Trend-following crisis alpha: Does it come from beta timing or market selection?

Highlights

• Proprietary Fidelity research explores whether the performance of trend-following strategies is driven by the timing of beta exposures or the ability to select markets within each asset class to take relative value positions based on trend signals.
• This research follows a four-step framework:
  1. Defines prototypical characteristics of trend-following strategies and how they can be viewed as crisis alpha.
  2. Decomposes trend exposures into beta-timing and relative value portfolios, using a simple trend-following model.
  3. Studies how much each contributes or detracts from crisis alpha.
  4. Considers the benefits of enhancing a trend strategy by investing less in the cross-sectional trend portfolio while using other signals as potential drivers of relative value positioning.
• As expected, analysis across key metrics including risk-adjusted returns, correlation, and downside correlation, underscores the unique value of trend-following strategies as a complement to equity, with demonstrated historical ability to perform during periods of market dislocations.
• Importantly, the results suggest that it may be possible to keep the crisis alpha characteristics intact while enhancing trend strategies with a richer set of alpha signals in the relative value component of the portfolio.

Proprietary research from Fidelity analyzes the drivers of crisis alpha in managed futures’ trend-following strategies and argues that complementary approaches can be used to enhance strategy efficiency and stability without giving up defensiveness.

The behavior of trend-following strategies has sometimes been referred to as “crisis alpha.”¹ For the purposes of this paper, we will define several simple characteristics as emblematic of trend-following crisis alpha and then investigate whether they are driven by trend portfolios’ timing of beta exposures (the timing of being long or short an asset class, and the sizing of the exposure) or from their ability to select markets within each asset class to take relative value positions based on trend signals. Some of the metrics we consider are average correlation to equities, downside correlation to equities, unconditional risk-adjusted return, and risk-adjusted return conditional on equities being down.

We find that the risk-taking in trend-following strategies is dominated by betatiming decisions, which contribute over 76% of the risk of the simple trendfollowing strategy we use in this study. Mostly because beta timing decisions are the bulk of the risk, they drive most of the prototypical characteristics of trend-following strategies. We also find that compared with the relative value component, the beta-timing component of trend returns has more negative correlation to equities on average and is more negatively correlated in months when stocks are down. In addition, beta timing has a higher historical riskadjusted return, both on average and conditional on stocks being down.

These results imply that it may be possible to keep the crisis alpha characteristics intact while also enhancing trend strategies with a richer set of alpha signals in the relative value component of the portfolio. Taking it one step further, there may be an opportunity to further improve trend strategies by intentionally increasing exposure to trend-based, cross-sectional positions in the asset classes where they are most likely to generate return and equity diversification and dialing them back in asset classes where they are less accretive to crisis alpha.

This paper proceeds as follows. First, we define some prototypical characteristics of trend-following strategies and show how they can be viewed as “crisis alpha.” Second, we define simple trend-following and long-only asset class portfolios to use as a laboratory for understanding where trend returns come from. Third, we decompose trend exposures into beta-timing and relative-value portfolios and study their characteristics—how much they contribute to or subtract from crisis alpha. Fourth, we consider enhancing a trend strategy by investing less in the cross-sectional trend portfolio, with other signals as potential drivers of relative value positioning.

Read the paper