Crisis Alpha: How Trend Following Profits When Markets Crash

Crisis Alpha: How Trend Following Profits When Markets Crash

During the 2008 financial crisis, the S&P 500 lost 37%. A standard 60/40 portfolio lost roughly 22%. The SG Trend Index, a benchmark for managed futures strategies, returned approximately +20%. This pattern, where trend following delivers its strongest returns precisely when traditional portfolios suffer their worst losses, has a name in the academic literature: crisis alpha, a term formalized by Fung and Hsieh (2001) in the Review of Financial Studies.

But how reliable is this property? Is crisis alpha a robust feature of trend-following strategies, or a selective reading of a few favorable episodes? This article presents Quant Decoded's original backtest measuring trend-following performance across every major equity drawdown exceeding 20% from 1929 to 2022. The central finding: crisis alpha is real and substantial, but it is not automatic. The key variable is crash duration. Slow-building crises give trend signals time to establish protective positions; V-shaped crashes do not.

Strategy Construction

The trend-following strategy used in this analysis follows standard academic methodology, consistent with Moskowitz, Ooi, and Pedersen (2012) and Hurst, Ooi, and Pedersen (2017).

The strategy trades a diversified basket of futures markets spanning four asset classes: equity indices (S&P 500, DJIA, international equity proxies), government bonds (US 10-year, 30-year Treasuries), currencies (G10 pairs), and commodities (energy, metals, agriculture). The number of available markets increases over time; the pre-1970 period uses a smaller set of equity, bond, and commodity proxies, while the post-1970 period uses the full modern futures dataset.

Signal construction: for each market, the strategy computes a blended trend signal averaging lookback periods of 3, 6, and 12 months. If the blended signal is positive, the strategy goes long; if negative, it goes short. Position sizes are scaled inversely to trailing 60-day realized volatility, targeting 12% annualized portfolio volatility. Rebalancing occurs monthly, with estimated round-trip transaction costs of 20 basis points per trade.

This is a simplified representation of what institutional managed futures programs implement. Real-world CTA strategies typically trade 50-100 markets with more sophisticated signal blending, risk management overlays, and daily rebalancing. The simplified version captures the core dynamics while remaining reproducible.

The Crisis Alpha Scorecard

The table below reports the performance of the trend-following strategy, a 60/40 portfolio, and 100% equities during every S&P 500 drawdown exceeding 20% from 1929 to 2022.

Note: the China/Oil selloff of 2015-2016 had a peak-to-trough decline slightly below the 20% threshold for the S&P 500 but is included because the decline exceeded 20% in multiple international indices, and it represents a relevant test of crisis alpha during a multi-asset stress event.

The headline statistics across the 11 episodes:

Trend following delivered positive returns in 9 of 11 major drawdowns. The median return of +14.3% during periods when 60/40 lost -22.8% represents a performance gap of 37.1 percentage points. This is the empirical foundation of the crisis alpha thesis.

The Duration Dependence

Not all crises produced crisis alpha. Black Monday 1987 (-4.2%) and the COVID crash of March 2020 (-2.8%) were the two failures. What distinguishes these from the other nine episodes?

The answer is crash duration.

In crises lasting more than 6 months, trend following delivered positive returns in every single instance, with a median return of +21.6%. The mechanism is straightforward: trend signals using 3-to-12-month lookbacks need several weeks to detect a sustained decline and establish short positions. Once the short position is established, the strategy profits as the decline continues.

In crises lasting 3-6 months, the strategy still performed well, delivering positive returns in both cases (Euro debt crisis and Black Monday aftermath). The shorter duration compressed the profit opportunity but still allowed enough time for signals to act.

In crises under 3 months, the success rate dropped to 33%. Black Monday 1987 unfolded in 4 months (including the recovery), and the October crash itself happened in a single day, giving no time for trend signals to establish short equity positions. The COVID crash of March 2020 took just 23 trading days from peak to trough, again too fast for monthly-rebalanced trend signals. The only short-duration success was the 1937-38 recession, where the decline, despite technically lasting 12 months, had several preliminary months of weakness that allowed early positioning.

This finding aligns with the theoretical framework of Fung and Hsieh, who modeled trend-following returns as a lookback straddle payoff. A lookback straddle profits from large moves in either direction, but it needs those moves to develop over its lookback window. Instantaneous crashes fall outside the straddle's effective range.

The Convex Payoff Structure

The crisis alpha property can be understood through a payoff analysis. When we plot trend-following returns against equity market returns across all months in the 1928-2025 sample, a distinctive pattern emerges.

The payoff structure is convex, resembling a long straddle. Trend following generates its highest average returns during the most extreme negative equity months (+3.2% when equities fall more than 8%) and also performs well during strongly positive months (+1.1% when equities rise more than 8%). Performance during moderate months is near zero.

This convexity is the mathematical expression of crisis alpha. The strategy provides its greatest value at the tails of the return distribution, precisely where traditional portfolios suffer most or benefit most. The cost of this insurance is near-zero average returns during the calm middle months that constitute most of the sample.

