Beta

April 11, 2026 Risk Management

Beta estimates how sensitive an investment's returns have been to a selected market benchmark, but it does not measure total investment risk.

On this page

Beta estimates how sensitive an asset or portfolio’s returns have been to movements in a selected market benchmark. A beta of 1.2 means the fitted relationship has historically associated a 1% benchmark move with an approximately 1.2% move in the asset, before considering the model’s intercept and residual return.

Beta is a relative measure of market exposure, not a forecast, a measure of maximum loss, or a complete measure of investment risk.

Key Takeaways

Beta is the covariance of an asset’s returns with benchmark returns divided by the variance of benchmark returns.
The benchmark, return frequency, lookback window, currency, and estimation method can materially change the result.
Beta measures fitted sensitivity to one selected factor. It does not capture all company-specific, liquidity, credit, valuation, or tail risk.
A beta above 1 indicates greater estimated sensitivity than the benchmark; it does not guarantee outperformance or a specific future move.
Beta is commonly used in portfolio analysis and the capital asset pricing model (CAPM).

Beta Formula

For asset \(i\) and benchmark \(m\):

\beta_i = \frac{\operatorname{Cov}(R_i,R_m)} {\operatorname{Var}(R_m)}

where:

\(R_i\) is the asset’s periodic return
\(R_m\) is the benchmark’s periodic return
covariance measures how the two return series move together
benchmark variance scales that co-movement by the benchmark’s own dispersion

In a single-factor return regression, beta is the fitted slope:

R_i - R_f = \alpha_i + \beta_i(R_m-R_f) + \varepsilon_i

Analysts may use total returns or excess returns depending on the model. The method, benchmark, and return definition should be stated.

Chart showing asset return versus market return with lines for beta 0.6, beta 1.0, and beta 1.4.

A steeper fitted slope indicates greater estimated sensitivity to the selected benchmark.

How to Interpret Beta

Estimated beta	Typical interpretation
Above 1	The asset has shown amplified sensitivity to benchmark moves
Near 1	The asset has shown sensitivity broadly similar to the benchmark
Between 0 and 1	The asset has tended to move in the benchmark’s direction, but by less
Near 0	The fitted linear relationship shows little benchmark sensitivity
Below 0	The asset has tended to move opposite the benchmark in the sample

These are model interpretations, not promises. A beta near zero can coexist with substantial standalone volatility, and a negative estimate may be unstable or specific to one period.

Worked Example

Assume a stock has an estimated beta of 1.3 relative to a broad equity index. If the index rises 2%, the beta component of the fitted model is:

1.3 \times 2\% = 2.6\%

This does not predict that the stock will rise exactly 2.6%. Its realized return also reflects the intercept, company-specific information, estimation error, and risks not represented by the selected benchmark.

The same caution applies in a falling market. Beta describes a symmetric linear exposure unless the analyst models different up-market and down-market behavior.

Beta, Correlation, and Total Volatility

Beta and correlation are related but answer different questions:

\beta_i = \rho_{i,m} \frac{\sigma_i}{\sigma_m}

A high beta can result from high correlation, high asset volatility relative to the benchmark, or both.

Measure	What it describes	What it does not show
Beta	Fitted sensitivity to a selected benchmark	Total standalone risk
Correlation	Direction and strength of standardized co-movement	Relative move size
Standard deviation	Total return dispersion in the sample	Source of that dispersion
R-squared	Share of sample variation explained by the fitted model	Whether beta will remain stable
Alpha	Regression intercept or model-relative return estimate	A guaranteed skill or future excess return

Beta should be read with the regression’s R-squared, standard error, sample period, and residual analysis. A precise-looking point estimate can be unreliable when the relationship is weak.

Benchmark and Data Choices

Before comparing beta estimates, align:

benchmark index or factor
total-return versus price-return series
local-currency versus base-currency returns
daily, weekly, or monthly frequency
start and end dates
treatment of dividends, fees, and corporate actions
use of raw returns versus excess returns
estimation method and outlier handling

A U.S. stock can have different betas relative to a domestic equity index, a global equity index, or an industry index. None is universally correct; the useful benchmark is the one relevant to the analysis.

Thin trading and stale prices can depress measured covariance. A short lookback may produce an unstable estimate, while a long lookback may mix obsolete business models, leverage, or market regimes.

Portfolio Beta

For a linear portfolio measured against one benchmark, portfolio beta can be estimated as the weighted sum of position betas:

\beta_p = \sum_{i=1}^{n} w_i\beta_i

This is most useful when weights and beta estimates use consistent data and definitions. Options, leverage, short positions, rebalancing, and other nonlinear exposures can make beta change as prices move. Scenario analysis and factor-based risk models may then be more informative than a single static number.

Beta in CAPM

CAPM uses beta to connect expected return with exposure to market risk:

E(R_i) = R_f + \beta_i\left(E(R_m)-R_f\right)

The model assumes that market beta is the relevant priced risk. In practice, realized returns can reflect multiple factors, changing exposures, estimation error, taxes, liquidity, and market frictions. CAPM is therefore a model-based framework, not proof that an asset with higher beta will earn a higher realized return.

Risks and Limitations

Historical estimate: past co-movement may not continue.
Benchmark dependence: a different benchmark can produce a different beta.
Model dependence: a single linear factor may omit important exposures.
Instability: leverage, operations, capital structure, and market regimes can change beta.
Tail blindness: beta does not describe maximum loss, drawdown, skew, or loss severity beyond a threshold.
Path blindness: the order and timing of returns are not captured.
Liquidity blindness: a low-beta asset can still be difficult or expensive to sell.
False precision: estimates should be reviewed with uncertainty statistics and data quality.

Common Mistakes

Treating beta as total volatility.
Calling beta above 1 evidence that an asset will outperform.
Comparing betas calculated against different benchmarks or periods.
Applying an equity beta to a materially changed capital structure without review.
Ignoring low R-squared or a wide confidence interval.
Assuming beta remains constant in stressed markets.
Confusing market beta with the beta probability distribution or a Type II statistical error.

Authoritative Context

Capital Asset Pricing Model: Uses beta to estimate a model-based required return for market risk.
Systematic Risk: The broad market-related risk that beta is intended to represent.
Correlation: Standardizes co-movement but does not incorporate the relative volatility used in beta.
Covariance: Forms the numerator of the conventional beta calculation.
Standard Deviation: Measures total return dispersion rather than benchmark-relative sensitivity.

FAQs

Is a beta below 1 a low-risk investment?

Not necessarily. It indicates lower estimated sensitivity to the selected benchmark, but the investment may still have substantial company-specific, credit, liquidity, valuation, or tail risk.

Can beta be negative?

Yes. A negative estimate means the asset and benchmark moved in opposite directions in the fitted sample. Negative betas are uncommon and may not remain stable.

Does beta predict the next market move?

No. Beta estimates sensitivity conditional on benchmark movement; it does not predict the direction of the benchmark or the asset’s exact return.

Educational Use

This article provides general financial education. It is not personalized investment, trading, portfolio-construction, valuation, statistical, tax, or legal advice.

Expected Monetary Value