Projections are estimates based on current data, updated daily. Not investment advice. They show what our model says a company's cash flows are worth under stated assumptions — not where its share price will go.
1. The cash flow we value
We start from each company's trailing-twelve-month free cash flow — operating cash flow minus capital expenditure — taken from its SEC filings (10-K and 10-Q, or foreign annual reports on 20-F and 40-F). We then subtract stock-based compensation. Paying staff in shares is a real cost to shareholders that never touches cash flow; leaving it in would flatter exactly the companies that pay the most in stock. What remains is the cash the business generates for its owners.
Young, fast-growing companies often pay a large share of revenue in stock, sometimes nearly all of their free cash flow. As a business matures that share tends to fall. Since method 1.6 (October 2026), where stock-based pay is above 10% of revenue we assume it falls in a straight line to 10% over the ten forecast years, and we count the cash that frees up in each year. The same rule applies to every company; below 10%, nothing changes.
Operating cash flow is reported after interest paid, but we subtract all debt when we work out the value per share (section 3). To count debt only once we add after-tax interest back: the result is the cash the business generates for all its investors, lenders and shareholders alike, and the 8.5% rate in section 3 is a rate for the whole business. Airlines and online travel firms are a special case: customers pay them before they fly or travel, and that float swells reported cash flow without being earnings. For them we use cash flow with working-capital movements stripped out.
We then normalise it: instead of taking one year's figure, we take the cash the business generated over its last three financial years, divide it by its revenue over the same three years, and apply that margin to today's revenue. A single year can be distorted (a capital-spending surge, a working-capital swing, a one-off) and five years of projected value should not hang on it. Weighting by money rather than by year means a small bust year can't outweigh a much larger normal one.
Companies whose revenue rises and falls with an industry cycle (memory chips, steel, airlines) get one adjustment: when the latest year's margin is at most half the company's own margin over the past ten years, we blend the three-year and ten-year margins, because in our tests of past data a business at the bottom of its cycle tended to recover towards its long-run margin. At the top of a cycle we keep the three-year figure: our tests showed booms rarely reverse in full.
2. Growth
The next two years use what is known about them. Year one blends three sources: analysts' consensus for the next twelve months (third-party estimates), management's own revenue guidance from its latest earnings release, and the company's record (below). The consensus weighs more the more analysts cover the company (half the weight with ten or more, less with fewer); guidance weighs 30% when it is newer than the consensus and 15% once the consensus has had time to absorb it; the record weighs 20%. Year two blends the consensus for the following year with the record. Consensus is compared with the revenue the company has already reported over the last twelve months, on the same definition, so growth already in the figures is not counted twice. When the consensus describes a different business from the one in the filings (after a spin-off, disposal or a merger not yet reported) we don't use it. Years three to five move back in a straight line to the record-based rate. The section "Forecasts and guidance" on each company page shows the figures and the weights used.
The record-based rate grows that cash flow at the average of two rates: revenue growth over the last twelve months and the three-year average growth rate, limited to between 2% and 20% a year, so that one exceptional year doesn't dominate. When growth is erratic (revenue up more than 40% or down more than 10% in the latest year, or the latest year more than 15 points away from the three-year rate) we move it halfway towards 6%, a typical long-run rate for US companies: very fast or very uneven growth rarely lasts, and this tempered estimate forecast past cash flows more accurately. Over the following five years growth fades in a straight line to 2.5% a year, roughly long-run nominal economic growth, and stays there.
3. Discounting and the value per share
Future cash flows are discounted at 8.5% a year, a long-run required return for a US business, and the final year's cash flow is capitalised as a perpetuity growing at 2.5%. We then add cash and short-term investments and subtract everything owed to others before shareholders: all debt (long and short term, including finance leases), minority interests, preferred stock and the after-tax deficit of underfunded pension plans. Then we divide by diluted shares. The result is our value per share, which we show at two significant figures because the inputs do not justify more precision.
4. Lower, base and higher cases
The base case uses the growth path above. The lower and higher cases move growth for years one to five by five percentage points in each direction, everything else unchanged. The spread between them is our measure of how sensitive the value is to the growth assumption.
