Portfolio and risk · Portfolio theory
Capital asset pricing model (CAPM): what return a beta should earn
The CAPM formula, the security market line and a test on 14 funds from 2005 to 2026: beta explained some of the differences in return, and less than the model says.
On this page
The capital asset pricing model, or CAPM, is a formula for the return an investment should be expected to earn, given one number about it: its beta. It says that expected return equals the risk-free rate plus beta times the market's premium over that rate. An asset with twice the market's sensitivity should earn twice the premium. An asset with none should earn the risk-free rate, however much it bounces around.
William Sharpe and John Lintner derived it in the mid-1960s by asking what would follow if every investor used Modern portfolio theory: why the mix matters more than the parts. It became the standard tool for estimating a company's cost of equity and the benchmark against which fund managers are judged, and it remains both, although the evidence for it is mixed. This lesson states the model, tests its prediction on 14 funds, and explains what use survives the test.
The idea
Modern portfolio theory: why the mix matters more than the parts shows that part of any asset's risk disappears when it is combined with others. CAPM draws the consequence. If risk can be removed free of charge by diversifying, the market has no reason to pay investors for carrying it. The only risk that should earn a premium is the part that remains in a fully diversified portfolio: the tendency to rise and fall with the market as a whole. Beta measures exactly that.
So the model predicts a straight line. Put beta on the horizontal axis and expected return on the vertical. Treasury bills sit at beta 0, the market at beta 1, and every other asset on the line through them.
The figure shows the line and what 14 funds actually did. If the model held exactly, every point would be on the line. They are scattered around it, and not at random: the points on the left tend to be above it and several on the right below.
How it works
Three inputs go in.
- The risk-free rate. A Treasury yield. Short-term bills for a one-period model; in company valuation, usually the 10-year yield.
- Beta. Estimated by regressing the asset's past returns on the market's.
- The market risk premium. A forecast of how much the market will earn above the risk-free rate. Nobody knows it; estimates between 4% and 6% a year are typical.
A worked example
Suppose the 10-year Treasury yields 4%, the market risk premium is assumed to be 5%, and a stock has a beta of 1.3. These are illustrative figures.
- Premium for this stock: 1.3 × 5% = 6.5%.
- Expected return: 4% + 6.5% = 10.5% a year.
A company's finance department would call 10.5% its cost of equity: the return shareholders require, and so the minimum a new project financed by them must earn. A stock with a beta of 0.6 under the same assumptions would have a required return of 4% + 3% = 7%.
The same formula run backwards defines Alpha: the return that market exposure does not explain. After the fact, compare what a fund earned with what its beta predicted. The difference is alpha.
Testing the prediction on 14 funds
The data: monthly total returns from January 2005 to October 2026, with SPY as the market. Over the period SPY earned an average of 9.6% a year above Treasury bills, so CAPM predicts each fund's excess return as its beta times 9.6%.
| Fund | Beta | Predicted return above bills | Actual | Difference (alpha) |
|---|---|---|---|---|
| XLK (technology) | 1.13 | 10.8% | 15.5% | +4.7% |
| QQQ (Nasdaq-100) | 1.13 | 10.8% | 13.8% | +3.0% |
| SPY (the market) | 1.00 | 9.6% | 9.6% | 0.0% |
| RSP (equal weight) | 1.08 | 10.3% | 9.5% | −0.8% |
| VTV (large value) | 0.93 | 8.9% | 8.5% | −0.4% |
| IWM (small companies) | 1.19 | 11.3% | 8.2% | −3.1% |
| XLP (consumer staples) | 0.56 | 5.4% | 7.8% | +2.4% |
| EEM (emerging markets) | 1.04 | 9.9% | 7.4% | −2.6% |
| VNQ (real estate) | 1.07 | 10.2% | 7.1% | −3.1% |
| EFA (developed ex-US) | 0.96 | 9.2% | 5.5% | −3.7% |
| GLD (gold) | 0.10 | 1.0% | 9.3% | +8.4% |
| TLT (long Treasuries) | −0.08 | −0.7% | 2.1% | +2.8% |
| IEF (7–10 year Treasuries) | −0.03 | −0.3% | 1.5% | +1.8% |
| AGG (US bonds) | 0.07 | 0.7% | 1.2% | +0.5% |
Returns are after fund expenses; taxes and trading costs are not included.
