O
Occam's Investing
102 Beginner · Financial Modeling

Discounted Cash Flow (DCF) Models

A company is worth the cash it will produce. This class builds a discounted cash flow model from the ground up — then shows you how to source every input and verify it before you trust it.

Instructor: John Gillespie
Tool: Perplexity
Format: Live class + slides
Level: Beginner · 1.5 hrs
Live class recording · published 2026-06-12 Watch on YouTube →  ·  Download slides (PDF)

What you will learn

Read the anatomy of a DCF, build a basic one yourself, and source the inputs responsibly. By the end you can run a five-year DCF on a single company and defend every assumption.

Section 1 · 10 min

The Core Idea

Why a business is worth its future cash, discounted back to today. Cash that arrives later is worth less now — so we shrink it. Earnings can be shaped; cash is harder to fake.

Section 2 · 30 min

The Five Building Blocks

Forecast free cash flow, the discount rate, terminal value, present-value mechanics, and model structure — and how each assumption links to the valuation output.

Section 3 · 30 min

Build a DCF — Worked Example

A single company, short horizon, in a structured spreadsheet. Project cash, discount each year, add the terminal value, and read out an intrinsic value.

Section 4 · 15 min

Source Inputs with Perplexity

AI-assisted research to gather your inputs fast — and the cite-or-verify discipline that keeps a hallucinated number out of your model.

Section 5 · 5 min

Pitfalls & Sanity Checks

Where DCFs go wrong: unrealistic growth, terminal-value dominance, discount-rate sensitivity, and false precision. The checks that keep a model honest.

Section 1 · The Core Idea

A company is worth the cash it will produce

A DCF estimates intrinsic value: the present value of every dollar of free cash flow a business is expected to generate, discounted back to today. Cash that arrives later is worth less now — so we shrink it. Earnings can be shaped by accounting choices; cash is harder to fake, which is why a DCF starts there.

Present value of $1, discounted at ~9% Yr 1 → 92¢  ·  Yr 2 → 84¢  ·  Yr 3 → 77¢  ·  Yr 4 → 71¢  ·  Yr 5 → 65¢
Section 2 · The Five Building Blocks

Five pieces, one valuation

Every DCF is the same five parts wired together. Get the structure right and the model stays honest: change one assumption and the output updates, cleanly and traceably.

01

Forecast Free Cash Flow

The cash a company keeps after running and reinvesting in itself. Start from operating cash flow, then subtract capital expenditures.

02

Discount Rate

The annual return required to compensate for time and risk. A higher rate makes future cash worth less today. At beginner level it is stated, not derived.

03

Terminal Value

Everything beyond the explicit forecast window, collapsed into one number at the end — then discounted back like any other cash flow.

04

Present-Value Mechanics

Shrinking each future cash flow back to today's dollars by dividing by the discount factor for its year.

05

Model Structure

How assumptions flow through drivers to the valuation output. Keep every input in one labeled block — never bury a number inside a formula.

Free cash flow FCF = Operating Cash Flow − Capital Expenditures  ·  each year: FCFt = FCFt−1 × (1 + g)
g = your assumed annual growth rate for cash flow — keep it modest and defensible, not heroic.
Present value & terminal value PV = CFt ÷ (1 + r)t  ·  TVn = [ FCFn × (1 + g) ] ÷ (r − g)
r = discount rate, g = perpetual growth rate. Keep terminal g low — at or below long-run GDP.
Section 3 · Worked Example

A five-year DCF, start to finish

Same machinery every time: project cash, discount each year, add the terminal value, and sum. The figures below are hypothetical — chosen to illustrate the mechanics, not to represent any specific company.

Assumptions Base FCF $120M  ·  Growth (g) 8%/yr  ·  Discount rate (r) 10%  ·  Terminal growth 3%
YearFCF ($M)Discount factor @10%Present value ($M)
1120.00.909109.1
2129.60.826107.1
3140.00.751105.2
4151.20.683103.2
5163.30.621101.4
TV2,4020.6211,491.6
Intrinsic value PV of forecast cash flows ($526M)  +  PV of terminal value ($1,492M)  →  ≈ $2.02B
Section 4 · Source Inputs with Perplexity

Gather fast. Verify before you trust.

A DCF is only as good as its inputs. Use Perplexity to gather them quickly — then confirm each number against the filing. Treat AI answers as leads, not citations.

Prompt 01 · Pull the cash inputs
List {TICKER}'s operating cash flow and capital expenditures for the last 5 fiscal years, with the 10-K source for each.
Prompt 02 · Frame the growth assumption
What free cash flow growth rate have analysts assumed for {TICKER}? Cite sources.
Prompt 03 · Stress the forecast
Summarize the risks that could change {TICKER}'s cash generation over the next 5 years.
Cite-or-verify discipline
  • Open the primary source Perplexity links to and confirm the number
  • Pull cash figures from the cash-flow statement, not summaries
  • Log where every input came from — your model should be auditable
  • If a figure cannot be sourced, do not use it
Why this matters
  • In 2023 a 30-year attorney filed a brief citing six cases an LLM invented — none existed
  • Asked to confirm them, the model doubled down and generated fake opinions
  • Every input feeds the output: one hallucinated number quietly corrupts the whole valuation
  • The discipline is the same for every model — anchor each number to a primary source
Section 5 · Pitfalls & Sanity Checks

Where DCFs go wrong

A DCF is a reasoned estimate, not a price target. The same four failure modes catch beginners and professionals alike — in the worked example above, roughly 74% of the value came from terminal value, which is exactly the kind of thing to stress-test.

01

Garbage in, garbage out

Unrealistic growth quietly inflates the whole answer. Keep g modest and defensible.

02

Terminal-value dominance

If most of the value sits in TV, your perpetual-growth assumption is doing the heavy lifting.

03

Discount-rate sensitivity

A one-point change in r can swing value by double digits. Always test a range.

False precision A DCF is a reasoned estimate, not a price target — report a range, not a single number.
Even sophisticated shops publish bull-and-bear DCF outputs that span multiples of each other. The model's job is to make your assumptions visible, not to manufacture certainty.

Follow the cash. Not the noise.

You have built the muscle for valuation by cash flow. Next: derive the discount rate from WACC, and run a reverse DCF to see the growth the market is already pricing in.