Jensen's Alpha

August 31, 2024 Portfolio Management

CAPM-based performance measure comparing portfolio return with the return implied by its estimated market beta.

On this page

Jensen’s Alpha measures portfolio return above or below the return implied by the Capital Asset Pricing Model (CAPM), given the portfolio’s estimated market beta. It isolates return not explained by the model’s risk-free rate and market-risk exposure, but it does not prove manager skill or adjust for risks omitted from CAPM.

Key Takeaways

Jensen’s alpha is the difference between realized return and CAPM-implied return.
The result depends on the risk-free rate, market benchmark, beta estimate, period, and fee basis.
Positive alpha means outperformance relative to the model, not guaranteed repeatable skill.
CAPM uses one market factor, so style, sector, credit, duration, currency, liquidity, and nonlinear exposures can remain unexplained.

Formula

For a single measurement period:

\alpha_p=R_p-\left[R_f+\beta_p(R_m-R_f)\right]

Where:

(R_p) is portfolio return
(R_f) is the matching risk-free rate
(R_m) is the selected market-benchmark return
(\beta_p) is portfolio beta relative to that market benchmark

The bracketed term is the return CAPM predicts for the portfolio’s estimated systematic risk.

Worked Example

Suppose a portfolio returned 12.0%, the risk-free rate was 3.0%, the market benchmark returned 10.0%, and portfolio beta was 1.10.

First calculate the market risk premium:

R_m-R_f=10.0\%-3.0\%=7.0\%

Then calculate the CAPM-implied portfolio return:

R_f+\beta_p(R_m-R_f)=3.0\%+1.10(7.0\%)=10.7\%

Finally:

\alpha_p=12.0\%-10.7\%=1.3\%

The portfolio exceeded its CAPM-implied return by 1.3 percentage points for this period. That is a model-relative result, not evidence that the same alpha will recur.

Single-Period vs. Regression Alpha

Analysts often estimate Jensen’s alpha over many observations with a regression:

R_{p,t}-R_{f,t}=\alpha_p+\beta_p(R_{m,t}-R_{f,t})+\varepsilon_t

In this form, alpha is the estimated intercept and (\varepsilon_t) is unexplained residual return. A time-series estimate allows analysts to review standard errors, confidence intervals, residual behavior, and goodness of fit. It still inherits the limitations of the model and data.

Approach	Output	Main use	Main caution
Single-period calculation	Actual return minus CAPM-implied return	Illustrating the mechanics for one period	One outcome provides little evidence about persistence
Time-series regression	Estimated intercept across observations	Evaluating average model-relative performance	Sensitive to benchmark, sample, changing beta, and regression assumptions
Multifactor regression	Intercept after several specified factors	Separating return from multiple systematic exposures	Results depend on factor choice and implementation

How to Interpret the Result

A positive Jensen’s alpha means return exceeded the CAPM estimate. A negative result means it fell short. The size should generally be reported in the same periodic units as the input returns and clearly labeled if annualized.

For a defensible comparison, use:

the same suitable market benchmark
aligned portfolio, market, and risk-free returns
the same gross- or net-of-fee basis
comparable samples and data frequencies
a beta estimated over a relevant period

Risks and Limitations

Benchmark sensitivity

Beta and alpha are estimated relative to the selected market index. An unsuitable benchmark can make the risk adjustment economically meaningless.

CAPM is a one-factor model

Apparent alpha may compensate for exposures CAPM does not represent. Multifactor analysis can test whether size, value, momentum, duration, credit, currency, or other systematic exposures explain the result.

Beta can be unstable

Portfolio holdings, leverage, and market relationships change. A constant-beta estimate can misstate expected return when exposure varies through time or behaves nonlinearly.

Positive alpha may be noise

Analysts should review estimation uncertainty, sample length, survivorship, backtest selection, and multiple testing. A positive point estimate alone does not establish persistent skill.

Jensen’s Alpha vs. Nearby Measures

Measure	Comparison	Risk input	Output
Jensen’s alpha	CAPM-implied return	Beta	Percentage-point value added
Active return	Benchmark return	None by itself	Percentage-point outperformance
Treynor Ratio	Risk-free return	Beta	Excess return per unit of beta
Information Ratio	Benchmark return	Tracking error	Active return per unit of active risk

Alpha: Covers both simple benchmark-relative and model-based alpha conventions.
Capital Asset Pricing Model: Supplies the expected-return relationship used by Jensen’s alpha.
Beta: Measures market sensitivity in the CAPM expected-return calculation.
Treynor Ratio: Uses beta as a ratio denominator rather than calculating model-relative value added.
Information Ratio: Measures benchmark-relative return efficiency using tracking error.

Sources

FAQs

What does a negative Jensen's Alpha mean?

It means the portfolio returned less than CAPM predicted for its estimated beta during the period. It does not identify the cause without further attribution analysis.

Is Jensen's Alpha the same as benchmark outperformance?

No. Benchmark outperformance simply subtracts benchmark return. Jensen’s alpha subtracts a CAPM-implied return that incorporates the risk-free rate and estimated beta.

Does statistically significant alpha guarantee future outperformance?

No. Statistical significance is conditional on the model, sample, and assumptions. Future exposures and market conditions can differ, and the model may omit relevant risks.

This page is for financial education and does not recommend a manager, fund, benchmark, or strategy.

Information Ratio (IR)

Tracking Error