CAPM Calculator – Capital Asset Pricing Model & Jensen's Alpha

What Is the Capital Asset Pricing Model (CAPM)?

Two stocks both delivered 12% last year. Stock A is a defensive utility with a beta of 0.4 — it barely moves with the market. Stock B is a high-flying semiconductor with a beta of 1.8 — it amplifies every market swing. Same return, very different risk. Which one was the better bet? The Capital Asset Pricing Model, introduced by William Sharpe in 1964 and independently derived by John Lintner (1965) and Jan Mossin (1966), provides a clean answer: the market only rewards you for systematic (non-diversifiable) risk, measured by beta. CAPM tells you what return you should have earned given the risk you took. Anything above is alpha; anything below is underperformance. (Note that CAPM's expected return is on a pre-tax basis; for U.S. investors, dividends, capital gains, and other equity income are classified for tax purposes under IRS Publication 550, which becomes relevant when comparing CAPM-implied returns to your actual after-tax keep.)[1, 2, 3, 28]

The CAPM formula is deceptively simple: Expected Return = Rf + β × (Rm − Rf). Rf is the risk-free rate (typically the 10-year Treasury yield, currently around 4.5%). Rm is the expected return on a broad market portfolio (S&P 500 long-term realized return is roughly 10%). β captures how sensitive the stock is to market moves. The expression (Rm − Rf) is the market risk premium — what investors demand for bearing market risk over the risk-free alternative. Multiplying by β scales that premium up or down for the stock's specific systematic risk. Add the risk-free rate, and you get the return that fairly compensates an investor for holding this stock.[4, 5]

CAPM rests on a clean intuition: investors hold diversified portfolios, so idiosyncratic risk (a single firm's missed earnings, scandal, recall) cancels out across holdings. What remains — and what the market must price — is exposure to broad market movements. Beta is therefore the only kind of risk that should earn a premium. A stock with β=2 has twice the market's systematic risk and should command twice the market risk premium. A defensive stock with β=0.5 should earn half. The Security Market Line (SML) plots this relationship: a straight line from the risk-free rate at β=0 to the expected market return at β=1, extending linearly. Stocks plotting above the SML have positive Jensen's Alpha and look attractively priced; stocks below are overpriced relative to their risk.[6, 21]

Interpreting Beta Bands and Alpha Signals

There is no formal industry consensus on beta thresholds, but a workable convention — aligned with how BlackRock's jargon buster describes the metric — is: Defensive (β < 0.5) — utilities, consumer staples, gold miners; Low Beta (0.5 ≤ β < 0.9) — healthcare, mature dividend payers; Market (0.9 ≤ β < 1.1) — broad index funds, large caps tracking the S&P 500; Growth (1.1 ≤ β < 1.5) — high-quality tech, financials in expansion; Aggressive (β ≥ 1.5) — speculative tech, biotech, leveraged or small-cap names. Our calculator clamps β to [0, 3] in the slider for typical use; mathematically the formula accepts negative β (inverse ETFs, gold in some cycles) — pass it via URL `?b=-0.5` to override.[18, 20, 33]

Jensen's Alpha is the gap between actual realized return and CAPM-expected return: α = Actual − [Rf + β(Rm − Rf)]. Michael Jensen (1968) introduced it precisely to separate manager skill from market exposure. A positive α means the stock or fund did better than CAPM predicted given its risk — a sign of either skill, mispricing, or factors CAPM misses (size, value, momentum, profitability). A negative α means it underperformed for its risk. We treat α within ±0.5 percentage points as effectively zero — there is too much measurement noise in real returns and beta estimates to declare confidence on smaller gaps.[6, 12]

CAPM and the Security Market Line (SML)

The Security Market Line is the geometric expression of CAPM. Plot β on the horizontal axis and expected return on the vertical axis. The SML is a straight line that begins at (β=0, return=Rf) and passes through the market portfolio at (β=1, return=Rm). Its slope is exactly the market risk premium (Rm − Rf), and its intercept is the risk-free rate. The line extends linearly: a stock with β=2 is expected to earn Rf + 2(Rm − Rf), at twice the market risk premium plus the risk-free rate. The SML is not a forecast — it is the equilibrium relationship CAPM says should hold if all investors hold mean-variance efficient portfolios.

