Discounted cash flow modeling, terminal value, scenario and sensitivity analysis, and the reverse DCF — the mechanics behind every IU Digest valuation, worked through with real numbers.
From Free Cash Flow to Intrinsic Value
A discounted cash flow model is nothing more than a disciplined answer to one question: what is a stream of future cash worth today? Every input is an assumption — this guide shows you which ones matter, and how IU builds and stress-tests them.
Four modules — work through them in order or jump to the one you need.
"Revenue is a headline. Net income is an accounting construct. Free cash flow is what's left after the business pays for its own survival — and it's the only thing a DCF can legally discount."
If a friend asked to borrow money and pay you back over five years, you wouldn't just ask "how much will you make?" You'd ask "how much will actually be left over, after your rent and bills, to pay me back?" That leftover cash is the entire idea behind a DCF (Discounted Cash Flow model). Instead of valuing a company off the revenue or profit headline it reports, a DCF asks a simpler question. After this business pays for everything it needs just to keep running — taxes, equipment, day-to-day operating costs — how much real, spendable cash is actually left over each year? That number is called free cash flow (FCF). It's the only thing a DCF is allowed to work with. Once you have that yearly leftover-cash number, you translate future dollars into today's dollars using a discount rate called the Weighted Average Cost of Capital (WACC) (covered in full in the CAPM & WACC guide). Why bother? Because a dollar arriving in year five is worth less to you than a dollar in your pocket right now.
Start with Earnings Before Interest and Taxes (EBIT) — operating profit — remove the tax you'd owe if the company had no debt (this keeps financing out of the picture), add back non-cash depreciation and amortization, then subtract Capital Expenditures (CapEx), the capital actually spent to keep the business running and growing. That last piece, "Δ Net Working Capital," just means the change in cash tied up in day-to-day operations — inventory, receivables, and similar short-term items. (The Δ is the Greek letter delta, shorthand for "change in.")
Each future year's FCF gets divided by (1 + WACC) raised to that year's power. A dollar five years out is worth less than a dollar today, and a higher WACC discounts it harder. Sum the projection period, add the discounted terminal value, and the result is enterprise value (EV) — the value of the entire business, debt and equity combined. (That's a different number from the "market cap" used later in the CAPM & WACC guide, which is the value of the equity slice alone — subtract net debt from enterprise value to get there.)
A company generates $200M of EBIT on a 25% tax rate, $40M of D&A, $60M of CapEx, and a $10M working capital build. Unlevered FCF = $200M × (1 − 0.25) + $40M − $60M − $10M = $120M. Discounted at a 9% WACC, that single year of cash flow is worth $120M / 1.09 ≈ $110M in today's dollars. That same math repeats for every year in the projection.
A DCF is only as good as its FCF forecast. Revenue growth assumptions get the attention. But margin trajectory and CapEx intensity usually swing intrinsic value more than the top line — model those with the same rigor.
Next: the terminal value assumption typically drives 60–80% of total DCF value. Module 02 covers both ways to calculate it.
Next Module →"Most of a DCF's value sits beyond the last year you can forecast with any confidence. Terminal value isn't a footnote — it's usually the majority of the answer."
Nobody can honestly forecast a company's cash flow year-by-year forever. Five to ten years out is already a stretch. So a DCF forecasts what it reasonably can, then bundles "everything after that" into one lump-sum number called terminal value. Here's the part that surprises most people the first time they see it: that single bundled number usually makes up the majority of the entire valuation, often 60–80% of it. Get the near-term forecast right but the terminal value assumption wrong, and you've still gotten the valuation wrong. There are two standard ways to calculate it, and a healthy model checks both.
Assumes the business grows at a modest, sustainable rate (g) forever after the projection window — typically 2–3%, roughly long-run GDP growth. The formula is unstable as g approaches WACC, so g should always sit comfortably below it.
