BEYOND CORRELATION: ENHANCING CURRENCY PORTFOLIOCONSTRUCTION THROUGH KENDALL’S TAU ANDCORRESPONDENCE ANALYSIS

Abstract

Traditional correlation metrics such as Pearson and Spearman are widely employed in financial portfolio construction, yet they exhibit critical limitations. Pear- son correlation, being linear and sensitive to scale and volatility, often overstates de- pendency during high-variance periods, while Spearman correlation fails to capture non-monotonic relationships. In this paper, I advocate a shift from classical cor- relation analysis to a correspondence-based perspective, employing Kendall’s Tau as a more robust and semantically consistent measure of dependency. I demonstrate, through a portfolio optimization framework, that Kendall correlation yields superior results in terms of return maximization and drawdown minimization, particularly under conditions of structural market changes and nonlinear dependencies. Empirical results on historical asset data reveal that portfolios constructed using Kendall-based correspondence measures exhibit enhanced stability and riskadjusted performance compared to those based on conventional correlations. This work highlights the importance of re-evaluating correlation paradigms in favor of more meaningful statistical dependence structures in financial modeling.