DCF

DCF conventions and the errors that sink you

6 min read · updated July 2, 2026

You already know the shape of a DCF: project cash flows, discount them, add a terminal value, get to a per-share number. If you don't, read The DCF, explained first. This article is about the part that actually separates candidates. The conventions and the order of operations. This is where a smart student who "gets" the concept still fumbles the walk-through and loses the room.

Nobody flunks a Superday for not knowing what a DCF is. They flunk it for saying "discount the cash flows" and then doing it wrong when pushed.

The mid-year convention

Standard textbook discounting assumes every year's cash flow lands on December 31. That's a lie. A business collects cash all year long, roughly evenly. So on average the cash arrives at the midpoint of each year, not the end.

The mid-year convention fixes this by moving each discount period back half a year. Year 1 gets discounted at period 0.5 instead of 1.0, year 2 at 1.5, and so on.

$\text{PV} = \frac{\text{FCF}_t}{(1+\text{WACC})^{\,t-0.5}}$

Because you're discounting over a shorter time, every present value goes up a little. The uplift is roughly $(1+\text{WACC})^{0.5}$ . At a 10% WACC that's about a 5% bump to your valuation. Not huge, but not nothing, and interviewers love to check whether you know it exists.

Key insight

Mid-year convention makes the value go up, never down. If someone asks "does mid-year raise or lower your value?" the answer is always higher, because you're waiting less time to get the same cash.

One subtlety on the terminal value. If you build TV with the Gordon Growth (perpetuity) method, discount it at the mid-year period of the final year too. If you build it with an exit multiple, the multiple is applied to a full-year metric at year-end, so most people discount that TV at the full final period, not the half. Be ready to explain that split. It trips people up.

Stub periods

Deals don't close on January 1. Say it's April and the fiscal year ends in December. You've already lived through part of the year, so your first forecast period isn't a full 12 months. It's a stub. About 0.75 of a year in that case.

You discount the stub cash flow over 0.375 years if you're using mid-year (half of 0.75), then the next full year sits at period 1.25 (0.75 for the stub plus half of the next full year), and so on. The mechanics matter less than the instinct: the first period is often not a clean 1.0, and your discount factors have to reflect the calendar. Miss this and every downstream period is off.

Sensitivity tables: a DCF is a range, not a number

Here's the truth most beginners resist. Your DCF output is only as good as your two most fragile assumptions: the discount rate and the terminal growth rate. Terminal value routinely drives 60% to 80% of the total, so the whole model hangs on a couple of numbers you basically guessed.

That's why you never present a DCF as a single point. You present a grid. WACC across the top, terminal growth down the side, enterprise value in the cells.

| EV ($M) | g = 2.0% | g = 2.5% | g = 3.0% | |---|---|---|---| | WACC 9% | 1,000 | 1,077 | 1,167 | | WACC 10% | 875 | 933 | 1,000 | | WACC 11% | 778 | 824 | 875 |

Look at the spread. Move WACC one point and growth half a point and the value swings by hundreds of millions. That range is the honest answer. When a banker asks "what's the company worth?", the strong reply is "roughly $875M to $1.0B in the base zone, and here's what drives the tails," not a false-precision "$941.7M."

Interview tip

When you show a sensitivity table, always sanity-check the corners. A low-WACC, high-growth cell where growth approaches WACC blows up toward infinity and is nonsense. Flag it as unrealistic rather than reporting it with a straight face.

You should also cross-check the DCF's implied exit multiple against your trading comparables. If your perpetuity-growth TV implies a 22x EV/EBITDA exit and comps trade at 10x, your growth assumption is fantasy. That reconciliation is the single fastest way to catch a broken model.

The sequence errors that actually sink candidates

Common mistake

The killers are almost never conceptual. They're sequence and bridge errors:

Not discounting the terminal value. People compute a clean TV in year 5 and then forget it's a year-5 number. It has to be discounted back to today just like every other cash flow. Adding an undiscounted TV overstates value massively.
Wrong number of discount periods. Off-by-one on the exponent, or forgetting the stub, or applying full-year discounting when you said you'd use mid-year. Your first-period exponent and your TV exponent are where this hides.
Forgetting the enterprise-to-equity bridge. You discounted unlevered free cash flow, so the sum is enterprise value, not equity value. You still have to subtract net debt (and preferred, and minority interest) to reach equity value, then divide by shares. Stopping at EV and calling it the share price is the most common mistake in the whole exercise.

Two more that pair with the bridge. Mixing levered and unlevered. If you discount unlevered FCF, you must use WACC and you land on enterprise value. If you ever discount levered cash flow, you'd use cost of equity and land on equity value directly. Pick one lane and stay in it. The classic error is unlevered FCF discounted at WACC, then forgetting the bridge, which double-counts the equity story. See enterprise value vs. equity value for why the claim structure forces this.

And building terminal value before you've settled WACC. Your TV via perpetuity growth is $\text{FCF}_{n}(1+g)/(\text{WACC}-g)$ . You literally cannot compute it without WACC, yet people scribble a TV early and then change their discount rate later without updating it. Lock WACC first, then TV, then discount everything together.

The clean sequence, out loud

Say it in this order and you sound like you've built one:

$\text{EV} = \sum_{t=1}^{n}\frac{\text{FCF}_t}{(1+\text{WACC})^{\,t-0.5}} + \frac{\text{TV}}{(1+\text{WACC})^{\,n-0.5}}$

then

$\text{Equity Value} = \text{EV} - \text{Net Debt}, \qquad \text{Share Price} = \frac{\text{Equity Value}}{\text{Shares}}$

Interview tip

Practice the walk-through as a fixed sequence you can recite without pausing: project unlevered FCF, pick WACC, build and discount TV, sum to enterprise value, bridge to equity, divide by shares, then present a range from your sensitivity grid. The candidate who delivers that in 60 seconds, remembers the mid-year uplift, and never forgets the bridge is the one who reads as "has actually done this." That fluency, under time pressure, is the depth that separates you in a Superday.

Glossary

New to the lingo? Every term used above, in plain English.

DCF (Discounted Cash Flow): A way to value a company by projecting its future cash and discounting it back to what it is worth in today’s dollars.
Unlevered free cash flow: The cash a business generates before any debt payments, so it belongs to all investors, both lenders and shareholders. This is the cash flow used in a DCF.
WACC (Weighted Average Cost of Capital): The blended rate a company pays to fund itself with both debt and equity. In a DCF it is the discount rate used to bring future cash back to today.
Terminal Value: In a DCF, the estimated value of all the cash flows that come after the years you forecast explicitly. It often makes up most of the total value.
Exit multiple: A valuation shortcut for terminal value that applies a market multiple, such as EV/EBITDA, to the final forecast year to estimate what the business would sell for at the end.
EV (Enterprise Value): The value of a company’s whole operations, to every investor including lenders and shareholders. It does not depend on how the company is financed.
Equity Value: The slice of a company that belongs to its shareholders. For a public company this is the market capitalization (share price times shares outstanding).
Net debt: A company total debt minus its cash. It is what you subtract from enterprise value to get to equity value, since a buyer could use the cash to pay down the debt.
Mid-year convention: A DCF tweak that assumes cash flows arrive in the middle of each year rather than at year-end, since companies earn cash throughout the year. It slightly raises the valuation.
Perpetuity: A stream of cash flows that continues forever. Its value is the cash flow divided by the discount rate, and a growing perpetuity is the basis of the Gordon growth terminal value.

Make it stick

Drill what you just learned

→