5. The value history
On each valuation page, the history line re-runs today's method on the figures each company filed for every past quarter. It shows how the method would have valued the company at the time. It is a backtest of the method, not a record of values we published then; our dated public record starts at launch and is kept in our tamper-evident ledger.
6. The five-year projection
The projection shows where our model's value would be at the end of each of the next five years if a case plays out. It follows from the model itself: value grows at the 8.5% discount rate each year from the latest filing, less the cash the business generates in that year (cash that is either paid out to shareholders or reinvested in the value already counted). In symbols: value next year = value this year × 1.085 − that year's cash flow per share. It is not a forecast of the share price, which can stay above or below value for long periods.
7. Companies that don't generate cash yet: the revenue-to-maturity model
When free cash flow after stock pay isn't positive, the cash-flow model has nothing to compound. For those companies we value the path to maturity instead: revenue grows from today's level at its recent rate (capped at 30% a year), fading to 2.5% over ten years, while the cash margin moves in a straight line from where it is today to the median margin of cash-generating companies in the same industry in our coverage. Those cash flows are discounted at 10% — 1.5 points above our standard rate, for the risk that maturity never arrives — plus net cash. If even that path doesn't produce a positive value, we say so rather than show a number.
8. Several models, one combined value
A cash-flow model doesn't describe every business. A bank's or insurer's debt is its raw material, not its financing, so its free cash flow means little; a property trust or a regulated utility can be highly profitable while investing more cash than it generates. So for every company we run each model that fits its kind of business, show each model's value per share on the stock page, and combine them. Research on valuation and on forecasting finds that different models fail in different ways and that a simple average of sound estimates is usually more accurate than any one of them (Yee 2004; the forecast-combination literature, e.g. Clemen 1989 and Timmermann 2006).
The models
- Residual income. Book value per share plus the present value of the returns the company earns above its cost of equity. Today's return on equity (averaged over up to five years, so one strong or weak year doesn't set it) fades over ten years to a level that keeps half of its excess over the cost of equity. Book value grows by the earnings the company keeps. This is the Edwards-Bell-Ohlson model (Ohlson 1995; Frankel and Lee 1998; Penman's textbook treatment), and it is Damodaran's recommended "excess return" approach for financial firms.
- Justified price-to-book. If a company keeps earning the same return on equity and grows steadily, its shares are worth book value × (return on equity − growth) ÷ (cost of equity − growth) (Wilcox 1984). Growth is what the company can fund from retained earnings, capped at 3% a year. A company that only earns its cost of equity is worth its book value.
- Price-to-book against return on equity. Across peers, banks and insurers with higher returns on equity trade at higher multiples of book value. We fit a straight line through the peers' price-to-book and return on equity and read off the multiple for this company's return. Studies of US insurers find this conditioned price-to-book among the most accurate relative-valuation methods (Nissim 2013). We use it only with at least eight peers and a reasonable fit.
- Peer multiples. Earnings per share × the price-to-earnings of similar companies in our coverage; book value per share × their price-to-book; for property trusts, funds from operations per share × their price-to-FFO. We use the harmonic mean of the peers' multiples, which research finds more accurate than the average or the median (Baker and Ruback 1999; Liu, Nissim and Thomas 2002), and need at least five peers. Peers are the same kind of business: banks and lenders; life insurers; property and casualty insurers; asset managers; brokers and exchanges; insurance brokers; property trusts; utilities; and for other companies their industry or sector. Price-to-sales is used only for companies without positive earnings, at half weight, because sales are the weakest guide to value (Liu, Nissim and Thomas 2002). Peer multiples use third-party market data and move with the market.
- Dividend discount. Dividends grow for five years at the rate the business can fund (return on equity × the share of earnings it keeps, capped at 10%), then the payout settles at the level steady 3% growth allows (Damodaran). For property trusts, which must pay out most of their income, today's dividend growing at 2.5% a year. Companies that pay no regular dividend are not valued this way.
- Funds from operations (property trusts). Net income plus property depreciation, the industry's standard measure of cash earnings, growing at 2.5% a year and discounted at 8%.
