Reverse DCF and Expectations Investing

Invert the model, and the argument changes character

A conventional discounted cash flow starts from assumptions and produces a value, which makes it a generator of numbers and a magnet for anchoring: pick the growth rate you like and the model agrees with you. Inverting it removes that freedom. Hold the discount rate and the terminal growth, and solve for the one unknown — the cash flow, the growth rate, or the number of years of advantage — that makes the model equal the price. The output is a single sentence that can be argued with: at a 9% discount rate, the price assumes $852m of perpetual cash flow, which is four times what the business earns today, which over ten years is about 15% growth a year and then 2.5% forever. That reframing is the whole contribution of expectations investing, and it has three practical consequences. It forces you to state assumptions in the units the market is actually using, so a disagreement becomes a disagreement about a specific quantity rather than about whether a stock is good. It exposes the discount-rate dependence of the answer, because the same price implies a more heroic forecast at a higher rate — which is why rising long-term yields compress valuations without any change in the business. And it makes the terminal value explicit: in most models the terminal value is between two-thirds and four-fifths of the total, so the assumptions about the years beyond the forecast horizon, which are usually the least examined, decide most of the answer. The discipline is to vary one input at a time and watch how far the implied forecast has to move before you would no longer believe it. If the price requires 15% growth a year and the industry has never grown faster than 9%, the gap is large enough to be the basis of a position. If it requires 15% and the company has contracted 15% growth for three years, there is no disagreement to trade — the price is right about something you already knew, and the interesting question is what it implies about the following decade, not this one. One price, four implied forecasts — At a 9% discount rate: about $852m of perpetual cash flow, 4.06× today · At a 10% discount rate: about $983m, which needs 16.7% growth a year for a decade ← · If the shares fell 20%: about $699m implied — 3.3× today, a different argument entirely · Terminal value share: usually two-thirds to four-fifths of the total, so the horizon matters most The answer is only as good as the discount rate, and a higher rate flatters whatever you are trying to prove. Run the inversion at three rates and three terminal growth assumptions, and treat the spread as the honest uncertainty.

From an implied forecast to a position

The output of the inversion is a set of required conditions, and the next step is to test each one against evidence that is specific rather than thematic. If the price needs a mature margin of 30% in an industry whose best operator earns 18%, that is a falsifiable claim, and the work is to find out why the company would exceed the best operator — a structural cost advantage, a contractual revenue base, a regulatory moat. If the price needs 15% growth for ten years, the test is the size of the market and the share it implies, which is arithmetic rather than judgement (F15 for the machinery on young companies). Then the sensitivity: which of the assumptions is the argument actually resting on? A company whose price is mostly justified by near-term cash flows is a different position from one whose price is mostly terminal value. The second is a bet on durability, which is exactly the kind of claim that competitive analysis can support or undermine, and it is also the kind that a change in the discount rate damages most. Writing down the two or three conditions that must hold, and the evidence that supports each, converts an opinion into a thesis with a stated way of being wrong. Finally, compare the implied forecast with the consensus, which is the same exercise performed by other people. If your implied growth is below the street’s, you are taking the other side of a crowded view and the question is what you know that they do not — a channel check, a segment-level analysis, a different reading of the industry’s capacity. If it is above, you are the one who has to explain the optimism. What the method cannot do is tell you the answer; it can only tell you what you are disagreeing about, which is more useful than a price target, because a disagreement about a quantity can be settled by evidence. • Test each required condition against a specific piece of evidence, not a theme. • Find which assumption the price actually rests on, and how much is terminal value. • Compare your implied forecast with the consensus — the disagreement is the position. • State the conditions that must hold and the evidence that would falsify each one. The method does not remove the need for judgement; it relocates it. Instead of choosing a growth rate you find comfortable, you are deciding whether a specific quantity is too high or too low relative to something observable.

Expected return is what you paid, not what you hope

The inverted model gives you the market’s forecast, and the next step is the one that turns analysis into a decision: your own expected return, computed from the price rather than from the story. It decomposes into three parts — the **cash yield** you receive while holding (dividends plus net buybacks relative to price), the **growth** in cash flow per share the business can sustain, and the **change in the multiple** the market pays. The first two are properties of the company and the price; the third is the crowd, and it is the only one you do not control. Two disciplines follow. First, build the expected return from the price you can actually pay, not from a fair value you have argued for; a stock that is “worth” $120 at $100 has an expected return that depends on whether the market ever agrees, and a margin of safety is the acknowledgment that it might not. Second, make the third term explicit and small. A forecast that needs the multiple to expand to produce its return is a forecast that leans on an opinion, and opinions reverse. This framing is also how you compare across ideas. Two stocks with the same expected return are not equivalent if one achieves it through a high cash yield and the other through an assumed re-rating: the first is close to arithmetic, the second is a bet on other people. Ranking by expected return *and* by how much of it is opinion is what makes a research process repeatable rather than a collection of good stories. The same headline return, two sources — Cash yield 4% + growth 5% + multiple −1%: ≈8%, most of it from the business · Cash yield 1% + growth 3% + multiple +8%: ≈12%, leaning almost entirely on a re-rating ← · Which one survives a flat market: the first A useful habit: write the expected return as a sentence with three numbers in it. If you cannot, the forecast is a feeling with a target price attached.

