Skip to content

Two Centuries of Multi-Asset Momentum: summary and key takeaways

The paper builds monthly return histories for six asset classes back to 1800 and tests cross-sectional price momentum inside and across them. Momentum is positive and significant in every class except commodity spot prices, which show mean reversion instead. Momentum betas swing with the market state, and correlations between momentum strategies have risen sharply in recent years.

Geczy, C.C.; Samonov, M. · 2015 · SSRNSummary updated 20 November 2026
Published2015Working paper
Asset focusMulti-assetMomentum
TypeStrategyTrend & Momentum
Reading time5 minthis summary

Two Centuries of Multi-Asset Momentum at a glance

The question
Does cross-sectional price momentum show up consistently across asset classes and across two centuries of data, and what drives its risk?
Dataset & period
Monthly data from 1800 to 2014 (from Global Financial Data, CRSP, Bloomberg and other sources): country equity indices, currencies, government bond indices, commodity spot prices, country-sectors and U.S. stocks. Sector and U.S. stock series start a few years later. Commodity spot tests are also run on a longer history reaching back to 1531.
Method
Each month assets in a class are ranked on the price change from 12 months ago to 2 months ago. The strategy buys the top third (winners) and sells the bottom third (losers), equal-weighted, rebalanced monthly. Sector momentum is ranked within each country first. A cross-asset version ranks the four classes against each other, and a combined portfolio equally weights the six within-class strategies plus the cross-asset one. Commodity results use the inverse signal because spot prices mean-revert.

Key findings of Two Centuries of Multi-Asset Momentum

  • Country equity momentum has the largest long-short spread at 0.88% per month (t-stat 10.6). Currency momentum earns 0.51% per month (t-stat 9.6).
  • Government bond momentum is the weakest at 0.13% per month (t-stat 2.3), while country-neutral sector momentum earns 0.36% per month (t-stat 6.6).
  • The cross-asset momentum strategy earns 0.45% per month (t-stat 10.2). Pooling all seven strategies with equal weights gives the same 0.45% per month, with a t-stat of 15.4.
  • Commodity spot prices show 10-month mean reversion, not momentum. Commodities without traded futures show -0.82% per month (t-stat -3.4) for the direct momentum signal, against -0.07% per month for those with futures, since the futures era began.
  • Momentum profits are higher after up markets than after down markets: 0.56% per month versus 0.15% for the combined portfolio. Equity momentum beta runs from 0.63 in up states to -0.56 in down states.
  • A simple time-series trend version earns 0.38% per month (t-stat 14.5), below the cross-sectional momentum spread but still significant in every class.
  • Average correlation between the seven momentum strategies is low over the full history, but the five-year correlation reached an all-time high of 41% at the end of the sample.

Robustness, as reported in the paper

  • Outlier filters at different cutoffs leave the combined result close to unchanged, at 0.43% to 0.45% per month.
  • Volatility-weighted momentum returns less, at 0.32% per month, but the t-stat is essentially the same (15.2).
  • Real-GDP weighting strengthens country equity momentum and makes bond momentum insignificant. Results are reported in local currency and also converted to USD.
  • Removing seasonality from commodity spot prices does not explain the mean reversion. The paper also tests skipping zero, one or two months before measuring returns.

Caveats and limits

  • Early index data are lower quality than modern indices, and the paper itself cites data errors and splicing across sources.
  • Results are mostly gross of transaction costs, and early-period asset classes and currency regimes were not tradable in the way tested (inferred).
  • Commodity momentum is measured on spot prices, not rolled futures, so the commodity result differs from the usual futures-based finding and is flipped to the inverse signal.
  • Momentum is negative over many decade-long windows, especially in the 19th century, and strategy correlations have risen, which the authors link to crowding.

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

Why it matters

The paper gives a very long out-of-sample record for momentum, including long stretches where it lost money. It also shows that momentum risk depends on the state of the market it is built in, and that diversification across momentum strategies has shrunk recently. Portfolio builders can use this to judge drawdown length and crowding risk.

Questions about this paper

Which asset class has the strongest momentum premium in the paper?

Country equities, at 0.88% per month in the long-short spread. Government bonds are the weakest at 0.13% per month.

Does momentum work in commodity spot prices?

No. Spot prices show mean reversion, so the paper reports the inverse signal for commodities. The reversion is weaker or absent for commodities with traded futures.

How does the paper find trend compares with cross-sectional momentum?

With a comparable signal definition, trend earns 0.38% per month, below momentum, though it is still significant in each asset class.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.2139/ssrn.2607730. 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.