Comparable company analysis (trading comps)

Key takeaways

• Comps value a target by applying the multiples at which similar listed companies trade to the target's own metrics.
• Peers should match on business model, end markets, growth, margins, scale and geography; a smaller, closer set beats a larger, looser one.
• Every peer must be spread on the same basis: same period (calendarized), same adjustments for non-recurring items, same EV bridge and diluted shares.
• Use the median and interquartile range rather than the mean and full range, because outliers pull the mean and extremes.
• Apply EV multiples to get implied EV, bridge to equity value, then divide by diluted shares computed at the implied price.
• Trading comps reflect minority stakes in liquid shares; they do not include a control premium and may need adjustment for a private target.

What comps analysis does

The premise is the law of one price applied loosely: similar assets should trade at similar prices relative to what they earn. If listed packaging companies with comparable growth and margins trade at around 8 to 9 times EBITDA, a packaging company with similar characteristics should be worth something in that range too.

Comps produce a range, not a point. The output is typically the interquartile range of each multiple applied to the target, shown alongside other methods on a chart called a football field. The decision maker then judges where within the range the target belongs, based on how it compares with the peers on growth, margins, risk and quality.

Comps measure the value of a small, liquid, non-controlling stake, because that is what trades on an exchange every day. An acquirer buying control usually pays a premium to that price. The companion method, precedent transactions analysis, uses multiples paid in past acquisitions of whole companies and therefore includes control premiums. The two methods answer different questions and should not be averaged without thought.

Selecting the peer group

Peer selection is the step that most determines the answer, and the one with the most judgment. Start from business characteristics: what the company sells, to whom, and how it earns money. Then narrow by financial characteristics that drive multiples: expected growth, margins and returns on capital, leverage and cyclicality. Finally, consider scale, liquidity of the shares and geography, since investors often pay differently for small or thinly traded companies and for companies listed in different markets.

Sources for a first list include the target's own disclosures about its competitors, the peers named by research analysts, industry classification systems such as GICS, and the companies that bid in recent auctions for similar assets. That list is then tested one company at a time. A company that shares the target's industry label but has a different business mix (for example, a packaging manufacturer that also owns a large forestry operation) may belong in a secondary tier or out of the set.

Many practitioners tier the group: a core set of closest peers that carries most weight, and a broader set that shows where the sector trades. Five to ten peers is a common working size, but there is no correct number. Four genuinely comparable companies are more informative than fifteen loosely related ones.

Spreading the comps: calendarization

Spreading means building each peer's inputs on an identical basis. The first requirement is the same time period. If the target and most peers have December year ends but one peer's fiscal year ends in March, its fiscal year figures cover a different stretch of the business cycle. Calendarization converts them to a calendar year by weighting the fiscal years that overlap it.

A calendar year that runs January to December overlaps three months of the fiscal year ending in March of that year and nine months of the fiscal year ending the following March. Weight each fiscal year by the number of months it contributes. For quarterly reporters, a more precise approach is to add the relevant reported and forecast quarters directly.

The same logic applies to LTM figures, which should be computed from each company's latest filings, and to share prices and balance sheets, which should be taken as of the same date for every company in the table.

Spreading the comps: non-recurring items and the bridge

The second requirement is the same definition of earnings. Reported figures include items that will not recur: restructuring and severance charges, impairments, litigation settlements, gains and losses on disposals, and acquisition costs. Remove them from each peer's EBITDA, EBIT and net income, on the same principles for every company. For net income, adjust on an after-tax basis. Forecast figures from data providers are usually already on an adjusted basis, so check that the historical figures used alongside them match.

Be skeptical of company-defined adjustments. A charge that appears as restructuring every year is a cost of doing business. Share-based compensation is a real cost that some companies exclude from adjusted figures and others do not; choose one treatment for the whole table.

The third requirement is a consistent enterprise value bridge and share count: fully diluted shares at the current price, the same treatment of leases (matched to the EBITDA definition), pensions, non-controlling interest and associates for every peer, and balance sheet figures from each company's latest filing.

The comps table and summary statistics

With inputs spread, compute each peer's multiples and summarize them. The mean (simple average) is sensitive to outliers: one peer trading at a very high multiple, perhaps because of takeover speculation or depressed earnings, pulls it upward. The median, the middle value when the multiples are sorted, is not affected by how extreme the extremes are, which is why it is the usual central measure.

The spread is summarized with the 25th and 75th percentiles, also called the first and third quartiles. Half of the peers sit between them, so this interquartile range is a natural benchmark range that excludes the most extreme observations on both sides. There are several conventions for computing percentiles. The one used below, linear interpolation between ranked values, is the method of the QUARTILE.INC and PERCENTILE.INC functions in Excel.

For P/E, the harmonic mean is sometimes shown as well, since it averages earnings yields rather than multiples and is less inflated by high values. Loss-making peers are shown as NM and excluded from the statistics.

Applying the range: implied value for Marlowe Packaging

Marlowe Packaging is the hypothetical target. The analyst applies the 25th percentile, median and 75th percentile EV/EBITDA multiples to Marlowe's calendar 2026 EBITDA to get implied enterprise value, then bridges to equity value using Marlowe's own balance sheet.

Marlowe has employee options, so its diluted share count depends on the implied share price. The treasury stock method adds N - N x K / P new shares, which makes the calculation circular. When all the options are in the money, the circularity has an exact solution: the implied price equals implied equity value plus the option exercise proceeds, divided by basic shares plus all the options. This works because the TSM assumes the proceeds are used to buy back shares at exactly that price.

The football field

The results are summarized in a football field: one horizontal bar per method, spanning its implied share price range, with the current share price or offer price drawn as a vertical line. The table form below carries the same information. Presenting the full low to high range alongside the interquartile range shows how much of the conclusion depends on excluding the extremes.

Reading it is a judgment, not a calculation. The three interquartile ranges overlap between $14.90 and $16.14. The medians sit between $15.09 and $16.70. If Marlowe grows faster than its peers or has better margins, a position toward the upper end is defensible; if it is smaller, more leveraged or less diversified, the lower end. The reasoning belongs in the memo, next to the chart.

Common mistakes

Most errors in a comps analysis come from inconsistency rather than arithmetic. A multiple is a ratio of two numbers, and each number has several definitions; any mismatch between peers, or between the peers and the target, flows straight into the implied value. The table below lists the recurring ones and the check for each.

Beyond mechanics, the conceptual trap is to treat the output as a fact. Comps tell you what the market pays for similar companies today. If the sector is richly or cheaply valued, comps will say the target is too. Pair comps with an intrinsic method whenever the decision is large.