What the forecast has to be built with: capital, margin and the share count

An inverted model produces a growth rate, and the growth rate is only half a forecast. Cash flows have to be produced by capital, and the relationship is the one from the moat lesson written as an identity: **growth equals the return on invested capital multiplied by the reinvestment rate**. So an implied forecast of fifteen percent a year for ten years is not one number, it is a choice between two very different businesses. If the company earns a thirty percent return on capital, it can grow at fifteen percent while reinvesting half its profit, and the free cash flow the inversion solved for is genuinely free. If it earns ten percent, the same growth requires reinvesting one and a half times its profit, which is more than it earns — so the capital has to come from somewhere else, and the inversion has quietly assumed the answer is free. Working the arithmetic makes the size of the assumption visible. Take a company with two billion of revenue growing fifteen percent a year for ten years, which takes it to eight point one billion — a six point one billion increase. If the business needs a dollar of invested capital for every dollar of new revenue, the incremental capital is six point one billion. That money arrives from operating cash flow if the company is already highly profitable, from debt if its balance sheet has room, or from issuing shares. Only the first is free in the sense the inversion assumed, and the third is not free for the shareholder even when it is free for the company: issue shares to fund the growth and the per-share cash flow the model solved for is diluted by exactly the number of new shares. The version of the inversion that is worth running therefore solves for per-share cash flow with an explicit share-count path, because a growth forecast funded by equity issuance looks considerably better before the dilution than after it. Two further variables are implied by the price and rarely named. The first is the **margin**: a price can be justified by fifteen percent growth at today’s margin or by eight percent growth with a terminal margin two hundred basis points higher, and those are different claims about the business — the first is about demand, the second about pricing power and cost discipline, and they fail for different reasons. The second is the **discount rate**, which the previous read already flagged: at a higher cost of capital the same price implies a more heroic forecast, so any statement of the implied growth is conditional on the rate used to extract it. With growth, margin, capital intensity, the share count and the discount rate all available and one price to explain, the honest output of an inversion is not a number but a short list of conditions that must all hold together. That list is also the monitoring tool, and it is what separates expectations investing from a valuation exercise. Each condition is checkable against a public report: did the margin hold, did the revenue growth arrive, did the capital intensity stay where the forecast needed it, did the share count grow. When one breaks, the thesis is falsified on that quarter’s evidence rather than on the price, which means the position can be reduced on the reason it was taken rather than after the market has already repriced it. The failure mode the discipline prevents is the common one: holding a fast-growing company through a margin decline because the growth is still there, when the growth was only ever worth paying for at the margin the price assumed. The same implied growth, two businesses — 15% growth at a 30% ROIC: Reinvestment rate = 15 ÷ 30 = 50% of profit — the rest is free cash flow · 15% growth at a 10% ROIC: Reinvestment rate = 150% of profit — the growth cannot be funded out of earnings ← · $2.0bn to $8.1bn of revenue in ten years: $6.1bn of new capital has to come from cash flow, debt or new shares ← · The same forecast with equity funding: The per-share cash flow the inversion solved for is diluted by the new shares Run the check as a pair of numbers rather than a paragraph: implied growth, and the reinvestment rate that growth requires at the company’s actual return on capital. If the second exceeds one hundred percent of profit, the forecast is not a growth forecast — it is a financing forecast, and it should be written that way.

Sources

• Expectations investing: reading the forecast inside the priceRappaport & Mauboussin, "Expectations Investing" (2001, rev. 2021)
• Reverse-engineering the growth embedded in a multipleDamodaran, "The Little Book of Valuation"; standard practice in equity research
• The weight of terminal value in a discounted cash flowStandard DCF practice; McKinsey, "Valuation"
• Discount-rate sensitivity and the level of long-term ratesMacro & Rates MR4 on the curve, MR13 on duration and MR15 on the term premium; standard rate-sensitivity analysis