Applies a market-based multiple (often the industry's current EV/EBITDA) to the final projected year. This grounds terminal value in what similar businesses actually trade for, rather than a perpetual growth assumption. EBITDA stands for Earnings Before Interest, Taxes, Depreciation & Amortization — a rough proxy for the cash a business's core operations throw off, before financing and accounting decisions are layered on top.
| Method | Input | Result |
|---|---|---|
| Gordon Growth | Final FCF $150M · g 2.5% · WACC 9% | ≈ $2.37B |
| Exit Multiple | Final EBITDA $218M · 11x | ≈ $2.40B |
Two independent methods landing within 2% of each other, as shown above, is a strong signal the underlying assumptions are internally consistent. When they diverge by 20%+, that's a flag to revisit either the perpetual growth rate or the peer group of comparable companies behind the exit multiple. Don't just average the two and move on.
Always calculate terminal value both ways and reconcile them. A DCF built on a single terminal value method is one unexamined assumption away from being wrong by a wide margin.
Next: since terminal value dominates the output, the model's conclusions should always be pressure-tested across a range of assumptions — not treated as a single point estimate.
Next Module →"If your DCF spits out one exact price target, you haven't finished the analysis — you've just hidden the uncertainty inside the assumptions."
Think about how you'd actually plan a road trip's arrival time. You wouldn't tell someone "I'll be there at 3:14 PM exactly." You'd say "somewhere between 3 and 3:30, depending on traffic." A DCF deserves the same honesty. Every number that goes into it — how fast the company grows, how much of each sales dollar turns into profit, what discount rate applies — is a guess, even when it's an educated one. So instead of publishing one single price target, IU runs the model across a bear / base / bull range and builds a sensitivity table. That's a simple grid showing how the answer changes as the two assumptions that matter most — usually the discount rate and the long-term growth rate — move together.
| WACC \ g | 2.0% | 2.5% | 3.0% |
|---|---|---|---|
| 8% | $142 | $149 | $158 |
| 9% | $121 | $126 | $132 |
| 10% | $105 | $108 | $113 |
Reading this grid tells you more than the base case alone: a 2-percentage-point WACC swing here moves fair value by roughly 28% (at a fixed terminal growth rate), while a full percentage point of growth-rate uncertainty moves it by less than 10%. That asymmetry tells you exactly where to focus diligence.
Bear, base, and bull cases in IU deep dives aren't arbitrary ±10% haircuts. Each is built from a distinct, named set of operating assumptions (e.g. bear = margin compression from pricing pressure; bull = faster-than-guided capacity ramp), so the range reflects real business scenarios rather than a cosmetic band around the base case.
The width of your sensitivity range is information. A tight range across reasonable assumptions signals a resilient thesis. A range that swings from deeply undervalued to overvalued means the position's risk is in the assumptions, not the market.
Next: flip the model around — instead of solving for a price, start from the market's price and solve for what it already assumes.
Next Module →"You don't need a model to know what the market thinks. You need a model to find out whether the market is right."
Every stock's current price is already making a bet on that company's future. The market just doesn't tell you out loud what the bet actually is. A reverse DCF is how you get it to talk. Instead of building assumptions and calculating a price like a normal DCF, you flip it around. Start with today's price, and work backward to figure out exactly how much growth and profit margin the company would need to deliver to actually be worth that price. This approach has a name: Price-Implied Expectations (PIE), popularized by analysts Michael Mauboussin and Alfred Rappaport. But the concept is simple — reverse-engineer the market's assumption, then go check whether it's realistic.
Hold WACC and the discounting mechanics fixed at reasonable levels, then iterate the growth rate and margin assumptions until the model's output matches the current market price. Whatever combination gets you there is what the market is implicitly pricing in.
A stock trades at a price that only reconciles to a standard DCF if the market is assuming a 22% five-year revenue Compound Annual Growth Rate (CAGR) and margin expansion to 28%. That's well above the company's five-year historical average of 14% growth and 19% margins. That gap is a testable claim: check the backlog, capacity plans, and competitive dynamics. If they don't support 22%/28%, the market's implied expectations are the risk, not the opportunity.
The reverse DCF turns "is this expensive?" into a specific, testable question: does the business's trajectory support what the price already assumes? That reframe is often more useful than a fresh price target. It tells you exactly what has to go right.
Next: WACC has appeared in every module here as an input — the CAPM & WACC guide breaks down exactly how that number gets built.
Continue to CAPM & WACC →