FXForex basics

Currency correlation

Also calledFX correlation

Currency correlation is the statistical tendency for the returns of two currency pairs, or of two currencies against a common reference, to move in the same direction, opposite directions, or with little consistent relationship over a selected period.

What Currency correlation means

Correlation is usually expressed on a scale from -1 to +1. A positive reading means the two observed price series generally rose and fell together; a negative reading means they generally moved in opposite directions. A reading near zero means there was little linear relationship in that sample. Traders commonly calculate it from percentage returns rather than price levels, because exchange rates have different quotation conventions and starting values.

Currency correlation helps show whether several positions are effectively concentrated in a similar market exposure. For example, two long positions can both lose when the same currency weakens, even if they are in different pairs. The statistic is descriptive, not a guarantee: relationships can change with the time period, market conditions, monetary policy, and the currencies used in each pair.

Assume daily returns for EUR/USD and GBP/USD have a 60-day correlation of +0.82. Over that historical sample, the pairs tended to move in the same direction fairly strongly. Holding one long position in each pair may therefore provide less diversification than holding positions whose returns have a lower or negative correlation. The +0.82 does not predict tomorrow’s moves.

Common questions

What does a negative currency correlation mean?+

A negative correlation means the two return series tended to move in opposite directions during the measured period. It does not mean that every daily move was opposite, nor that the relationship will continue in future periods.

Is a correlation of zero proof that two currency pairs are independent?+

No. A correlation near zero only indicates little linear relationship in the chosen data sample. The pairs can still respond to the same event at different times or display a nonlinear relationship.

Go to the original material.

01NIST Engineering Statistics Handbook — correlation02NIST — correlation coefficient calculation and interpretation