Skip to content

Global Equity Momentum: A Craftsman's Perspective: summary and key takeaways

The article replicates the Global Equity Momentum (dual momentum) strategy and tests many lookback variants against it. It finds the original twelve-month version is statistically indistinguishable from the alternatives, while an ensemble of all variants produced similar long-run returns with smaller drawdowns and steadier shorter-horizon outcomes.

Butler, A. et al. (ReSolve Asset Management) · 2019 · ReSolve Asset ManagementSummary updated 20 November 2026Practitioner research
Published2019Firm research
Asset focusMulti-assetMomentum / TAA
TypeBothTactical Allocation
Reading time5 minthis summary

Global Equity Momentum: A Craftsman's Perspective at a glance

The question
Does the choice of a single lookback specification for Global Equity Momentum matter, and can combining many specifications reduce that risk?
Dataset & period
S&P 500, MSCI World ex US and a bond proxy (half AAA corporates, half 10-year Treasuries); monthly total returns from 1950 to September 2018 (828 months); 324 lookback combinations per variant, plus moving-average variants.
Method
The authors replicate the original strategy, which holds US stocks, foreign stocks or bonds based on trailing returns. They then build every combination of one to eighteen month trend and relative-momentum lookbacks, add moving-average versions and a second bond-switch rule. Block bootstrap tests compare the original with all alternatives. An ensemble averages the target weights of all variants each month, and is also run on synthetic indexes and under a simple US tax model.

Key findings of Global Equity Momentum: A Craftsman's Perspective

  • Over a ten-year window ending 2018, the same strategy built on a ten-month lookback returned 146% while the nine-month version returned 43%, showing large short-horizon dispersion between near-identical specifications.
  • In block bootstrap tests the original beat the alternatives by an average of 0.8% annualized, and outperformed only about 61% of the time. Against the weakest alternative it won in 93.26% of samples, short of the authors' threshold to reject equal merit.
  • Out of sample, the original's compound return of 13.14% converged on the median across specifications at 12.92%, after ranking well in-sample.
  • The ensemble's Sharpe ratio was 0.93 against a median of 0.83 across specifications, and its average of the five worst drawdowns was 13.2% against 17.4% for the median.
  • Compared with the original, the ensemble made 1867 trades and an extra 0.5% per year in tax drag, before a trade filter was applied.

Robustness, as reported in the paper

  • Block bootstrap comparison of the original against every alternative specification.
  • Separate in-sample and out-of-sample periods, using the dates before and after the original paper's sample.
  • Repeat tests on many synthetic US and foreign indexes built from random portfolios; the authors say results matched the main tests and the ensemble's stability metrics sat at the top of the distribution.
  • A trade filter based on a noise-trading model roughly halved ensemble trades and nearly closed the tax gap, while keeping the Sharpe and drawdown benefits.

Caveats and limits

  • Main results are gross of taxes and transaction costs; costs are not modelled beyond the tax section.
  • The replication uses proxy indexes (MSCI World ex US instead of ACWI ex US, a blended bond proxy), which the authors note can shift long-run results.
  • The article is a firm note from an asset manager promoting an ensemble approach, not a peer-reviewed study (inferred).
  • The tax model is a simple US model with fixed rates and year-end payment, and it assumes the filter threshold generalizes (inferred).

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

Why it matters

Practitioners who run a single lookback are exposed to specification luck that the article says is not rewarded. The work gives a concrete way to test that claim, by comparing one chosen model with the full distribution of nearby models. It also shows how to weigh any extra turnover or tax cost of a multi-model design.

This is a firm research note or web article, not a peer-reviewed paper.

Questions about this paper

Is the original twelve-month Global Equity Momentum specification statistically better than other lookbacks?

The paper's bootstrap tests could not reject that it has the same mean return as the alternatives. Its edge in-sample faded toward the median out of sample.

What is the ensemble approach in this article?

It averages the target weights of all tested specifications each month, so the portfolio reflects many lookbacks and signal definitions at once.

Does the ensemble cost more in taxes?

Yes, before filtering the authors estimate roughly 0.5% a year more tax drag, which they say shrinks after rebalancing gains and a trade filter.

Where can I read the original paper?

The original is at its publisher: https://hedgefundalpha.com/strategies/global-equity-momentum-a-craftsmans-perspective/. 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.