A position above the SML signals positive Jensen's Alpha — the stock outperformed its CAPM-implied required return. This could mean the market underestimates its quality, the stock has factor exposures CAPM misses (small-cap premium, value premium, momentum, profitability), or it caught a tailwind unrelated to systematic risk. A position below the SML signals negative α: the stock underperformed expectations given its β. Don't leap to "buy" or "sell" conclusions from a single observation. CAPM is a one-period model, betas are estimated with error, and the market portfolio Rm is itself unobservable in pure form (Roll's critique, 1977). But the SML still serves as a discipline: given how risky this stock is, did it pay me enough?[7]

Jensen's Alpha: Skill, Luck, or Missing Factors?

Jensen introduced alpha in his 1968 study of mutual fund performance from 1945 to 1964. He found that, on average, the funds in his sample produced negative alpha after fees — i.e., they underperformed their CAPM-implied benchmarks. This launched a half-century debate about whether active managers have skill that justifies their fees. Modern indexed investing owes much of its intellectual scaffolding to Jensen's finding. William Sharpe himself, in his 1990 Nobel Prize lecture, emphasized that CAPM's main contribution was clarifying which risks should be priced — diversifiable noise should not be — making Jensen's α the natural attribution measure. For an individual stock, a positive Jensen's α over a multi-year window is a hint that the company is either genuinely outperforming relative to its market risk, or that CAPM is mispricing it because some other factor (size, value, momentum, quality) drives its return.[6, 25]

Statistical significance matters more than the point estimate. A 1% Jensen's α over a single year on a single stock is essentially noise — beta itself has a standard error, the realized market return varies, and one year is too short. Jobson and Korkie (1981) showed that even Sharpe ratio comparisons require dozens of monthly observations for confidence intervals to tighten. The same applies to Jensen's α. Use the metric to rank alternatives over comparable periods, not to declare one observation a "winner." For an individual stock, a multi-year track record of consistently positive α is far more meaningful than a single year of large positive α.[12]

Choosing Rf and Expected Market Return for 2026

For Rf you have two defensible choices. Academic textbooks favor the 3-month T-bill (DTB3) because it best matches the single-period nature of CAPM. Practitioners often prefer the 10-year Treasury (DGS10) because most equity holding periods are multi-year, and a long-duration risk-free benchmark better reflects the opportunity cost of stock investing. As of early 2026, the 10-year Treasury yields roughly 4.5% — still elevated relative to the 2010s zero-rate era — and our default uses that figure. The Federal Reserve's H.15 statistical release publishes both daily.[15, 14, 16]

For the expected market return Rm, opinions diverge more sharply. The S&P 500 has delivered roughly 10% nominal annualized over the long run (since 1928, per S&P Dow Jones Indices). But forward-looking estimates of the implied equity risk premium (ERP) are usually lower than the historical average. Aswath Damodaran's 2026 ERP study (the 2026 Edition, March 2026) reports an implied ERP of around 4.23% at the start of 2026 — close to the long-run historical norm of 4.5–5%. With Rf at 4.5% and ERP around 5.5%, we land at Rm = 10% as a reasonable default. You can override this in the calculator if you prefer 7%, 8%, or any other defensible figure.[13, 17]

A complementary cross-check comes from Robert Shiller's CAPE (Cyclically Adjusted Price-to-Earnings) dataset. The cyclically adjusted earnings yield — the inverse of CAPE — is a rough proxy for the expected real return on the S&P 500. With CAPE elevated in the 30+ range as of early 2026, the implied real earnings yield runs near 3.0–3.3%, which suggests a forward-looking Rm closer to 7–8% real (roughly 9–10% nominal once you add 2%+ inflation expectations). That broadly aligns with Damodaran's implied ERP plus the 4.5% Treasury — but if you are willing to bet on mean reversion in valuations, the forward Rm could come in lower than today's starting yields suggest. Use both lenses; never anchor on a single number.[32]

A practical reminder: CAPM's expected return is pre-tax. After-tax investors — anyone holding equities in a taxable account rather than a 401(k) or IRA — receive a lower realized return. Long-term capital gains tax rates per IRS Topic 409 currently sit at 0%, 15%, or 20% depending on income, with an additional 3.8% Net Investment Income Tax for high earners. A nominal 10% expected return shrinks to roughly 7.6% after a 23.8% top combined rate. For long-horizon planning in taxable accounts, work with after-tax cost of equity; for tax-deferred accounts, the pre-tax CAPM number is the right input.[29]

