On this page
  1. What the strategy actually does
  2. Why it paid off exactly when it mattered
  3. What it costs the rest of the time
  4. Sources

In each of the five worst years for US equities since 1985, a diversified trend-following strategy made money. It averaged +21.6% across those five years while the equity market fell between 6% and 39%. That is the entire case for holding it, and also the reason it is hard to hold.

5 of 5 Worst US equity years since 1985 in which AQR's diversified trend strategy was positive.
+21.6% Average trend-following return in those five years, while equities fell 6 to 39%.
−0.09 Correlation between monthly trend and equity market returns, full sample, 1985 to 2026.
Bar chart of AQR diversified time-series-momentum calendar-year returns, 1985 to 2025. Most years are positive, ranging from about minus 11% to plus 41%. The five bars marking the worst US equity years, 2008, 2002, 2022, 2001, and 2000, are all positive, from 11.4% to 29.2%, highlighted against the rest.
AQR diversified Time Series Momentum, calendar-year return, full years 1985-2025. Equity market: Shiller nominal total return. Both fetched 2026-09-16.

What the strategy actually does

A time-series momentum system looks at each market on its own, not against any other market. For dozens of futures contracts, spanning equity indices, currencies, commodities, and interest-rate instruments, it checks whether that one market has trended up or down over the past 12 months, sizes a position to a fixed volatility target, and goes long the ones trending up and short the ones trending down. It carries no opinion on where any single market is headed next; the trend itself is the only input.

Why it paid off exactly when it mattered

The five years above are not a cherry-picked run of good luck. They are the equity market’s five worst calendar years in this sample, found by sorting Shiller’s own return series independently, and trend following happened to be positive in each one: +11.4% in 2000, +17.9% in 2001, +29.2% in 2002, +24.6% in 2008, and +24.7% in 2022. A sustained equity decline gives a trend system months to build a short position and ride the same decline. Its correlation to the equity market over the full sample is slightly negative, not zero and not strongly positive, which is the property that lets it offset an equity drawdown rather than add to it.

What it costs the rest of the time

The same system that shorted equities into 2008 also lost money in 2016, 2018, 2020, and 2023, years when a sharp reversal turned a position around right after it opened, known as whipsaw. None of those four years coincides with a bad year for stocks; two of them, 2016 and 2023, sat inside a strong equity market. A large share of a futures market’s return over any stretch comes from its term structure, the shape of the curve between a near contract and a later one, rather than from the spot price alone; a whipsaw year is expensive on both counts at once, in the position and in the roll. Holding trend following means paying that cost every few years for a payoff that only shows up in the years an investor least expects to need it and most wants to abandon the position that isn’t working.