Skip to content

Is This Time Different? Trend Following and Financial Crises: summary and key takeaways

The paper builds a diversified trend following portfolio across futures and forwards and compares its returns in crisis and no-crisis periods. Returns after the start of a financial crisis are much lower than in other periods, and the paper links this to weaker return persistence in the underlying markets.

Hutchinson, M.C.; O'Brien, J. · 2019 · SSRNSummary updated 20 November 2026
Published2019Working paper
Asset focusMulti-assetTrend following / crisis alpha
TypeBothTrend & Momentum
Reading time4 minthis summary

Is This Time Different? Trend Following and Financial Crises at a glance

The question
Is weak trend following performance after the 2008 crisis typical of what happens after financial crises, and what changes in the underlying markets?
Dataset & period
Global sample: futures and synthetic forwards covering twenty one equity indices, twenty one commodities, thirteen government bonds, plus currency crosses built from nine exchange rates, January 1921 to June 2013, with January 1940 to December 1949 excluded. Portfolio results are reported from 1925 to 2013, monthly. Six global crises (1929, 1973, 1981, 1987, 2000, 2007 starts) plus eight regional crises.
Method
Each instrument gets a signal of +1 or -1 for each look-back from one to twelve months, based on the sign of past excess return. The signals are averaged, then sized by inverse ex-ante volatility to a target volatility. Risk is split equally across four asset classes. Crisis periods are fixed windows of twenty four or forty eight months after the equity market high. Pooled regressions of volatility-scaled returns on their own lags test return predictability in crisis versus no-crisis periods.

Key findings of Is This Time Different? Trend Following and Financial Crises

  • Over the full sample from 1925 to 2013 the portfolio averaged 12.1% a year net of fees, with volatility of 11%.
  • The portfolio correlated 0.76 with the Newedge trend following CTA index over 2000 to 2013.
  • In the opening twenty four months after a crisis began, the portfolio averaged 4%, against 13.6% in no-crisis months. Over forty eight months the figures were 6% and 14.9%.
  • The full-sample Sharpe ratio of the global portfolio was 1.1.
  • Each global crisis had a long flat spell with no excess return over cash. These ran from 18 months to 54 months, about three years on average.
  • In regional crises the average return in year 1 was -1.3%, with a wide range across crises. Performance picked up from about the third year.
  • Outside crises, a futures contract's past returns predict its next return for up to twelve months. In crises that link weakens and some lags flip sign. Commodities differ.

Robustness, as reported in the paper

  • Repeating the analysis on the 1980 to 2013 exchange-traded futures period gave consistent results, apart from currencies at four years.
  • Swapping exchange-traded futures for synthetic forwards in tests left the series nearly perfectly correlated and the results essentially unchanged.
  • Using price return instead of excess return for the signal, or local currency instead of hedged US dollar returns in the regional test, produced nearly the same conclusions.
  • Extending the crisis window gave regression results very close to the base case.

Caveats and limits

  • Only six global crises exist in the sample, and they differ in character. The paper adds regional crises for support.
  • The crisis windows are fixed lengths of twenty four and forty eight months, not dated crisis by crisis.
  • Data limits allow significance tests only on cumulative betas, not on single-lag betas.
  • The paper does not test why return predictability weakens. Behavioural, intervention and hedging-pressure explanations are offered as untested possibilities.
  • Pre-war and pre-1980 returns rely on derived forward prices and estimated historical trading costs (inferred).

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

The paper treats a post-crisis period as a distinct state for trend following, with weak returns lasting around four years on average. Allocators reviewing CTA performance can compare current results against this crisis-conditional pattern. It also points to reduced serial correlation as the mechanism to monitor.

Questions about this paper

How long does trend following underperform after a crisis?

The paper finds weak performance for about four years on average after a global crisis starts, with regional crises showing a shorter effect.

Which asset class does not follow the pattern?

Commodities earn comparable returns in both states, because their returns keep trending for roughly the first five months of a crisis.

Why does trend following struggle in crises?

Serial correlation over lags up to twelve months drops in crises and at times goes negative, so fewer trends are available.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.2139/ssrn.2375733. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.