- Utility earnings. Regulated utilities and power producers often have negative free cash flow by design while they build assets that earn a regulated return. We value their net income growing at 3% a year, discounted at 8%. When such a company is profitable, the earnings-based models lead and the revenue-to-maturity model is not used.
Cost of equity
The models above (except the funds-from-operations and utility earnings models, which use 8%) discount at a cost of equity set for each company: the 10-year US Treasury yield plus beta × a 4.5% equity risk premium (the capital asset pricing model). Beta measures how much the shares have moved with the S&P 500, from three years of weekly prices. Because one company's beta is a noisy estimate, we take half the company's own beta and half its peer group's median, then pull it a third of the way towards 1 (Blume's adjustment). The result is kept between 7% and 14%.
Which models count, and how they are combined
- Banks and lenders, insurers: residual income (primary), justified price-to-book, price-to-book against return on equity, peer price-to-book, peer price-to-earnings, dividend discount.
- Asset managers, brokers and exchanges: peer price-to-earnings (primary), residual income, justified price-to-book, peer price-to-book, dividend discount.
- Property trusts: funds from operations (primary), peer price-to-FFO, dividend discount. Mortgage trusts, which hold loans rather than buildings, are valued like lenders.
- Utilities and power producers: utility earnings (primary), peer price-to-earnings, justified price-to-book, residual income, dividend discount.
- Other companies: the cash-flow model (primary) with peer price-to-earnings, residual income and justified price-to-book as cross-checks. Where free cash flow isn't positive yet, the revenue-to-maturity model takes the cash-flow model's place: as the primary model for loss-making companies (with peer price-to-sales), and counting once beside the earnings-based models, with peer price-to-earnings primary, for profitable ones.
Wherever the cash-flow model applies it is the primary model, and it alone sets our value and rating for those companies; the other models are shown as a cross-check. For every other company the value we show is the combined value: the weighted average of the models that apply, with the primary model counting double. Equal weights, rather than weights fitted to past data, because fitted weights rarely do better in practice. The range shown runs from the lowest to the highest of the models used.
A model is left out, with the reason shown, when its inputs are missing or don't make sense (no positive book value or earnings, a return on equity below long-run growth, no dividend, fewer than five peers), when its value is below 0.2× or above 5× the share price (that far from the market usually signals a data problem such as a merger or a share-count change, not a bargain), or when it is more than three times away from the median of the other models. If every model is left out, we show the company as under review rather than a number. If nothing applies to a bank or insurer, we fall back to book value per share. Model values are estimates under stated assumptions, not price targets.
What these models can't see: life insurers' embedded value (not in SEC filings), tangible book value (preferred shares and goodwill are included in book value), analysts' forecasts (we use reported and averaged earnings), and the consolidated funds that make alternative asset managers' reported earnings a poor guide to value.
9. Young and fast-growing companies: the growth model and forward estimates
A company that has only just listed, is not yet profitable, or is spending heavily to grow can't be valued from its past cash flow: the past is short, and today's spending is an investment in future revenue, not a permanent cost. For these companies the growth cash-flow model leads, following the revenue-driven approach for young growth firms set out by Aswath Damodaran.
- Revenue: this financial year and next come from the analysts' consensus (third-party estimates, used only when at least three analysts cover the company). From year three, growth starts at the lower of half the year-two rate and a cap that falls as the company gets bigger, then fades to 2.5% by year ten: research on how long fast growth lasts (Chan, Karceski and Lakonishok, 2003) shows it rarely persists.
- Margin: today's operating margin moves to that of profitable peers by year seven (the upper quarter of peers for companies growing 25% or more a year). A company already above its peers keeps half its advantage.
- Investment: each year's extra revenue needs capital, at the rate peers' businesses turn capital into sales.
- Risk: cash flows are discounted at 10% for five years, easing to 8.5% by year ten, and the value is reduced by a chance of failure set by how long the company's cash would last at today's burn (5% to 20%).
Alongside it, forward price-to-earnings and (for loss-makers, at half weight) forward price-to-sales compare next year's consensus figures with peers', which research finds more accurate than trailing multiples (Liu, Nissim and Thomas, 2002). Companies valued mainly with the growth model show a value but no Buy, Hold or Sell rating until the model has a track record.