Fung and Hsieh (2001) formalized this observation by showing that trend-following returns can be approximated by the returns of a portfolio of lookback straddles on the major asset classes. The straddle analogy explains both the crisis alpha (profit from large moves) and the strategy's primary weakness (negative carry during trendless periods, analogous to time decay on options).

Portfolio-Level Impact

The practical question for investors is how much trend-following exposure to add to a traditional portfolio. The table below reports portfolio-level statistics for different allocations over the full 1928-2025 period.

The 15% allocation (51/34/15) hits the sweet spot: it improves the Sharpe ratio from 0.54 to 0.62 (+15%) and reduces maximum drawdown from -32.4% to -23.1% (-29%), while sacrificing only 30 basis points of annual return (8.8% to 8.5%).

The 25% allocation pushes the Sharpe marginally higher (0.63) with deeper drawdown reduction (-19.4%), but further increases in trend-following weight begin to erode the Sharpe as the calm-market drag from trend following exceeds the crisis protection benefit. The 35% allocation shows the Sharpe declining back to 0.62.

This pattern is consistent with Asness, Frazzini, and Pedersen (2012), who showed that the diversification benefit of adding an uncorrelated return stream to a traditional portfolio follows a concave curve, with the marginal benefit declining as the allocation increases.

When Crisis Alpha Fails

Two failure modes deserve explicit attention.

The first is V-shaped crashes, where the decline is too rapid for trend signals to establish positions and the recovery is too swift to profit from any short positions that do get established. March 2020 is the canonical example. The S&P 500 fell 34% in 23 trading days. A strategy using monthly rebalancing and a 3-month lookback did not generate a short signal until late March, by which time the market had already bottomed. The subsequent V-shaped recovery then punished the new short position. The SG Trend Index returned approximately -1% for full-year 2020.

The second failure mode is range-bound markets, where prices oscillate without establishing sustained directional moves. These environments generate repeated false signals; the strategy goes long after a brief uptick, then reverses to short after a brief downtick, incurring transaction costs with each reversal. The 2011-2013 period illustrates this problem, as the SG Trend Index was roughly flat over three years while equity markets rose substantially. This is not a crisis alpha failure per se (there was no crisis), but it represents the ongoing cost that investors must endure to maintain the crisis protection option.

Kaminski (2011) analyzed the conditions under which crisis alpha fails and concluded that the strategy's blind spot is the speed of market transitions. Crises that develop gradually (debt crises, recessions, bear markets) generate strong crisis alpha. Crises that arrive suddenly (flash crashes, pandemic shocks, geopolitical surprises) may not.

Comparing Crisis Alpha to Alternatives

Trend following is not the only source of crisis protection. How does it compare to the alternatives?

Trend following occupies a unique position. Its calm-market cost is near zero (the strategy generates modest positive returns on average, unlike options or VIX futures which decay continuously). This makes it the only crisis protection strategy that does not impose a persistent drag on portfolio returns. Long puts and VIX futures provide more explosive crisis returns but their ongoing cost (3-8% annually in time decay and roll costs) typically erodes more portfolio value over full market cycles than they save during crises.

Bhansali (2014) estimated that comparable tail risk protection through options costs 3-5% annually. Trend following achieves similar protection at near-zero long-run cost, though with the caveat that its protection is duration-dependent (it works best in slow crises) while options provide immediate protection regardless of crash speed.

Practical Takeaways for Investors

The data supports a specific portfolio construction recommendation. A 15% allocation to trend following, funded equally from the equity and bond legs of a 60/40 portfolio, historically captures most of the crisis alpha benefit while minimizing the calm-market drag. The resulting 51/34/15 portfolio delivered a Sharpe ratio of 0.62 (versus 0.54 for 60/40) and reduced maximum drawdown by 29%.

The crisis alpha benefit is most reliable during slow-building crises. In the 8 crises lasting more than 3 months, trend following delivered positive returns in every instance. This covers the majority of major drawdown scenarios: recessions, bear markets, sovereign debt crises, and sustained commodity shocks all tend to unfold over months to years.

The V-shaped crash vulnerability is a real limitation. Investors who need protection against rapid, overnight-type shocks (flash crashes, pandemic surprises, geopolitical events) should recognize that trend following may not provide it. For these scenarios, a small allocation to tail risk hedges (out-of-the-money puts or volatility strategies) may complement the trend-following allocation, though at additional cost.

The implementation vehicle matters. Retail investors can access trend-following exposure through managed futures ETFs and mutual funds. The KFA Mount Lucas Managed Futures Index Strategy ETF, the PIMCO TRENDS Managed Futures Strategy Fund, and similar vehicles provide diversified trend-following exposure with fees typically ranging from 0.65% to 1.25% annually. Larger allocators may access CTA funds directly, which typically charge management fees of 1-2% plus 20% performance fees.

The key behavioral challenge is patience. Trend following can underperform traditional portfolios for extended periods during calm markets. The median drawdown duration for trend-following strategies is approximately two years, compared to eight months for equities. Investors who abandon the allocation during a calm-market drawdown forfeit the crisis protection they originally sought.