How Beta Is Actually Estimated

The textbook recipe is a linear regression of the stock's excess returns against the market's excess returns over a chosen window: typically 60 monthly observations (5 years) versus the S&P 500. The slope coefficient is your raw beta. Morningstar, Yahoo Finance, and Bloomberg all report betas, but they use slightly different windows, frequencies, and adjustments. Bloomberg famously applies a "shrinkage" adjustment toward 1.0 — Adjusted Beta = 0.67 × Raw Beta + 0.33 — based on the empirical observation that betas tend to mean-revert toward the market over long periods.[18]

Two practical caveats. First, beta changes over time. The beta of a tech stock during the 1990s tech boom is not the same as during the 2008 financial crisis. Sliding-window estimates are common, but the choice of window matters: a 5-year window smooths short-term noise but may stale rapidly during structural change (think: airline betas pre- vs post-pandemic). Second, leverage affects beta. A company that adds debt mechanically increases its equity beta because earnings volatility scales with the debt-equity ratio. SEC investor.gov emphasizes that retail investors should treat published beta as a snapshot, not a permanent attribute. When in doubt, compare betas from at least two reputable sources before using one.[19]

Levered vs Unlevered Beta: Adjusting for Capital Structure

A reported equity beta — the one Yahoo Finance or Bloomberg shows you — already reflects the company's current capital structure. Add debt to a balance sheet and equity earnings become more volatile, simply because fixed interest payments scale up the swings in net income. The Hamada equation (1972) formalizes the relationship: βL = βU × [1 + (1 − T) × D/E], where βL is the levered (observed) beta, βU is the unlevered (asset) beta, T is the marginal corporate tax rate, and D/E is the debt-to-equity ratio. The asset beta strips out leverage and isolates the company's pure business risk.[23]

The standard unlevering procedure runs in two steps. First, take a peer group of comparable companies, pull each one's reported beta and D/E ratio, and compute βU for each: βU = βL / [1 + (1 − T) × D/E]. Second, take the median (or mean) of those asset betas as the industry asset beta, then re-lever it for your target company's capital structure: βL,target = βU,industry × [1 + (1 − T) × D/E,target]. The result is a beta consistent with peer business risk but adjusted for your firm's leverage. Damodaran's Spring 2026 valuation packet walks through this approach with worked examples for U.S. and emerging-market firms.[26]

Why bother with this? Two practical reasons. First, when valuing an early-stage or private company that has no traded equity, you cannot regress its returns against the market — there are none. Pulling an industry-average asset beta and re-levering for the target capital structure is the standard workaround. Second, when comparing two listed peers with very different leverage, raw equity betas mislead. Two semiconductor companies with the same business but different debt loads will look like different risk profiles in CAPM, when really only their balance sheets differ. The CFA curriculum and most corporate-finance textbooks treat unlevering and re-levering as a routine step in cost-of-equity estimation.[21]

A pitfall worth flagging: the Hamada formula assumes constant debt beta (typically taken as zero) and a stable tax rate. For highly leveraged firms or those with floating credit risk, the simplification breaks down — practitioners then turn to alternative formulas (e.g., Miller-Modigliani with risky debt, or Harris-Pringle) that allow non-zero debt beta. For most retail-investor use cases — comparing peers within an industry, sanity-checking a published beta — the standard Hamada equation is the right starting point. If your peer's D/E is unusually high relative to the median, exclude it from the unlevering peer set rather than letting it skew the median asset beta.

Limitations and Critiques of CAPM

Richard Roll (1977) launched the most fundamental critique: CAPM cannot be empirically tested because the true market portfolio — which would include every risky asset, including human capital and unlisted holdings — is unobservable. Any test using a stock-market proxy is jointly testing CAPM and the choice of proxy. Decades later, Eugene Fama and Kenneth French (1992) delivered the most damaging empirical finding: across U.S. stocks from 1963 to 1990, beta had little explanatory power once size and book-to-market ratio were added. Small caps and value stocks earned higher returns than CAPM predicted, leading to the celebrated three-factor model.[7, 8]

Fama and French (1993) formalized the three-factor model adding size (SMB) and value (HML) factors, and Fama and French (2015) extended it to five factors with profitability and investment. Mark Carhart (1997) added momentum to make a four-factor model. Each addition further reduced the unexplained "alpha" in cross-sectional return data. Despite the empirical limitations, CAPM survives in textbooks and corporate finance practice because it provides a clean, single-factor language for risk pricing — useful for capital budgeting, cost of equity in DCF models, and benchmark construction even when its precise predictions disappoint.[9, 10, 11]

