Research preview

The Chinese yuan has surged to its strongest level against the U.S. dollar since mid‑2022, prompting traders to question whether traditional models can explain the move. This article shows how high‑dimensional regressions, regularisation, and debiased machine‑learning estimators can be combined to extract signal from a flood of macro‑financial predictors and to build more robust trading signals around currency dynamics. The Challenge of High‑Dimensional Macro Data Macro‑financial markets generate thousands of potential predictors: interest‑rate spreads, commodity inventories, trade balances, capital‑flow statistics, and sentiment indices derived from news or social media. When the number of predictors (p) approaches or exceeds the number of observations (n), ordinary least‑squares estimates become unstable....