Quant Memo
Core

Selecting a Defensible Comparable Set

A multiples valuation is only as good as the peer group it's built on — pick companies that aren't truly comparable and the "market-based" number is really just an opinion wearing a spreadsheet.

Prerequisites: FCFF vs FCFE: Which Cash Flow to Discount

Ask two analysts to value the same mid-size software company using comparables and they can arrive at valuations 40% apart — not because they made arithmetic mistakes, but because they picked different peer groups. One included a slower-growing enterprise incumbent trading at 6x revenue; the other included a hot, fast-growing niche player trading at 14x. Same target company, same method, wildly different answer, because the method's entire output depends on a judgment call that happens before any multiplication.

A comparable set should be built from companies that are similar in the things that actually drive the multiple — growth, margins, risk, and capital intensity — not just companies that share an industry label. Getting the peer group wrong makes every downstream number look precise while being arbitrary.

What "comparable" should actually mean

Same-industry membership is a starting filter, not the answer. Two companies in the same industry can deserve very different multiples if they differ on the variables that actually determine a multiple: growth rate (higher growth generally justifies a higher multiple), margins and returns on capital (more profitable, more capital-efficient businesses justify higher multiples), and risk (more leveraged or more cyclical businesses justify lower multiples for the same growth). A slow-growing, capital-intensive telecom operator and a fast-growing, asset-light software company both sit in "technology," but they are not comparable on any dimension that drives a multiple.

Practical screens that narrow a raw industry list into a defensible set: similar revenue growth (within a reasonable band), similar gross or operating margins, similar leverage, similar business model (subscription vs. transactional, for instance), and, where feasible, similar geographic exposure.

revenue growth EV/EBITDA target wrong peers: too different on growth
A defensible peer group clusters near the target on the variables that drive the multiple, not merely on industry classification.

Worked example

Valuing a mid-cap specialty retailer with 8% revenue growth and 15% EBITDA margins.

  1. Raw industry screen pulls 20 "retail" companies, with multiples ranging from 4x to 16x EV/EBITDA — too wide a spread to be useful without further filtering.
  2. Apply growth filter: keep only companies growing 5–12% (excludes hyper-growth e-commerce names and declining legacy chains). This narrows to 11 companies.
  3. Apply margin filter: keep only companies with EBITDA margins between 12% and 18%. This narrows further to 6 companies.
  4. Median multiple of the final set: 9.2x EV/EBITDA, with a tight range of 8.1x–10.4x — a far more defensible number than the 4x–16x raw industry range.
  5. Applying 9.2x to the target's EBITDA of $40 million gives an enterprise value of 40 \times 9.2 = \368$ million, a figure an analyst can actually defend line by line against a challenge on peer selection.

What this means in practice

Every comparable-company analysis should show its work: list the companies excluded and why, not just the final set, since the exclusions are exactly where disagreements happen in practice — in a pitch, a client challenge, or an interview. Small peer sets (under 5–6 companies) are more sensitive to one outlier and should be reported with the full range, not just the median.

It's tempting to quietly include or exclude a company because its multiple happens to support the valuation you want to reach — sometimes called "comp shopping." A defensible set is built from the screening criteria first, before anyone looks at what multiple each candidate implies for the target.

Related concepts

Practice in interviews

Further reading

  • Damodaran, Investment Valuation (ch. 'Relative Valuation')
ShareTwitterLinkedIn