonds. When stocks rise but Treasuries lag bills, it sits in cash and misses the gains.
Frequent cash. It often holds neither asset, which lowers returns in rising markets.

Using it on Tactfolio#

The live strategy above runs these rules on SPY, IEF, and BIL. It is a simplified form of the paper: no short positions, one country, and no volatility scaling. Copy it to try TLT or TIP as the bond signal, or add it beside a trend strategy such as Faber's 10-month moving average.

Year by year

Year	Strategy	SPY
2026*	5.5%	14.0%
2025	10.0%	17.7%
2024	11.2%	24.9%
2023	4.9%	26.2%
2022	1.4%	-18.2%
2021	6.2%	28.7%
2020	0.6%	18.3%
2019	17.0%	31.2%
2018	7.2%	-4.6%
2017	3.4%	21.7%
2016	6.3%	12.0%
2015	-7.5%	1.2%
2014	8.2%	13.5%
2013	15.1%	32.3%
2012	16.0%	16.0%
2011	1.9%	1.9%
2010	13.6%	15.1%
2009	3.4%	26.4%
2008*	16.0%	-34.3%

Year	Strategy	SPY
2026*	5.5%	14.0%
2025	10.0%	17.7%
2024	11.2%	24.9%
2023	4.9%	26.2%
2022	1.4%	-18.2%
2021	6.2%	28.7%
2020	0.6%	18.3%
2019	17.0%	31.2%
2018	7.2%	-4.6%
2017	3.4%	21.7%
2016	6.3%	12.0%
2015	-7.5%	1.2%
2014	8.2%	13.5%
2013	15.1%	32.3%
2012	16.0%	16.0%
2011	1.9%	1.9%
2010	13.6%	15.1%
2009	3.4%	26.4%
2008	7.1%	-36.8%
2007	6.4%	5.1%
2006	7.1%	15.8%
2005	8.1%	4.8%
2004	12.3%	10.7%
2003	11.7%	28.2%
2002	11.8%	-21.6%
2001	6.5%	-11.8%
2000	-7.3%	-9.7%
1999	8.7%	20.4%
1998	28.7%	28.7%
1997	24.5%	33.5%
1996	12.3%	22.5%
1995	23.4%	38.0%
1994	-0.6%	0.4%
1993	9.7%	9.7%
1992	7.6%	7.6%
1991	26.1%	30.3%
1990	6.6%	-3.2%
1989	23.8%	31.5%
1988	7.7%	16.4%
1987	24.8%	5.1%
1986	18.6%	18.6%
1985	25.9%	31.6%
1984	11.5%	6.2%
1983	22.2%	22.4%
1982	36.9%	21.6%
1981	14.9%	-4.9%
1980	11.9%	32.4%
1979	10.5%	18.3%
1978	7.3%	6.5%
1977	-1.5%	-7.2%
1976	23.8%	23.8%
1975	-5.6%	37.1%
1974	8.0%	-26.5%
1973	7.1%	-14.7%
1972	12.0%	18.9%
1971	11.9%	14.1%
1970	7.1%	3.9%
1969	6.7%	-8.4%
1968	5.4%	10.9%
1967	8.9%	23.8%
1966	4.7%	-10.1%
1965	6.2%	12.4%
1964	3.4%	16.4%
1963*	4.9%	23.5%

* 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 paper trades long and short in 20 developed markets with volatility-scaled positions. This version keeps only its long signals in the US: stocks when both excess returns are positive, bonds when bonds are positive and stocks negative, and Treasury bills otherwise, as Allocate Smartly tests it.
SPY and IEF stand in for the paper's US stock index and government bond index; excess returns are measured against BIL over 252 sessions.
Signals and trades use the close of the last trading day of each month, as in the source.

Common questions#

What is cross-asset momentum?#

It is the finding that one market's past returns help predict another's. In Pitkäjärvi, Suominen, and Vaittinen's study, rising bonds predicted rising stocks, and rising stocks predicted falling bonds.

What ETFs does US Cross-Asset Momentum use?#

SPY for US stocks, IEF for 7–10 year Treasuries, and BIL for Treasury bills, which serve both as the hurdle and as the cash holding.

How is it different from dual momentum?#

Dual momentum asks whether stocks themselves beat bills. This strategy also requires Treasuries to beat bills before it owns stocks, and it holds Treasuries only when stocks are weak.

Does cross-asset momentum still work?#

Since the paper first circulated in 2016 the US version has earned about 7% a year. Over the whole ETF era it trailed a 60/40 portfolio. The live record on Tactfolio shows how it is doing now.