Three things stand out.
Beta did carry information. Funds with betas near zero (the bond funds) earned 1% to 2% above bills. Funds with betas near 1 earned 5% to 15%. A line fitted through all 14 points explains half of the variation in their returns.
The fitted line was too flat. CAPM's line starts at 0 and rises 9.6 points per unit of beta. The line that best fits these 14 funds starts at 3.3% and rises only 6.0 points. Low-beta assets earned more than the model allows and the average high-beta asset less. Among the ten stock funds alone, beta explained only 15% of the differences in return.
The big misses have other explanations. The highest-beta fund in the table, IWM, earned less than the market. Consumer staples, with a beta of 0.56, earned nearly as much as funds with twice the beta. And gold earned 9.3% above bills with a beta of 0.10: the model has nothing to say about an asset whose return is unrelated to the stock market.
The flat line is not peculiar to this sample. Black, Jensen and Scholes reported the same shape in 1972, using US stocks from 1931 to 1965, and Fama and French's 2004 review concluded that the model's empirical record was poor enough to invalidate most of its applications. The observation that low-beta stocks earn more than they "should" is the basis of Low-volatility investing: the rules, and USMV against SPY.
Using it
Despite that record, CAPM is used every day, for three things.
- Cost of equity. Company valuation needs a discount rate, and CAPM gives a defensible one from public inputs. Analysts know it is rough and usually test a range.
- A first benchmark for performance. Before crediting a manager with skill, subtract what the fund's beta would have earned. What remains is alpha. The table above shows why this is only a first step: XLK's alpha of 4.7% is a sector's good fortune, not stock-picking.
- A way of thinking. The model's central claim, that diversifiable risk is not rewarded, is a useful default. Holding one volatile stock in place of a diversified fund adds risk for which no extra return should be expected.
The failures of the one-beta model led to models with several. Factor models: explaining a portfolio’s returns with several betas adds further sources of common risk, such as company size and value, and explains some of the misses in the table.
Where it fails or misleads
- The security market line is flatter in the data than in the model. Low-beta assets have earned more, and high-beta assets less, than predicted.
- Its assumptions are strong. All investors hold the same beliefs, care only about mean and variance over one period, and can borrow and lend freely at the risk-free rate. None of these is true. Fischer Black showed that restricting borrowing alone is enough to flatten the line.
- "The market" cannot be observed. In theory it is every risky asset in the world: stocks, bonds, property, private businesses. In practice it is a stock index. Richard Roll argued in 1977 that the model is therefore untestable.
- Beta is an estimate. It varies with the period, the return interval and the index used, and it drifts over time.
- The market risk premium is a guess. Moving it from 4% to 6% changes the cost of equity of a beta-1.3 stock by 2.6 points, enough to change most valuation conclusions.
- One factor is not enough. Assets whose returns come from sources unrelated to the stock market (gold, Treasuries, currencies) are outside its reach.
Compared with its neighbours
| Model | Sources of rewarded risk | Inputs needed |
|---|---|---|
| CAPM | One: the market | Beta, risk-free rate, market premium |
| Fama–French three-factor | Market, company size, value | Three betas and three premiums |
| Arbitrage pricing theory | Several, not named by the theory | A beta for each chosen factor |
| Dividend discount model | None explicitly; return implied by price and payouts | Dividend, growth rate, price |
How Finabase does it
Finabase does not compute a CAPM expected return, a cost of equity or an alpha for any symbol.
It does show two of the three inputs. A symbol's key statistics and the comparison table include beta, supplied by the data provider; the page does not state the index or the window used, so treat it as approximate. The Treasury page lists current yields from one month to thirty years, which gives the risk-free rate for whichever horizon is being used. The third input, the market risk premium, is an assumption that no screen can supply.
With those, the formula is one line: yield + beta × assumed premium. Running it for two or three values of the premium shows how wide the plausible range is.
Frequently asked questions
What does CAPM tell me in plain terms?
That an investment's expected return should depend on how much it moves with the market, and on nothing else about it. Twice the beta, twice the premium over the risk-free rate.
