Beta
What is Beta?
Beta measures the sensitivity of an investment's returns to movements in a specified benchmark within a defined statistical model.
Historical beta should use synchronized total returns, a stated frequency and rolling window, and enough observations. Currency, stale prices, leverage, nonlinear options, and changing exposures can distort it. A single beta should not be used to predict loss under gaps or regimes outside the estimation sample.
Downside beta, rolling estimates, and stress scenarios can reveal asymmetry hidden by a full-period linear average.
Portfolio controls should compare returns-based beta with holdings-based market exposure and investigate material differences. The result can reveal stale prices, nonlinear instruments, changing leverage, benchmark mismatch, short history, or exposures that the chosen linear factor does not represent.
Calculation
Beta is the covariance of portfolio and benchmark returns divided by benchmark variance, or the slope in a regression. A beta of 1 indicates one-for-one average sensitivity, above 1 greater sensitivity, and below 1 lower sensitivity. Negative beta indicates average movement in the opposite direction over the sample.
Interpretation
Beta describes historical linear co-movement, not a guaranteed response to the next market move. A portfolio with beta 1.2 is not certain to gain 12% when the market gains 10%. Alpha, idiosyncratic return, changing exposures, and noise also contribute. Beta is always relative to a named benchmark and period.
Estimation choices
Daily, weekly, and monthly returns can produce different estimates. Lookback, weighting, currency, total-return series, nonsynchronous trading, outliers, and stale marks matter. Confidence intervals reveal uncertainty. Rolling beta can show change, but short windows are noisy and can encourage interpretation of random movement as structural exposure.
Portfolio uses
Beta supports hedging, risk budgeting, expected-return models, manager analysis, and scenario approximation. Portfolio beta can be estimated from holdings or returns, with different strengths. Derivatives and leverage can change sensitivity quickly. Multivariate models separate market beta from sector, style, rate, currency, and other exposures.
Limitations
Options, dynamic strategies, liquidity shocks, and downside asymmetry violate a stable linear relationship. Correlations and volatility can rise in stress. Low beta does not mean low total risk, because concentration, credit, fraud, or illiquidity can dominate. Pair beta with volatility, drawdown, scenarios, and look-through exposure.
Sources and further reading
- Risk Management: An Introduction, CFA Institute
- Portfolio Performance Evaluation, CFA Institute