> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cf0.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# DCF walkthrough

> Build a discounted-cash-flow valuation on NVDA in one Lab thread, with WACC built from live market data and clarifying gates on the assumptions that matter.

A standard analyst task: *"Build a DCF on NVDA. Use the last five years of 10-Ks for the historical, current consensus for forward revenue, and a 10% / 7% / 4% three-stage growth schedule. WACC from market data. Show Bear / Base / Bull fair value."*

In Excel that's a half-day with manual data pulls. In cf0 it's three prompts.

## Step 1 — open Lab and ask

Open **Lab** from the left sidebar. Either type the question in plain English, or invoke the system Skill directly:

```
/dcf-model NVDA
```

The Skill runs but cf0 doesn't just ship a number — it pauses on the assumptions that change the answer the most.

<Frame caption="Lab — NVDA DCF with WACC build, three-stage growth schedule, and Bear / Base / Bull fair-value cards. Click any number to see the source.">
  <img src="https://mintcdn.com/cf0ai/MLKVdlfco0_3MhGd/images/lab-nvda-dcf-dark.jpg?fit=max&auto=format&n=MLKVdlfco0_3MhGd&q=85&s=a48731c5379f569bf84cbc31f0656c95" alt="cf0 Lab — NVDA DCF analysis" width="1600" height="900" data-path="images/lab-nvda-dcf-dark.jpg" />
</Frame>

## Step 2 — review the clarifying gates

For a DCF, the inputs that swing fair value most are **WACC**, **terminal growth**, and **forward revenue growth**. cf0 builds defaults from market data and consensus, then asks before committing:

* **WACC** — built from current risk-free rate, equity risk premium, and the company's beta (with the build shown alongside).
* **Forward revenue growth** — three-stage default with the option to override per stage.
* **Terminal multiple** — sector median by default; override if your thesis differs.

The model **pauses** at this point. Accept the defaults to keep going, or override and re-run. The output isn't generated until you've signed off on the inputs.

## Step 3 — read the fair-value range

Once the assumptions are committed, Lab streams three cards: **Bear**, **Base**, **Bull**. Each shows the implied fair value, the upside / downside to current market price, and the assumption it's most sensitive to.

<Frame caption="The DCF sensitivity tornado — fair-value swing per ±1σ change to each input, ranked. Terminal growth and WACC dominate; capex/sales is rounding. Tells the analyst which assumptions to scrutinise before showing the PM.">
  <img src="https://mintcdn.com/cf0ai/Tr576YnoqImeEPZS/images/brand/charts-sensitivity.png?fit=max&auto=format&n=Tr576YnoqImeEPZS&q=85&s=c72a4df5bbb0c080513d5a88f0f99c8d" alt="cf0 DCF sensitivity chart — fair value per share, tornado view" width="1280" height="511" data-path="images/brand/charts-sensitivity.png" />
</Frame>

Below the cards: the **WACC build & assumptions** card (auditable), the **revenue projection table** with QoQ and YoY deltas, and a **sensitivity matrix** of fair value across WACC × terminal-growth pairs.

Every number cites the underlying filing or data feed. Click any figure to open the source.

## Step 4 — generate a one-pager for the PM

In the same thread:

> *Generate a one-page DCF brief for the PM. Lead with the Bull / Base / Bear range, include the WACC build, the assumptions table, and the sensitivity matrix. Flag the two assumptions the answer is most sensitive to.*

cf0 reads the conversation and drafts the report. Edit any section inline before exporting the PDF. The charts and tables aren't editable — they reflect the live thread data — but the prose is. See [Reports](/features/reports) for the full editor.

## Step 5 — save the workflow as a Skill

If "run a DCF with our firm's WACC build and ship a one-pager" is recurring, save it as a custom Skill from the [Skills](/features/skills) page. Org admins can share it across the team. Future runs invoke with `/firm-dcf AAPL` — same WACC build, same three-stage schedule, same one-pager template.

## What this used

This example used every part of cf0:

* **SEC + global filings** — NVDA 10-Ks, consensus revenue, current market data feed.
* **Lab** — the surface where the question came in, the WACC build that rendered live, the report draft.
* **Reports** — branded PDF assembly from the Lab thread.
* **Skills** — saved the firm's specific WACC and three-stage schedule for the next run.

Citations trace every number to source. Reports go out in your firm's brand template.

## Next walkthrough

→ [Build an IC memo](/examples/ic-memo)