Methodology

The cashflow build

Free cashflow to firm is built from drivers rather than taken from a reported total, so every line is traceable:

Revenue(t) = Revenue(t-1) x (1 + growth)
EBIT(t)    = Revenue(t) x margin
NOPAT(t)   = EBIT(t) x (1 - tax rate)
FCFF(t)    = NOPAT(t) + D&A(t) - Capex(t) - change in NWC(t)

An increase in net working capital is a cash outflow and is subtracted. Yahoo reports the working-capital movement with the opposite sign, as a cashflow contribution, so it is flipped once during normalisation.

Discount rate

Cost of equity = risk-free + beta x equity risk premium
Cost of debt   = interest expense / average total debt
WACC           = E/V x Re + D/V x Rd x (1 - tax rate)

Risk-free rates are sourced live for the US (10-year Treasury) and Australia (RBA F2 table, the 10-year Australian Government series). Every other market uses a documented assumption, labelled as such on the model page.

Two guards apply, because reported inputs are sometimes not what they claim to be. Beta is clamped to a range of 0.3 to 2.5: reported betas are occasionally negative, and a cost of equity below the risk-free rate is not a defensible discount rate. The cost of debt is floored at the risk-free rate and capped at 800bp above it - beyond that, interest expense and reported debt are measuring different things rather than the company being distressed, which is common for banks whose interest expense includes deposits that total debt excludes.

Terminal value

Both methods are shown, because terminal value is usually most of the answer and the gap between the two says how much rests on assumption. Perpetuity growth requires terminal growth below the WACC; where it is not, the calculation is blocked rather than fudged.

The exit multiple defaults to the company's own current EV/EBITDA rather than a fixed number - a single multiple applied to a software business, an oil major and a property trust alike is an arbitrary anchor, not a second opinion. It is clamped to 3-20x, because a depressed or blown-out trailing multiple should not be projected five years out unchallenged. Note what this makes it: anchored to today's price, so it answers "what if the rating holds" rather than providing an independent intrinsic check. The perpetuity method is the intrinsic one.

The terminal year is put on a steady-state footing, which matters more than it sounds and cuts both ways. Carrying a buildout capex rate into perpetuity understates value - Alphabet spends ~22% of revenue on capex against ~5% depreciation, and assuming that gap persists forever cost ~44% of value per share. But setting capex equal to depreciation and then growing the result is the opposite error: zero net reinvestment with positive growth implies infinite return on incremental capital, and overstates terminal value by 9-33%.

So the reinvestment identity is applied instead: g = ROIC x reinvestment rate.

FCFF terminal = NOPAT x (1 - g / ROIC)

Depreciation covers maintenance; the reinvestment rate funds growth on top. ROIC is NOPAT over invested capital from the company's own accounts, floored at the WACC - below that, growth destroys value, which on this data usually means invested capital is the wrong denominator (financials) rather than a genuine finding. Both discount at the same convention as the explicit flows: under mid-year the terminal value discounts at n - 0.5, because the perpetuity it capitalises also arrives through the year. The forecast years always show the real capex being spent.

Scenario presets

• Base- revenue CAGR, mean EBIT margin, mean capex and D&A as a share of revenue, and the median effective tax rate, all from the company's own reported history.
• Bull - Base plus 3pp of revenue growth, 2pp of EBIT margin and 0.5pp of terminal growth.
• Bear - Base less the same amounts.
• Consensus - analyst revenue estimates for the first two years, fading linearly to the historical growth rate. No consensus exists for margins or capex, so those stay rule-based.

Priced in

Each input is also solved in reverse: what would revenue growth, margin, or the discount rate have to be for today's price to hold, with the others left where the model has them. Bisection against the same engine that produces the headline. Alphabet currently needs ~46% annual revenue growth for five years against 12.5% of history, or a discount rate near 6% - alternatives, not a set. Returns nothing where the price is unreachable in any input, rather than extrapolating past the bracket.

Where this breaks

• Banks and insurers. They report no meaningful operating income, and capex and working capital do not mean what this model assumes. A dividend-discount or excess-returns model is the right tool. The model still runs, with a warning.
• Property trusts. Reported capex can be a fraction of real investment - one large REIT reports capital expenditure of 0.1% of revenue because it invests through investment properties rather than plant and equipment. Where capex is negligible or falls well below depreciation, it is substituted and flagged.
• Hybrids and preference lines. Yahoo types a hybrid note the same as ordinary shares, so a search for a bank can return both. A hybrid has no claim on group free cashflow - its value is its coupon and call schedule - and it carries no market capitalisation, which is how those lines are detected and flagged.
• Loss-makers. Terminal value ends up carrying essentially the whole valuation, which is disclosed rather than hidden.
• History depth. Only four to five annual periods are available free, so the historical anchors are short-run averages, and a five-year DCF anchored on trailing history will read low against a company the market is pricing for acceleration.
• Currency. Some companies report in one currency and trade in another, and some markets quote in a sub-unit such as pence. The model works in reporting currency throughout and converts once, at the end.

Data sources

Statements, prices, betas, analyst estimates and FX from Yahoo Finance. Australian risk-free rates from the Reserve Bank of Australia (F2, capital market yields). Both are free sources and neither is guaranteed accurate or current. Not affiliated with Yahoo.