Worked example: NVIDIA
- Free cash flow, FY2027 Q2 TTM: $127B
- Less stock-based compensation: $7.2B
- Cash flow we value (after-tax interest added back): $123B
- Base growth, years 1–5: 66.7%, 34.2%, 27.2%, 20.1%, 13% (lower and higher cases 5 points either side)
- Value per share: ~$320 base (~$250–$410)
- Base-case projection for 2031: ~$400
Full workings: NVIDIA valuation page.
10. Businesses that need a different lead model
- Oil and gas producers: led by the net asset value of proved reserves, the standardised measure every producer reports in its annual report (future cash flows from proved reserves at SEC prices, after tax, discounted at 10%), less debt and other claims. It is a floor: unproved acreage, hedges and price moves since the year end are left out. Royalty and land companies are valued like other businesses.
- Health plans: led by forward earnings against other health plans; their cash flow swings with the timing of claims and premiums.
- Car makers with finance arms (and other manufacturers with large lending businesses): not rated until we can value the two businesses separately; the lender's debt makes a single cash-flow model meaningless.
11. Foreign financial statements and US-listed shares
We read US-GAAP and supported IFRS tags from SEC annual filings. Foreign issuers that provide only annual financial statements are shown annually; quarterly figures are not invented. The source taxonomy and reporting currency are identified on the company page.
Foreign-currency income and cash flows are translated into USD using period-average Federal Reserve H.10 observations from FRED. Balance-sheet amounts use an observation on or before the balance date. We derive quarters in the reporting currency before conversion, then build USD trailing totals from the converted quarters. We do not use a future exchange rate or assume parity when a rate is missing.
An American depositary share can represent several ordinary shares. We verify the ratio from the SEC securities disclosure and adjust the denominator to the US-listed security. When recent filing share counts are absent, a current third-party count may be used; an old historical count is not carried forward indefinitely. Unverified listing ratios, currencies or consensus EPS units prevent the corresponding model from being used.
For IFRS cash flows with interest classified as financing, we deduct that interest in the valuation cash-flow series before applying the firm's after-tax interest adjustment. Reported statements remain as filed, translated into USD. The treatment follows the cash-flow distinctions in IAS 7. Financial quality checks compare source-currency statements so translation differences are not mistaken for an accounting error.
Exchange-rate translation does not remove economic or country risk. Discount rates and growth rates remain the model assumptions shown in the calculator; they are not exchange-rate forecasts.
12. Reported crypto treasury NAV
For identified crypto treasury companies and miners, we calculate the reported crypto fair-value holdings plus cash and short-term investments, less borrowings, preferred capital and the other claims in the equity bridge, divided by the listed share count. We never add a reported total to its overlapping current/noncurrent or restricted/unrestricted breakdowns, and we do not include customers' custodial assets.
This is a dated treasury measure. It excludes the operating business, mining equipment, future token purchases and movements in crypto prices or holdings since the filing. It is not a live NAV or a price target. If the balance, debt, share basis or dates cannot be reconciled, no supported positive value is shown. The treasury value alone does not generate a Buy/Hold/Sell rating.
Common-share and peer assumptions
Where the company reports them, earnings available to common shareholders are used in equity valuation. Preferred capital is removed from the book value allocated to common shares. This is book value, including goodwill, rather than a claim to tangible book value. Peer growth-margin anchors exclude failed quality checks and operating margins above 100%; a one-off accounting gain is not a sustainable mature operating margin.
What the model cannot see
- Anything not in the filings yet: new products, competitors, regulation, management changes.
- Cyclical peaks and troughs: a record year's cash flow can overstate value, a weak year understate it.
- Heavy investment phases: capital spending that will pay off later lowers today's free cash flow.
- Company-specific risk: we use one discount rate for every company (heavily indebted companies are held for review for that reason).
Updates
Values are recalculated every day from the latest filed figures; they change when a company files new results. Methodology changes are versioned and noted here. See also our general methodology and research and recommendations disclosure.