Skip to content

Understanding the Momentum Risk Premium: Trend-Following Strategies: summary and key takeaways

The paper extends a continuous-time trend-following model to many assets and uses it to explain the payoff, skewness, hit ratio and diversification of momentum strategies. It then fits the model to 20 futures-based assets and replicates much of the SG CTA Index. It also concludes that trend-following gives volatility protection but not tail-risk protection.

Jusselin, P.; Lezmi, E.; Malongo, H. et al. · 2017 · SSRN Electronic JournalSummary updated 20 November 2026
Published2017Working paper
Asset focusMulti-assetMomentum risk premium
TypeStrategyTrend & Momentum
Reading time4 minthis summary

Understanding the Momentum Risk Premium: Trend-Following Strategies at a glance

The question
What does a tractable multi-asset model say about how the momentum risk premium earns its return and how it behaves as a diversifier and hedge?
Dataset & period
Mostly theory and simulation with illustrative parameters. The empirical part uses 20 assets (short rates, government bonds, stock indices, currencies, commodities) from January 2000 to July 2017, rebalanced daily, and compares results with the SG CTA Index.
Method
Asset prices follow a geometric Brownian motion whose trend drifts randomly. The best trend estimate is an exponentially weighted moving average. Exposure is proportional to that estimate. Profit and loss then splits into an option-like profile (short-run) and a trading impact (long-run), and the authors derive its distribution, hit ratio and moments, first for one asset and then for many correlated assets.

Key findings of Understanding the Momentum Risk Premium: Trend-Following Strategies

  • The strategy tends to lose more often than it wins when the asset Sharpe ratio is low: with a one-year moving average the hit ratio is below 50% when the Sharpe ratio is under 0.35. Average gains still exceed average losses.
  • A positive trading impact needs the absolute Sharpe ratio to exceed a threshold that falls as the averaging window lengthens: 1.41 for a three-month window and 0.71 for a one-year window.
  • Losses of the strategy are bounded while gains are not, so the payoff distribution is right-skewed. More uncorrelated assets leave the mean unchanged but shrink volatility, skewness and kurtosis toward a Gaussian shape.
  • For time-series momentum the best diversification comes at zero correlation, and a few assets with strong trends beat many weak ones. Cross-section momentum prefers highly correlated assets and cares about the dispersion of Sharpe ratios.
  • In the backtests, long-run performance comes mainly from the trading impact, not from the straddle-like option profile. An equal-weighted mix of the 20 trend strategies at a 20% volatility target reaches a daily correlation of 78% with the SG CTA Index at an optimal lambda of 2.3.
  • Excessive leverage can make gamma costs prohibitive, and the strategy is hurt by rising volatility (negative vega).

Robustness, as reported in the paper

  • The paper checks the model decomposition against a simulated Brownian motion and finds modelled and actual strategy returns very close.
  • It reports that estimated optimal trend durations are sensitive to the study period, and that equities and bonds show different durations over the last five and ten years.
  • It finds that a moderately wrong averaging window causes relatively small estimation errors.
  • Replication of the CTA index by single asset classes gives lower correlations than the diversified mix, so diversification is key.

Caveats and limits

  • The authors call it a toy model: no transaction costs, no management or performance fees, continuous rebalancing, and a fixed given allocation.
  • Prices are assumed to be geometric Brownian motions, with no jumps or discontinuities, so trend estimation is easier than in real markets.
  • Simulations show trend-following does not protect against sudden skewness events, such as the CTA experience in January 2015.
  • Most results are analytical or simulated rather than out-of-sample live results (inferred).

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

Why it matters

The paper gives a framework linking trend-following to option-like payoffs, volatility and the Sharpe ratio of the underlying trend. It clarifies why the strategy suits a mix with concave-payoff risk premia and why universe and leverage choices matter. It also separates volatility protection from tail-risk protection.

Questions about this paper

Does trend-following win more often than it loses?

Not when asset Sharpe ratios are low. In the model the hit ratio falls below 50% when the Sharpe ratio is under 0.35 for a one-year window, but gains are larger on average than losses.

What Sharpe ratio is needed for a positive trading impact?

The absolute Sharpe ratio must exceed a bound that depends on the averaging window: 1.41 for three months and 0.71 for one year.

Can trend-following hedge tail risk?

The paper finds it can provide protection during gradual downturns but is not able to manage tail risk when sudden skewness events cause sharp falls.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.2139/ssrn.3042173. 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.