Please rotate your device to landscape mode to view the charts.

Optimizing Currency Factors

Journal: Financial Review

Date: 2025

Author: Minyou Fan, Fearghal Kearney, Youwei Li, Jiadong Liu

Abstract:
We introduce a novel framework that dynamically optimizes currency factor strategies via trading currency spot and forward. We examine the performance of 24,336 portfolio optimization approaches and find that the optimized currency factors significantly outperform the naïve factors after correcting for data snooping bias. Our framework suits both symmetric factor portfolios, including carry, momentum, and value, and asymmetric factor portfolios, such as time series momentum and return signal momentum. An out-of-sample procedure that aggregates all the outperforming optimization approaches validates the economicsignificance of our optimized factor portfolio.

Link: Google Scholar

Background and Context

Currency Market Advantage

Unlike equity markets where portfolio optimization often fails, currency markets have unique characteristics that make mean-variance optimization consistently profitable.

Comprehensive Testing Framework

The study examines 24,336 different portfolio optimization approaches across three major currency factors: carry, momentum, and value.

Robust Statistical Methods

Data snooping bias tests ensure that superior performance results from genuine optimization merit rather than random chance.

Comprehensive Testing of 24,336 Optimization Combinations

24,336 Optimization Approaches 12 Covariance Estimators Ledoit-Wolf, Bayes-Stein Sample, EWMA, etc. 4 Expected Return Estimators Sample, EWMA Bayes-Stein, Implied 7 Optimization Methods Mean-Variance, GMV Maximum Decorrelation × × 24,336 Combinations Per Currency Factor
  • Researchers systematically tested every possible combination of estimation methods and optimization approaches available.
  • This comprehensive approach ensures no potentially profitable optimization strategy was overlooked in the analysis.
  • The scale demonstrates the thoroughness required to identify truly superior portfolio construction methods.

Long-Short Portfolio Structure Maximizes Factor Exposure

Long Portfolio High Carry/Momentum Low Value Currencies Weight = +1 Short Portfolio Low Carry/Momentum High Value Currencies Weight = -1 − Factor Portfolio Return Long Return - Short Return
  • Factor portfolios capture return differences by simultaneously buying attractive and selling unattractive currencies.
  • This long-short structure isolates the factor premium while maintaining dollar neutrality for investors.
  • Optimization separately improves weights for both long and short legs to maximize overall performance.

Optimized Strategies Deliver 7-10% Higher Annual Returns

  • Optimized currency factors significantly outperform traditional equal-weighted approaches across all three major factors.
  • Even the previously unprofitable value factor becomes highly profitable when properly optimized using advanced techniques.
  • The magnitude of outperformance demonstrates substantial economic value from applying sophisticated optimization to currency trading.

Statistical Significance Confirmed Despite Extensive Testing

  • Data snooping bias tests confirm that superior performance results from genuine optimization merit rather chance.
  • Thousands of portfolios remain statistically significant even after adjusting for multiple hypothesis testing concerns.
  • These rigorous statistical controls provide confidence that the results are economically meaningful and reliable.

Mean-Variance Optimization Consistently Dominates Other Approaches

  • Mean-variance optimization delivers the highest risk-adjusted returns across all currency factors and portfolio positions.
  • The success stems from currency markets having smaller cross-sections and lower estimation errors than equity markets.
  • These findings contradict the conventional wisdom that mean-variance optimization fails in practical investment applications.

Contribution and Implications

  • First comprehensive study to apply modern portfolio optimization techniques to currency factor investing strategies.
  • Provides practical guidance for currency fund managers on selecting optimal portfolio construction methods and parameters.
  • Demonstrates that sophisticated optimization can generate substantial economic value in foreign exchange markets unlike equity markets.
  • Establishes robust statistical framework for evaluating trading strategies while controlling for data snooping bias concerns.

Data Sources

  • Performance comparison chart constructed using data from Table 2 comparing naive vs optimized factor returns.
  • Data snooping results chart based on Table 4 showing number of significant portfolios after bias testing.
  • Optimization method comparison chart derived from Figure 2 data on Sharpe ratios by approach.