Intermarket context
The idea
Correlation measures co-movement over a selected sample and frequency. It does not establish causation. Several positions can share one underlying risk even when the symbols differ. Changes in regime and stressed liquidity can make historical relationships unreliable.
A worked example
Long EUR/USD and long GBP/USD both carry exposure to a weaker dollar. Two separately small positions can form one larger shared bet.
Try it yourself
Write the economic risk behind each position, then compare it with the numerical correlation estimate.
A mistake to avoid
Adding positions because the pair names differ while ignoring the common currency exposure.