A point that is easy to overlook: even when CAPM's β estimate is statistically reliable, the resulting α can be drowned in measurement noise. Andrew Lo's 2002 work on Sharpe-ratio statistics extends naturally to Jensen's α — confidence intervals around realized α are wide unless you have many years of return data, and skewness/serial correlation in returns make the standard t-stat a generous test. Stock returns are also frequently fat-tailed; a single year of unusual returns can throw an alpha estimate off by hundreds of basis points. Treat any single-period α as suggestive at best; demand a multi-year track record before drawing strong conclusions about a manager's skill or a stock's mispricing.[12]

CAPM is also silent on what investors should do with risk they cannot diversify away. Fiduciary practice fills the gap: the CFP Board's Code of Ethics and Standards of Conduct requires planners to act in the client's best interest, including matching portfolio risk to the client's actual capacity and willingness to take risk — not just the β number that minimizes CAPM-implied required return. The Code's duty of loyalty and care implies that simply maximizing CAPM α is not a substitute for understanding a client's tax bracket, time horizon, liquidity needs, and emotional tolerance for drawdown. CAPM informs the math; fiduciary judgment governs the decision.[30]

Using CAPM in the 2026 Rate Cycle

The macro backdrop of early 2026 changes how CAPM defaults read. After the Federal Reserve held rates higher-for-longer through 2024 and 2025 — and as documented in the FOMC 2026 meeting calendar, the policy path remains data-dependent through year-end — the 10-year Treasury yield has stabilized in the 4.0–4.7% range, well above the post-2008 zero-rate norm but below the 1990s 6%+ era. With Rf at 4.5% and our default Rm at 10%, the market risk premium is 5.5%. That is roughly in line with Damodaran's implied ERP. A high-beta growth stock (β=1.5) has an expected return of 4.5% + 1.5 × 5.5 = 12.75%; a defensive utility (β=0.4) sits at 4.5% + 0.4 × 5.5 = 6.7%. The wider the gap between Rm and Rf, the more β scaling matters.[13, 22, 31]

A worked example brings this to life. Apple (AAPL) typically trades at β ≈ 1.2 in late-2025/early-2026 readings — slightly more cyclical than the broad market thanks to consumer-electronics exposure. CAPM-implied required return: 4.5% + 1.2 × 5.5% = 11.1%. Coca-Cola (KO), a consumer-staples defensive, trades at β ≈ 0.6: required return = 4.5% + 0.6 × 5.5% = 7.8%. So if AAPL's realized total return over the past 12 months was 14%, Jensen's α = 14% − 11.1% = +2.9% (positive, but within noise for a single year). If KO returned 9% over the same window, α = 9% − 7.8% = +1.2% — also positive but in the "essentially zero" gray zone. Comparing two stocks in the same direction (both above their CAPM line) is more informative than declaring either one a winner alone.

In late-cycle environments where ERP is compressed and growth concentration is high (large tech dominating index returns), CAPM may understate the true risk of high-β names that are also exposed to factor crowding. Conversely, defensive names with persistent positive Jensen's α can be quietly pricing in defensive demand from late-cycle investors. Use CAPM as a baseline, then sanity-check against multifactor models or sector-level risk premia. The 2026 backdrop also makes a pure 10-year Treasury Rf less stable: TIPS yields, real-rate expectations, and term-premium shifts all matter. If you are using CAPM for a corporate cost of equity, consider whether your time horizon really matches a 10-year nominal bond.[22]

CAPM in Cost of Equity and WACC Calculations

CAPM's most consequential real-world job is not picking stocks — it is producing the cost of equity that drives discounted cash flow valuation, capital budgeting, and merger pricing. Plug β, Rf, and the equity risk premium into the formula and you get cost of equity (Re) = Rf + β × (Rm − Rf). That number becomes the equity discount rate in a DCF model and one of two ingredients in the weighted-average cost of capital: WACC = (E/V) × Re + (D/V) × Rd × (1 − T), where E and D are market values of equity and debt, V = E + D, Rd is the cost of debt, and T is the marginal tax rate. CFOs and equity analysts use WACC to set hurdle rates for new projects and to discount free cash flows in valuation models.[26]

A worked example. Suppose a U.S. industrial firm with β = 1.1, market debt of $300M at a pre-tax yield of 5.5%, market equity of $700M, and a 21% federal corporate tax rate. With Rf = 4.5% and ERP = 5.5%: Re = 4.5% + 1.1 × 5.5% = 10.55%. After-tax cost of debt = 5.5% × (1 − 0.21) = 4.345%. Weights: E/V = 700/1000 = 0.70; D/V = 0.30. WACC = 0.70 × 10.55% + 0.30 × 4.345% = 7.385% + 1.304% ≈ 8.69%. That 8.69% becomes the discount rate when valuing this firm's free cash flows. Move β to 1.4 (a higher-beta tech name) and Re jumps to 12.2%, lifting WACC to roughly 9.85% — a ~1.2 percentage-point swing that meaningfully changes a DCF intrinsic value.

