How AI enters the analysis
Models that read currency as a signal, not noise
Traditional portfolio tools often treat currency exposure as a residual — what is left over once the asset decision is made. The editorial line here is different. We write about machine-learning approaches that take the currency leg seriously: clustering regimes in foreign-exchange volatility, flagging when a hedge ratio should move, and surfacing the currency pairs that drive a portfolio's reported return.
The articles do not recommend trades. They describe the method — how a model is trained on historical carry and volatility, what features it weighs, and where it tends to fail — so a reader can judge the approach rather than imitate it.