Discounted Cash Flow (DCF) Part 1: Building the Model

If you can't value a business, you can't invest in it—you're just guessing with money.

PROFESSOR: Charlie, you've covered dozens of earnings calls, watched analysts grill CFOs about guidance, seen stocks move ten percent on a revenue miss. When those analysts walk out of the room or hang up the call, most of them are going back to a DCF model. Why is discounted cash flow the gold standard for valuation, especially when we have simpler metrics like P/E ratios?

CHARLIE: Because a P/E ratio tells you what the market thinks right now. A DCF tells you what the business is actually worth based on the cash it will generate over its lifetime. I've sat in investor conferences where someone says a stock trades at fifteen times earnings like that's analysis. That's a description, not a valuation. The DCF forces you to answer the hard questions—how much cash will this company produce, how certain are you about that, and what's your opportunity cost for tying up capital in this specific business instead of another one? When Amazon traded at a P/E of three hundred in the early 2000s, traditional metrics made it look insane. But if you built a DCF that captured the cash flows from AWS and the retail flywheel ten years out, suddenly that price made sense. The DCF makes you think like an owner, not a trader.

PROFESSOR: Let's build one then. First component—projecting free cash flows. Walk me through what actually goes into that forecast. And I know you've seen management teams throw around adjusted EBITDA and non-GAAP nonsense. How do you cut through that to get to real free cash flow?

CHARLIE: You start with revenue, which means you need a thesis about growth. Take a company like Shopify. You're asking—how many merchants will use this platform, what's the average revenue per merchant, how does that change as they scale? Then you work down the income statement. What are the gross margins—that tells you the unit economics. Operating expenses—you need to separate the growth investments from the maintenance costs. A company spending heavily on sales and marketing to acquire customers looks different from one burning cash on executive perks. Then you get to EBITDA, but you can't stop there because EBITDA is not cash. Depreciation and amortization might be non-cash expenses, but capital expenditures are very real cash out the door. If you're valuing a railroad, they might report strong EBITDA, but they need to spend billions maintaining track and buying locomotives. That's your capex. Subtract that. Then you adjust for working capital changes—if receivables are growing faster than payables, cash is getting tied up in the business. What you're left with is unlevered free cash flow—the cash available to all investors, debt and equity, before any financing decisions. That's what goes into the DCF. I've watched companies report record EBITDA while their free cash flow was negative. The DCF makes you confront that reality.

PROFESSOR: So you've projected five or ten years of those cash flows. Now you need to discount them back to present value. That means picking a discount rate, which is where most people's eyes glaze over. This is the weighted average cost of capital—WACC. Break down what that actually represents and why we can't just use a round number like ten percent.

CHARLIE: The WACC is your hurdle rate—it's what you could earn elsewhere with the same risk profile. It has two pieces: the cost of equity and the cost of debt, weighted by how much of each the company uses. Cost of debt is straightforward—what interest rate does the company pay on its bonds? You can look that up. But you tax-adjust it because interest is tax-deductible, so if they pay five percent interest and have a twenty-one percent tax rate, the after-tax cost is closer to four percent. Cost of equity is trickier. You use the capital asset pricing model—CAPM. Start with the risk-free rate, which is the ten-year Treasury yield, call it four percent today. Then you add an equity risk premium—the extra return investors demand for stocks over bonds, historically around five to six percent. But you multiply that premium by beta, which measures how volatile this stock is relative to the market. A utility with a beta of 0.6 gets a lower cost of equity than a semiconductor company with a beta of 1.8. Then you weight those two costs by the company's capital structure. If they're thirty percent debt and seventy percent equity, you blend the costs accordingly. I covered a leveraged buyout once where the buyers pumped in debt to lower the WACC and juice returns. But too much debt raises the cost of equity because equity holders demand more return for the risk. There's a balance. And this isn't academic—a one percent change in WACC can swing your valuation by twenty or thirty percent. That's why analysts argue about beta assumptions in pitch meetings.

PROFESSOR: Let's say we've discounted ten years of cash flows back to today. But the company doesn't evaporate in year ten. That's where terminal value comes in, and it usually represents sixty to seventy percent of the total valuation. How do you calculate something that depends on assumptions stretching into infinity without turning it into complete fiction?

CHARLIE: You have two main methods. The perpetuity growth model and the exit multiple. Perpetuity growth assumes the company keeps generating cash forever, growing at a modest rate—usually tied to GDP growth, say two to three percent. You take year ten's free cash flow, grow it by that rate, and divide by your discount rate minus the growth rate. That gives you a terminal value. The math is clean, but the assumption is heroic—you're saying this business will exist and grow steadily forever. I've seen analysts plug in four percent perpetual growth for a retailer in a dying industry. That's fantasy. The exit multiple approach says—in year ten, what would someone pay for this business? You apply a multiple, often EV to EBITDA, based on comparable companies. If peers trade at twelve times EBITDA, you assume the same and calculate terminal value that way. The problem is you're importing today's market sentiment into a future valuation. If multiples are inflated now, your terminal value is inflated. I prefer the perpetuity model for stable businesses—Coca-Cola, Visa—and the exit multiple for cyclical or high-growth companies where predicting steady-state cash flow is harder. Either way, you need to sanity-check it. If terminal value is ninety percent of your total, your model is a bet on the distant future, not the business today. That should make you nervous.

PROFESSOR: So you've got your projected cash flows, you've discounted them at your WACC, you've added a terminal value. You sum it up and get an enterprise value. Last step—how do you get from enterprise value to a price per share that you can compare to the stock price?

CHARLIE: Enterprise value is what the whole business is worth to all capital providers—debt and equity. To get to equity value, you subtract net debt. Take the company's total debt, subtract cash and equivalents, and pull that out. Why? Because if you bought the whole company, you'd have to pay off the debt but you'd get to keep the cash. What's left is the equity value—what belongs to shareholders. Divide that by shares outstanding, and you have intrinsic value per share. Then you compare it to the market price. If your DCF says the stock is worth ninety dollars and it's trading at sixty, that's a potential buy. But you need a margin of safety—I don't buy just because my model says it's undervalued. I want a significant discount to account for my mistakes in assumptions. I've built DCFs that were wrong because I overestimated margins or didn't foresee a competitor. The model is only as good as your inputs. That's why you sensitivity-test it—run scenarios with different growth rates, different WACCs, different terminal values. If the stock looks cheap in all of them, you've got conviction. If it only works with optimistic assumptions, you've got a prayer, not an investment.

PROFESSOR: See you Wednesday. A DCF doesn't predict the future—it forces you to state your assumptions clearly enough that you can be wrong precisely instead of vaguely.