Skip to content
Toro FinanceCalculator Suite
Back to Calculator HubMethod: CAPM (Jensen 1968)
Risk-Adjusted Return Analysis

Jensen's Alpha & Risk-Adjusted Return Calculator

A fund that returned 15% in a year the market returned 10% has not necessarily beaten the market. If it carried more risk, most of that gap was simply market exposure. Jensen's alpha subtracts what beta alone should have delivered and shows what is left.

Inputs

Realised annual returns over the same period, plus the asset's beta

%
%
%

Beta is the slope of the asset's returns regressed on the benchmark's. If you have both price series, the shape tool below computes it for you.

Jensen's Alpha

+3.2%

Return beyond what beta explains

CAPM Expected Return

11.8%

4% + 1.3 × (10% − 4%)

Return From Market Exposure

7.8%

The part beta alone accounts for

Treynor Ratio

8.462

Excess return per unit of beta

Reading This Result

The asset beat what its market exposure alone would predict by 3.2 points a year. Over a short sample this is as likely to be luck as skill.

α = 15% − [4% + 1.3 × (10% − 4%)] = 3.2%

Alpha measured over a single period is dominated by noise. Academic work on fund performance generally finds that very few managers sustain positive alpha once costs are deducted, and that past alpha is a weak predictor of future alpha.

Shape Similarity (Dynamic Time Warping)

A separate question: how closely does one price path track another, allowing for leads and lags? DTW answers that. It is a shape metric, not a return metric — which is why it is computed here rather than folded into the alpha figure above.

Comma or newline separated. Series are z-normalised before comparison, so the price level does not matter — only the shape.

DTW Similarity

99/100

Normalised distance 0.024

Warping Path Length

12 steps

Total cost 0.282

Fitted Beta

1.308

11 return observations

R-Squared

0.986

Variance explained by the benchmark

Portfolio Theory9 min read

Jensen's Alpha: Separating Skill From Market Exposure

The single most common mistake in performance analysis is treating a return above the index as evidence of skill. Most of the time it is evidence of leverage.

1. The Beta Mirage

Suppose the market returns 10%, cash returns 4%, and a fund returns 15%. It looks like the fund beat the market by five points. But if the fund runs a beta of 1.5, the Capital Asset Pricing Model says it should have returned 4% + 1.5 × (10% − 4%) = 13% purely from carrying more market risk. The genuine outperformance is 2%, not 5%.

Run the numbers again with a beta of 1.83 and the expected return becomes 15% exactly: the entire apparent edge disappears. The fund did not pick better assets, it simply held a more volatile version of the same market — and in a downturn that same beta works in reverse.

Jensen's Alpha (Jensen, 1968)CAPM
αp = Rp − [ Rf + βp(Rm − Rf) ]

Rp = realised return of the asset or portfolio.

Rf = risk-free rate over the same period.

Rm = realised return of the benchmark.

βp = OLS slope of the asset's returns on the benchmark's.

Units: percentage points per year. Every term is a return, which is what makes the subtraction meaningful.

2. What Dynamic Time Warping Is For

DTW finds the lowest-cost alignment between two sequences by stretching the time axis, using the dynamic-programming recurrence below. It is genuinely useful for pattern matching: correlation demands that two series move on the same day, whereas DTW recognises the same shape arriving early or late.

DTW Recurrence (Sakoe & Chiba, 1978)Shape Metric
γ(i, j) = d(xi, yj) + min[ γ(i−1, j), γ(i, j−1), γ(i−1, j−1) ]

γ(i, j) = cumulative minimal distance to grid point (i, j).

d(xi, yj) = local distance between z-normalised points.

The tool above computes this matrix in full, with a Sakoe-Chiba band limiting how far the alignment may drift.

Why DTW is not part of the alpha formula

A DTW distance is a dimensionless measure of shape dissimilarity. An alpha is a return, measured in percentage points per year. Subtracting one from the other is a unit error — the result is not a return, not a distance, and not interpretable as either. Any screen that mixes an oscillator reading, a warping distance and a beta into one number and calls the output “alpha” is producing a number with no defined meaning. The two quantities are kept in separate panels above deliberately.

3. The Limits of Alpha

Alpha is a description of the past, computed with hindsight. Its weaknesses are well documented: it depends entirely on the benchmark chosen, a single period carries enormous sampling noise, and CAPM's single-factor view is known to be incomplete — multi-factor models attribute much of what looks like alpha to size, value, momentum and quality exposures instead.

Treat a positive alpha as a question worth investigating rather than an answer. Over short samples it is more likely to reflect luck than skill.

Frequently Asked Questions & Quantitative Reference

There is no threshold that reliably separates skill from luck, and be sceptical of anyone who offers one. What matters far more than the size of an alpha is its statistical significance: how many periods it was measured over, how volatile the excess returns were, and whether it survives after fees, trading costs and taxes. A 2% alpha over three years is well within the range of chance.
Disclaimer

Educational use only. These are backward-looking diagnostic measures computed from figures you supply. They are not investment recommendations, not a screen for selecting securities, and not a forecast. Alpha computed over a short period is dominated by chance. Consult a licensed financial adviser before making investment decisions.