Beyond the dollar: Diversifying FX exposure in trend following

This note summarizes the key concepts and empirical findings from “Beyond the Dollar: Diversifying FX Exposure in Trend Following” by Alberto D’Onofrio, Shangqi Han, and Alessandra Zanin. For a full description of the methodology and results, please refer to the original paper available at the end of this article.

Executive Summary

In the first half of 2025, USD-centric FX trend implementations experienced material drawdowns amid elevated dollar dominance and trend reversals linked to trade policy uncertainty.

This episode highlights a broader structural feature of currency markets: a dominant US dollar factor explains a large and time-varying share of FX return variation. Trend-following strategies implemented exclusively on USD-quoted returns therefore inherit substantial common-factor exposure, mechanically reducing effective diversification and compressing portfolio breadth, particularly during regimes of elevated dollar dominance.

Allocating part of the FX sleeve to non-USD currency crosses mitigates dollar concentration while preserving exposure to relative currency dynamics. In our empirical implementation, this increases the effective number of independent risk drivers and improves risk-adjusted performance.

As discussed in our 2025 performance commentary (D’Onofrio, 2026), incorporating currency crosses within the FX sleeve of Fulcrum’s trend-following strategy materially reduced concentration in US dollar risk entering 2025. Combined with our concentration-aware risk framework, this diversification meaningfully enhanced resilience during the first half of 2025, contributing to the Fulcrum Multi Asset Trend (MAT) strategy delivering a net return of 10.5%* for the year, significantly outperforming the Société Générale Trend Index, which returned 2.4%.

The remainder of this note develops the formal framework underlying this portfolio construction choice and its implications for systematic FX strategy design.

1. The Numeraire Problem in FX

FX returns are conventionally quoted against the US dollar. While operationally convenient, this representation embeds a structural numeraire effect: idiosyncratic currency movements are combined with a pervasive dollar component.

When the dollar becomes the dominant driver of global FX returns, USD-based implementations concentrate risk in this common factor. For systematic strategies designed to extract diversified trends across currencies, such concentration reduces effective breadth.

We quantify the magnitude of this effect using a Bayesian dynamic factor model with stochastic volatility.

2. A Factor Representation of FX Returns

2.1 Universe

We analyse a universe of 20 currencies spanning G10 and non-G10 markets. This selection mirrors the investable opportunity set used in Fulcrum’s trend-following strategies, ensuring that the results reflect realistic implementation constraints.

2.2 Econometric Framework

Let rₜ denote the vector of USD-denominated currency returns at time t. We specify the state-space model:

where fₜ represents a small set of latent common factors, Λ is the factor-loading matrix, and εₜ captures idiosyncratic return components.

Both factor and idiosyncratic innovations follow stochastic volatility processes, allowing systematic and idiosyncratic risk contributions to vary across regimes.

Estimation proceeds within a Bayesian framework with weakly informative priors, allowing the data to determine the structure while preserving economically interpretable loadings.

2.3 Factor Structure and Interpretation

We extract four latent factors that summarise the dominant sources of cross-currency co-movement. Although statistically derived, their estimated loading patterns suggest clear economic interpretations (Figure 1). We interpret them as a US dollar factor, a carry factor, a European regional factor, and an Oceania regional factor.

The USD factor displays uniformly positive loadings across currencies, confirming its pervasive influence. In contrast, the carry factor exhibits a sign structure consistent with yield differentials between high- and low-interest-rate, while regional factors capture geographic clustering beyond the dollar component.

While the dollar remains the primary common driver, these regional and carry channels generate meaningful cross-sectional differentiation.

Figure 1: Estimated Factor Loadings (Posterior Median)

Source: Fulcrum Asset Management LLP. As of 12/03/2026.

3. Quantifying Dollar Dominance

We regress individual FX returns on the estimated USD factor and interpret the resulting R² as the fraction of variance explained by broad dollar dynamics (Table 1).

For many currencies, the USD factor accounts for a substantial share of total return variance. The Japanese yen exhibits materially lower sensitivity, consistent with its potential safe-haven behaviour and idiosyncratic domestic drivers.

Augmenting the regressions with carry and regional factors materially increases explanatory power. Although the dollar factor dominates, additional systematic channels remain economically relevant and support cross-sectional diversification.

Table 1: Cumulative R² from expanding factor regressions

Source: Fulcrum Asset Management LLP. As of 12/03/2026.

4. Removing Dollar Concentration with Currency Crosses

4.1 Analytical Mechanism

Consider two USD-quoted currency returns:

Forming the cross return (x – y) effectively removes the common USD component when the two currencies have similar dollar exposures:

The resulting return reflects relative factor exposures and idiosyncratic dynamics rather than broad dollar cycles.

In practice, pairing currencies with comparable USD sensitivities enhances this attenuation effect.

4.2 Implications for Trend Following

In a USD-only framework, a standard momentum allocation’s return takes the form M(x) · x which retains embedded dollar exposure.

For a currency cross, the strategy return stream becomes M(x – y) · (x – y), such that both the momentum signal and the realised return are driven by relative price dynamics.

This construction preserves exposure to regional divergence, policy differentials, and carry regimes while reducing concentration in the global dollar movements.

5. Empirical Implementation

We compare two implementations of the FX sleeve: a baseline USD-only portfolio and a mixed portfolio incorporating a substantial allocation to non-USD crosses.

5.1 Dollar Exposure

The cross-based portfolio exhibits materially lower and more stable aggregate USD exposure through time (Figure 2), confirming the mechanical reduction in dollar sensitivity achieved by neutralising the common numeraire. While the USD-only portfolio remains highly sensitive to broad dollar cycles, the mixed implementation successfully redistributes risk across multiple currency pairs.

Following the 2024 US Presidential election, Figure 2 shows that USD-only trend strategies built considerably larger aggregate USD exposure than the cross-based implementation. This higher exposure, in turn, led to larger whipsaw losses around tariff announcements.

Figure 2: US Dollar Exposure (%)

Source: Fulcrum Asset Management LLP. As of 12/03/2026.

5.2 Risk-Return Characteristics (Last Five Years)

Table 2: Summary statistics for FX portfolios (Last 5 Years)

Source: Fulcrum Asset Management LLP. As of 12/03/2026.

Empirical results show that, in the last five years, the cross-based implementation delivered higher returns at comparable volatility, with improved drawdown characteristics and lower realised tail risk.

5.3 Effective Breadth

We quantify diversification using the squared diversification benefit measure:

This statistic approximates the effective number of independent risk drivers within the portfolio, thereby measuring its “effective breadth.”

Our results demonstrate a clear structural advantage for the cross-based approach. The USD-only portfolio captures approximately 3.4 independent risk drivers, whereas the cross-based portfolio captures approximately 7.6 independent risk drivers. Incorporating crosses therefore more than doubles effective breadth, materially reducing reliance on any single systematic driver.

6. Conclusion

USD-quoted FX returns embed substantial exposure to a dominant dollar factor. When this factor strengthens, systematic currency strategies implemented purely in USD terms become structurally concentrated.

Currency crosses provide a transparent and robust mechanism to reduce unintended dollar exposure while preserving exposure to relative macroeconomic and carry dynamics. Empirically, incorporating crosses contributed to the Fulcrum MAT strategy delivering a net return of 10.5%*, significantly outperforming the Société Générale Trend Index, which returned 2.4%.

The benefits are most pronounced during regimes of elevated dollar dominance, when effective diversification in USD-only portfolios is mechanically compressed. For diversified trend-following strategies, this represents a structural enhancement in FX portfolio construction rather than a tactical overlay.