What is correlation?

Correlation measures the direction and strength of association between two data series. The widely used Pearson coefficient expresses linear association from −1 to +1. Positive values indicate a same-direction relationship; negative values indicate an inverse relationship. A value near zero suggests little linear association, not necessarily the absence of every possible relationship.

Example

Across three hypothetical days, asset A returns +1%, −1% and +2%, while asset B returns +2%, −2% and +4%. Every B return is twice the corresponding A return, so Pearson correlation equals +1 in this small sample. That does not imply the pattern will continue tomorrow or that A caused B to move. Three observations are insufficient for a reliable market forecast.

The coefficient is not a profit probability: a correlation of 0.8 does not mean an 80% chance of a rise. The selected period, sampling frequency, use of prices versus returns, and outliers can change the result.

Two assets may move together because both respond to the same economic event. A causal claim needs a mechanism and evidence distinguishing alternative explanations. Choosing only the window that fits the story does not make a sound comparison.

The economic calendar guide separates an observation from an interpretation when a price move overlaps with a release. Our Fed rates and crypto analysis examines possible transmission channels alongside their causal limits.

The Bitcoin–dollar relationship analysis provides a concrete example of separating a coefficient’s sign from its magnitude. The dated research values are not presented as current market readings.

The ETF analysis comparing two IBIT quarters illustrates the limits of equating fund demand with price direction: positive net share transactions accompany negative NAV returns. Two periods do not establish a reliable correlation or causal conclusion.