Trend Following

Trend following is the strategy behind the managed-futures / CTA industry: hold a diversified book of futures across equities, bonds, currencies, and commodities, and in each one go long if it has been rising and short if it has been falling. It is the canonical time-series strategy, each market is judged against its own past, not against its peers, and it has one property no other systematic strategy reliably shares: it tends to make its most money in exactly the crises when everything else is losing.

The signal: time-series momentum

The core empirical fact (Moskowitz, Ooi, Pedersen) is time-series momentum: an asset's own past 12-month excess return positively predicts its next-month return, across essentially every liquid futures market and over a century of data. The simplest implementation sizes each position by the sign of the past return, scaled to a constant risk:

$w_{i,t} = \operatorname{sign}\big(r_{i,\,t-12,\,t}\big)\times \frac{\sigma_{\text{target}}}{\sigma_{i,t}}.$

Two ingredients matter as much as the sign. First, volatility scaling, dividing by each market's volatility $\sigma_{i,t}$ (see Vol Targeting) so a calm bond future and a wild commodity contribute equal risk, and the total book is held at a constant volatility target. Second, diversification across many markets, 50–100 futures with low cross-correlations, which supplies the breadth that turns a modest per-market edge into a respectable Sharpe. Because positions do not sum to zero, the book carries time-varying net directional exposure: long risk in calm uptrends, short risk in sustained selloffs.

Why do trends exist?

Trend following is a bet that price moves are serially correlated over months, that new information is impounded gradually, not instantly, contradicting strict Market Efficiency (The EMH). The persistence is usually attributed to a chain of under- then over-reaction:

• Initial under-reaction. Investors anchor to prior beliefs and update too slowly to news (earnings, macro shifts), so prices drift toward fair value rather than jumping, creating early trend.
• Delayed over-reaction / herding. Later, momentum traders, positive-feedback flows, and risk-management rules (stops, rebalancing, capital chasing performance) push the move past fair value, extending the trend before it eventually reverses.
• Structural flows. Central-bank policy cycles, commodity supply adjustments, and slow institutional rebalancing all generate multi-month persistent price pressure.

None of these are riskless; trend following is compensated for bearing the reversals and whipsaws that punctuate the persistence.

The convexity: trend following as a straddle

The deepest property of trend following is its long-volatility, convex payoff, and it is what makes the strategy "crisis alpha." Fung and Hsieh showed trend following replicates a portfolio of lookback straddles, its return profile resembles being long options: small, frequent losses (premiums paid via whipsaws in choppy markets) and occasional large gains when a big move develops in either direction. Schematically the payoff as a function of the underlying's move $\Delta$ is convex:

$\text{P\&L} \approx c\cdot|\Delta| - \text{whipsaw cost},$

profiting from large moves up or down. This matters because crises are typically trending events, 2008, 2000–02, the 2022 bond selloff, that unfold over weeks to months. A trend book, already short equities and long bonds as the selloff builds, makes money as the trend extends, precisely when Carry, equity, and most risk-premium strategies are crashing. That negative-to-zero correlation with equities in the left tail, crisis alpha, is trend following's economic reason to exist in a portfolio, and it is the mirror image of carry, which is short the tail.

Worked example

You run a 60-market trend book at a 10% annualized vol target. Crude oil's trailing 12-month return is negative, so you are short oil, sized $\sigma_{\text{target}}/\sigma_{\text{oil}}$ contracts. A geopolitical shock sends oil down another 25% over two months: your short profits, and, because the move is large and sustained, it dwarfs the small losses you took in the choppy months prior. Simultaneously equities sell off and bonds rally; your book was already long bonds and short equity index futures on their prior trends, so those legs profit too. The whole portfolio posts a strong month while a 60/40 investor is deep in drawdown. In the following choppy, directionless recovery, the same book bleeds small amounts as trends whipsaw, the premium you pay for the convexity.

Failure modes

• Whipsaws and range-bound markets. The long-option profile loses when markets chop sideways with no sustained trend, the strategy's worst environment is a low-volatility, mean-reverting, V-shaped-recovery regime (e.g. much of the 2010s, and sharp reversals like 2020's crash-and-snapback).
• Reversals at turning points. Trend following is late by design; it gives back a chunk at every trend reversal because it must wait for the new trend to establish before flipping.
• Correlated drawdowns across the book. In "risk-on/risk-off" regimes, the diversified markets collapse to a single bet, gutting the breadth the Sharpe relies on.
• Cost and capacity. Vol-scaling and trend flips generate turnover; execution and slippage across many futures are a first-order drag (Transaction Costs, Alpha Decay).
• Parameter sensitivity. Lookback length, vol-window, and rebalancing rules are easy to overfit; robust implementations blend multiple horizons rather than optimizing one.

In interviews

Define trend following as time-series momentum, sign of each market's own trailing (≈12-month) return, vol-scaled and diversified across dozens of futures, and contrast it with cross-sectional momentum (own-history vs. peer-ranking; net directional vs. dollar-neutral). Explain persistence via under-reaction (slow diffusion of news) then over-reaction/herding, noting it contradicts strict efficiency. The signature point is convexity: trend following behaves like a long straddle / lookback-option portfolio, small frequent losses in choppy markets, large gains in big moves either direction, which is why it delivers crisis alpha, profiting in sustained selloffs when carry and equity risk premia crash. A strong closer pairs it with carry: trend is long the tail, carry is short it, so a trend-plus-carry book smooths both their weaknesses. See Cross-Sectional vs. Time-Series Strategies and Carry.