Is CAPM accurate?
As a precise predictor, no. Across many markets and periods, the relationship between beta and return has been weaker than the model says. As a first approximation and a shared vocabulary, it remains in use.
What is the security market line?
The straight line the model predicts when expected return is plotted against beta. It crosses the vertical axis at the risk-free rate and passes through the market portfolio at beta 1. Assets above it have positive alpha.
How is the capital market line different from the security market line?
The capital market line plots expected return against total volatility and applies only to efficient portfolios (mixes of the market and cash). The security market line plots it against beta and applies to every asset.
Why do companies still use CAPM for cost of equity?
Because the alternatives need more inputs that are just as uncertain, and because a common method makes estimates comparable. Most practitioners treat the result as a midpoint of a range.
Summary
CAPM says expected return equals the risk-free rate plus beta times the market premium, on the argument that only undiversifiable risk is rewarded. Tested on 14 funds from 2005 to 2026, beta separated bonds from stocks well and separated one stock fund from another poorly: the fitted line rose 6.0 points per unit of beta where the model predicts 9.6, and low-beta assets beat their predictions. The model supplies the definitions of alpha and cost of equity and a sound default about diversifiable risk. Its numerical predictions should be treated as rough, and models with more factors were built to do better.
Check your understanding
5 questions on this lesson. Each answer comes with the reason, and nothing is scored.
The risk-free rate is 4%, the expected market premium is 5% and a stock’s beta is 0.8. What return does CAPM predict?
- 8%
- 9%
- 7.2%
Show the answer
A. 8% Expected return = 4% + 0.8 × 5% = 8%. Beta scales the premium, not the whole market return.
According to CAPM, why should a volatile stock with a beta of 0.5 have a lower expected return than a steadier stock with a beta of 1.2?
- Because volatile stocks are always poor investments
- Because only market-related risk is rewarded; the rest can be diversified away
- Because low-beta stocks pay no dividends
Show the answer
B. Because only market-related risk is rewarded; the rest can be diversified away In the model, risk specific to one company earns no premium, since investors can remove it by holding many assets. Only sensitivity to the market counts.
Across 14 funds from 2005 to 2026, the line fitted through beta and return was flatter than CAPM’s line. What does a flatter line mean?
- Low-beta assets earned more, and high-beta assets less, than the model predicts
- Beta and return were unrelated
- High-beta assets earned far more than predicted
Show the answer
A. Low-beta assets earned more, and high-beta assets less, than the model predicts A higher intercept and a lower slope mean the reward per unit of beta was smaller than the market premium. The pattern has been reported since the early 1970s.
Gold had a beta near 0.1 and earned about 9% a year above Treasury bills. What does that show about CAPM?
- Gold is mispriced
- A single market beta cannot explain assets whose returns come from sources unrelated to the stock market
- The beta was calculated wrongly
Show the answer
B. A single market beta cannot explain assets whose returns come from sources unrelated to the stock market CAPM has one source of rewarded risk. An asset that moves independently of stocks gets a predicted premium near zero, whatever drives its actual return.
Which CAPM input is the least observable?
- The risk-free rate
- The expected market risk premium
- The stock’s price
Show the answer
B. The expected market risk premium A Treasury yield can be read off a screen and beta can be estimated from history. The premium the market will earn in future is a forecast, and plausible values differ by several points.
Sources and further reading
- William F. Sharpe, “Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk,” Journal of Finance, vol. 19, no. 3 (1964)
- John Lintner, “The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets,” Review of Economics and Statistics, vol. 47, no. 1 (1965)
- Fischer Black, Michael C. Jensen and Myron Scholes, “The Capital Asset Pricing Model: Some Empirical Tests,” in Studies in the Theory of Capital Markets (Praeger, 1972)
- Eugene F. Fama and Kenneth R. French, “The Capital Asset Pricing Model: Theory and Evidence,” Journal of Economic Perspectives, vol. 18, no. 3 (2004)
- Board of Governors of the Federal Reserve System, 3-Month Treasury Bill Secondary Market Rate (TB3MS), via FRED
- Finabase chart data — monthly bars and dividends of 13 funds and monthly bars of GLD, January 2005 to October 2026; statistics computed by Finabase
