Valuations and modelling: comps, DCF and the football field, sourced

A valuation is a chain of choices: which peers, which multiples, how to calendarise, which outliers to drop, which forecast, which discount rate, which terminal assumption. The football field at the end is only as good as the record of those choices, and in most models that record lives in someone's head or in a cell comment.

This playbook walks through how a team builds a valuation in Aqmen so that the record lives with the model. Four properties hold: every multiple and every input cites its filing; scenarios are named assumptions held side by side; the template and formulas are yours and export editable; and when the peer set is refreshed, the field redraws.

Set up

The valuation is a project in the client's workspace, with the question in its notes (a fairness range, a bid price, an impairment test) because the question decides the method. Collections hold the comps, the DCF and the Q&A. Filings, broker notes and market-data exports are sources, stored and cited.

Workstream 1: comps screening

The peer set is a dataset: one row per company, with enterprise value, revenue and EBITDA for the periods you need, each figure citing the filing or export it came from. Transaction comps are a second dataset with the same discipline: deal, date, consideration, the target's financials, and the announcement or filing behind each.

Multiples are calculated columns on top of those tables, under rules you state once:

• Calendarisation. Fiscal years that end in March or June are brought to a December basis by a stated method, applied to every peer alike.
• Screening. Size bands, business-mix thresholds and outlier rules are written down where the multiples are computed, not applied by deleting rows.
• Medians and quartiles are derived from the screened set, so changing the set changes them.

Valuation

Trading comparables

	EV · €m	EV / Revenue	EV / EBITDA
Peer A	2,140	4.1x	18.2x
Peer B	980	3.4x	15.9x
Peer C	3,610	5.2x	21.0x
Median		4.1x	18.2x

Implied value range · €m

Trading comps

Transactions

DCF

1. 1Filings and market data; calendarised to December.

The football field is drawn from those tables. Add a peer, drop an outlier, update a quarter, and the medians, the implied ranges and the field follow. Nobody re-pastes a chart.

Aqmen has no live market-data feed. Prices and consensus come from files you load or from your assistant's research; either way they arrive as cited sources, dated, so the field always says as of when it is.

Workstream 2: the DCF

The DCF is a spreadsheet in your template, connected to its inputs: forecast drivers from the operating model or the plan, the discount rate built up from its components, the terminal value under a stated method. Every formula is visible as written; nothing is hidden in generated logic you cannot read.

Three habits keep it honest:

1. Inputs cite their source. The risk-free rate, the equity risk premium, the beta and its peer set, the tax rate: each input cell traces to a dataset and the document behind it.
2. Checks catch what does not tie. The balance sheet balances, cash flows bridge to the change in net debt, the implied exit multiple is shown next to the perpetuity-growth value so an implausible terminal assumption is visible.
3. Sensitivities are scenarios, not copied tabs. Discount rate, terminal growth, margin path: each case is a named set of assumptions on the same model, so the base case, a downside and management's case sit side by side and any of them can be explained.

Your assistant can work the model with you over MCP: apply a case, read what changed, and write the effect into the notes, with the chart.

Your assistant · over MCP

Apply a downside to 2027. What changes?

• Read the model and its assumptions
• Halved adoption growth in the downside case
• Redrew the market chart

In the downside, 2027 is 6% below the base case, all of it from adoption. I added that to the findings document, with the chart.

Workstream 3: valuation Q&A

The open questions for management are drafted in a document from the model. Each question is tied to the cell it concerns: the margin step-up in year three, the capex assumption, the working-capital days. Answers are written beside them, with their evidence, so the model's caveats are on the record rather than in an email thread.

• Each question links to the cell it concerns.
• Answers are recorded with their evidence.
• Open points stay visible until they are closed.

When an answer changes an input, the figures that depend on it, in the model, the document and the deck, are flagged until they are reviewed.

Where did that number come from?

Any value on the football field walks back: the range, the multiple behind it, the screened peers, each peer's figures, and the filings they cite. Any DCF output walks back to its inputs and their sources. The method that derived a figure is stored with it, and a confidence is stated where an input is a proxy or an estimate. Before a source is replaced, you can see everything it touches.

Hand over the workspace, not a file

The valuation deck still comes out in your house style, with charts added by source from the model, flagged when stale, and exported editable. The client can also join the workspace as a guest: open the model, see every source, and ask the next question ("what if the peer set excludes the two largest?") through their own assistant, from the same data.

Chat, workbook or workspace

What it runs on

Aqmen works inside the assistant you already use (Claude, ChatGPT or any MCP-capable assistant). The step-by-step practice is in the open Aqmen skills. Models, documents and decks export editable, in the formats your clients use. The workspace is private to your organization and never used to train models.

To see a valuation built this way on your own template, talk to our team.