Beta is a coefficient measuring an asset’s return sensitivity to a specified benchmark. It is commonly estimated from historical observations as the covariance of asset and benchmark returns divided by the variance of benchmark returns. Equivalently, it is the slope of a linear regression using those return series.
Reading the coefficient
A beta of 1 indicates unit sensitivity; a positive beta above 1 indicates greater sensitivity, and a value between 0 and 1 indicates less. With beta of 1.5, a one-percentage-point change in benchmark return corresponds to a 1.5-percentage-point change in the fitted asset return, holding the model’s other terms constant. Actual returns can differ substantially.
Negative beta indicates an inverse historical relationship. A beta near zero means little estimated linear sensitivity, not an absence of risk.
What beta leaves out
Beta concerns market exposure, not total volatility. An asset can have low beta yet large price swings from its own specific risks. Beta is also different from correlation.
Results depend on the benchmark, observation frequency, sample window, and currency. A token’s beta against Bitcoin need not match its beta against a broad crypto index. Historical relationships can change, so beta does not guarantee future movements or returns.