Three caveats from corporate-finance practice. First, use market values of debt and equity for V, not book values — book values lag market reality, especially for distressed or high-growth firms. Second, the cost of debt should reflect current yields on the firm's outstanding bonds (or comparable issuers), not the historical coupon. Third, the project-specific β can differ from the company's overall β: a stable-utility company evaluating a new biotech venture should use a biotech β for the project hurdle rate, not its low corporate β. CFA Institute curriculum (2026) covers these adjustments in the Portfolio Risk and Return refresher reading.[21]

CAPM as a one-period model raises a theoretical question for multi-period valuation: should the same Re apply year after year, or should it shift? Robert Merton's Intertemporal CAPM (1973) formally extends CAPM to a multi-period setting where investors hedge against shifts in the investment opportunity set, but the practical-finance industry typically uses a single, time-stationary Re for DCF tractability. If you are valuing a firm whose risk profile is expected to change materially (e.g., a startup transitioning from cash-burn to profitability), use staged WACC: a higher early-stage β phasing down to mature-firm β over the explicit forecast horizon.[24]

International CAPM and Country Risk Premium

Apply U.S. CAPM defaults to a Brazilian or Indian stock and you will systematically understate required return. Sovereign risk, currency volatility, weaker contract enforcement, and shallower capital markets all push emerging-market equities to demand a premium beyond U.S. ERP. The standard practitioner adjustment is to add a country risk premium (CRP) on top of the U.S.-based ERP. Two common formulations: a sovereign default-spread method and a relative-volatility method. Damodaran's country risk premium dataset, updated each January, publishes both versions for over 150 countries. The January 2026 update lists Brazil at a CRP near 2.98% and an India CRP under 2%, illustrating the wide range of EM risk pricing.[27]

Mathematically, the relative-volatility method computes CRP = Rating-based default spread × (σEquity / σBond). For Brazil with a sovereign spread of about 1.85%, an equity-market σ of ~30%, and a Brazilian government bond σ of ~18%, you get CRP ≈ 1.85% × (30/18) ≈ 3.08%. Adding to the U.S. ERP of 4.23% gives a Brazil-adjusted ERP of about 7.31% for cost-of-equity calculations. For a Brazilian firm with β = 1.0, that produces Re ≈ Rf,US + 1.0 × 7.31% — far above what a naive U.S. CAPM would suggest. The 2026 Damodaran ERP edition reports an evolving methodology that combines this with implied-ERP techniques in markets where index data is reliable.[13]

Two design choices matter. First, do you treat the country risk as additive (one CRP for every firm in that country) or company-specific (a CRP scaled by each firm's revenue exposure to the country)? Damodaran advocates the company-specific approach for multinationals, since a Brazilian-listed company with 80% U.S. revenue should not absorb the full Brazilian CRP. Second, integrated vs segmented markets: if the emerging market is fully integrated with global capital, a global CAPM with a single global ERP can suffice; if segmented (capital controls, restricted foreign access), you need the local CRP overlay. Most retail use cases — sizing exposure to EM equity ETFs, comparing global ADRs — use the additive shortcut, accepting the modeling simplification.

CAPM vs Multifactor and Alternative Models

The Fama-French three-factor model adds size (SMB) and value (HML) to beta. Empirically it explains roughly 90% of the variation in diversified portfolio returns versus CAPM's ~70% — a meaningful improvement. The five-factor model adds profitability (RMW) and investment (CMA) factors. Carhart's four-factor model adds momentum (UMD) to the original three. The Arbitrage Pricing Theory (APT) is a more general framework that admits multiple unspecified macro or statistical factors. None of these models replaces CAPM as a teaching tool or as a baseline for capital-budgeting cost of equity, but professionals routinely cross-check CAPM-implied required returns against multifactor-implied numbers.[10, 11]

Frequently Asked Questions

Common questions investors ask about CAPM, beta, and Jensen's Alpha: