Discounted cash flow explained

Step one: project the free cash flow

The starting figure is free cash flow: cash generated from operations after paying for the equipment and infrastructure needed to keep running. The usual approximation is cash from operations minus capital expenditure, both on the cash flow statement.

Use the UNLEVERED figure, measured before interest payments. This is the source of the most common mistake in a homemade model. Unlevered cash flow belongs to everyone who financed the business — lenders and shareholders together — and step four deals with splitting it. A figure that is already after interest deducts borrowing costs twice.

Then grow it, year by year. Growth compounds from the previous projected year, not from the starting figure: 10% growth for three years takes 100 to 110, then 121, then 133.1 — not to 130.

Step two: discount it back

Each projected year is divided by (1 + r) raised to the power of the year number, where r is your discount rate. Year one is divided once, year two twice. Money further away is divided more, which is the whole intuition.

The discount rate does two jobs at once: it is the return you require, and it is your estimate of risk. A riskier business deserves a higher rate, which mechanically produces a lower value. There is no objectively correct rate — the weighted average cost of capital is the standard estimate, and it is an estimate built on further estimates.

End-of-year discounting assumes each year’s cash lands on the final day. Mid-year assumes it arrives evenly through the year and so discounts by half a year less, giving a slightly higher value. Neither is more correct; what matters is applying the same convention to both the forecast and the terminal value.

Step three: the terminal value, where most of the answer hides

Businesses do not stop when your forecast does. The terminal value captures everything beyond the window, and it routinely accounts for well over half the total. Most of a typical DCF’s answer therefore comes from the part you did not forecast in detail — which is worth sitting with before quoting a number to two decimal places.

The perpetuity growth method assumes cash grows at a constant modest rate forever. Take the final forecast year, step it up by one more year of growth, and divide by the discount rate minus the terminal growth rate. The step-up matters, because the perpetuity starts the year after the forecast ends.

That subtraction is the method’s weak point. As terminal growth approaches the discount rate, the denominator approaches zero and the value explodes. If growth equals or exceeds the discount rate the formula is meaningless — it describes a company outgrowing the entire economy forever.

The exit multiple method is the cross-check: assume the business is sold at some multiple of its final-year cash flow. It ignores the growth rate entirely, so it fails differently, which is exactly what makes it useful. If the two methods disagree wildly, your terminal assumptions are doing too much of the work.

Step four: from the whole business to one share

The discounted forecast plus the discounted terminal value gives the enterprise value — the whole operating business, funded by lenders and shareholders together.

Shareholders do not own that. Lenders are paid first. Subtract debt and add back cash, because cash is owned outright, then divide by the diluted share count.

Skipping this step is the other classic amateur error, and it is not small: a company with 300 million of net debt and 50 million shares is overvalued by exactly six per share if you forget it.

One sign convention catches people out. If a company holds more cash than debt, net debt is negative, and subtracting a negative adds. A net-cash business is worth MORE per share than its enterprise value, not less.

Read the range, not the number

Small changes in the discount rate and the terminal growth rate produce large changes in the output, because both feed a denominator that can approach zero. That is the nature of dividing by a small difference between two uncertain numbers, not a defect in any implementation.

This is why a sensitivity grid is more honest than a single figure. If the plausible range spans a factor of three, the model is telling you it cannot answer the question precisely — and that is a legitimate finding rather than a failure.

Worked example: why the terminal value dominates

Start with 100 of free cash flow, forecast ten years, grow at 10% a year and discount at 10%. Because the growth rate equals the discount rate, every year’s present value is exactly 100: year one is 110 divided by 1.1, year two is 121 divided by 1.21, and so on. Ten years therefore contribute exactly 1,000.

The final forecast year is 100 times 1.1 to the tenth power, which is 259.37. Step it up by one year of terminal growth at 2.5%: 265.86. Divide by the discount rate minus terminal growth, 0.10 minus 0.025, which is 0.075. That gives a terminal value of 3,544.78 in year-ten money.

Discount it back ten years by dividing by 1.1 to the tenth, which is 2.5937: 1,366.7 in today’s money.

Result: Enterprise value is about 2,367 — of which 1,000 comes from ten years of detailed forecasting and about 1,367, or roughly 58%, comes from one line of arithmetic about the indefinite future. Change terminal growth from 2.5% to 3.5% and that single figure rises to about 1,592, roughly 17% higher, from a one-point change in a number nobody can know. The DCF Calculator ships a fully worked version of this arithmetic with every intermediate number shown, so you can check it with a calculator rather than trust it.

Open the tool

Build a DCF from your own assumptions

Runs the arithmetic in your browser, shows the equity bridge as an explicit chain, and refuses to compute a perpetuity where growth meets the discount rate. It fetches no market data.

What this does not cover

• Educational only. Nothing here or in the tool is advice, a recommendation, a rating or a signal.
• No market data, live or historical. Every input is typed by you.
• Outputs are no better than the assumptions behind them, and the model cannot tell you whether those assumptions are reasonable.
• One capital structure and one discount rate are applied across the whole forecast period.
• Not suitable for banks, insurers or any business where free cash flow is not the right measure.
• The perpetuity growth method is extremely sensitive to the gap between the discount rate and terminal growth; small input changes produce large swings.