Comparable Company Analysis

Comparable company analysis, known on the desk as trading comps or simply comps, values a company by benchmarking it against similar publicly traded businesses. It is the purest form of relative valuation: instead of projecting the target's cash flows, you observe what the market currently pays for comparable companies and apply that pricing to the target's financials. If genuinely similar businesses trade at 10-12x EBITDA and the target generates $200 million of EBITDA, the implied enterprise value is roughly $2.0-2.4 billion. Everything else in the methodology is the machinery for making that inference defensible.

Comps are typically the first analysis built on any engagement because they are quick, grounded in observable market prices, and immediately intelligible to clients. They appear in virtually every pitchbook, fairness opinion, and football field, and they form the relative-value pillar of valuation alongside the DCF (intrinsic value) and precedent transactions (acquisition value). The calculations themselves are mechanical; the judgment, and therefore most of the interview questioning, concentrates in selecting the peer group and interpreting the output.

The Five-Step Process

The full workflow runs in five steps, each building on the one before it:

1. 1.Select the peer group: identify 5-12 publicly traded companies genuinely similar to the target.
2. 2.Gather and normalize financial data: market data, historical results, consensus estimates, and operating metrics, scrubbed for one-time items.
3. 3.Calculate valuation multiples for each peer, on both trailing and forward bases.
4. 4.Derive statistical benchmarks: mean, median, and percentiles, with outliers investigated.
5. 5.Apply the benchmark multiples to the target's metrics to produce an implied valuation range.

Errors compound down this chain. A carefully calculated median from a badly chosen peer group is a precise answer to the wrong question, and unnormalized data lets one-time items distort every multiple downstream. This is the garbage-in, garbage-out problem, and it is why experienced bankers spend far more time on steps one and two than on the mechanical calculations in steps three and four.

Selecting the Peer Group

Peer selection is the most important and most subjective step in the entire analysis. Every subsequent step is mechanical; if the peer group is wrong, the mechanics produce meaningless output no matter how carefully they are executed. It is also the step interviewers probe hardest, because it reveals whether a candidate understands the business rather than just the arithmetic. A strong answer covers not only which companies to include but why specific candidates were excluded.

The Five Dimensions of Comparability

No two companies are identical, so the goal is not perfect matches but companies that are similar enough along the dimensions that drive valuation:

• •Industry and business model. The starting filter, but industry labels alone are insufficient. Classification systems (GICS, SIC, NAICS) are a starting point, not an answer: the GICS Software sub-industry spans enterprise ERP vendors, cybersecurity firms, and gaming studios. Within one sector, a SaaS company with 95 percent recurring revenue and 80 percent gross margins is a fundamentally different business from a hardware company with cyclical revenue and 35 percent gross margins; one may trade at 15x revenue and the other at 1.5x, and a median blending them describes neither. The operative test: do these companies compete for the same customers, in the same markets, with similar products?
• •Size and scale. Larger companies typically trade at higher multiples because of revenue diversification, pricing power, capital markets access, and lower perceived risk. The practical screen is roughly 0.3x to 3x the target's enterprise value or revenue: a $5 billion EV target maps to peers between roughly $1.5 billion and $15 billion. The screen bends in niche industries with few public players, but then the analysis should acknowledge the scale premium of a larger peer or the size-specific risks (customer concentration, limited capital access) priced into a smaller one.
• •Growth profile. Growth is one of the strongest drivers of multiples, and the trajectory matters as much as the level: a company that grew 30 percent but is decelerating toward 10 percent is a different animal from one compounding steadily at 15 percent. When the universe splits into fast and slow growers, segment it. A cybersecurity target growing 18 percent with a 22 percent EBITDA margin sits between hyper-growth peers trading at 15-20x forward revenue and mature names at 5-8x; presenting both segments, and placing the target at the upper end of the moderate-growth group, is far more informative than one blended median.
• •Profitability and margin structure. More profitable companies convert more of each revenue dollar into cash flow, and the market pays for that. Match gross margin (the business model's inherent economics), EBITDA margin, capital intensity (capex-heavy businesses convert less EBITDA into free cash flow), and margin trajectory, because the market prices the expected path, not just the current level. Growth and profitability trade off against each other, which is why growth-adjusted frameworks like the PEG ratio and the Rule of 40 exist; both are covered below.
• •Geographic exposure. Geography shapes growth rates, regulation, tax, and currency risk. Cross-border comparables bring accounting differences (US GAAP versus IFRS treatment of leases, capitalized development costs, stock-based compensation), tax regime effects on after-tax multiples, and FX distortion of growth rates. None of that disqualifies them, and some industries force global peer groups: luxury goods are dominated by European names (LVMH, Kering, Hermes, Richemont), mining companies list across Toronto, London, and Sydney, and semiconductors span the US, Taiwan, South Korea, the Netherlands, and Japan. In those sectors a US-only peer group is not a peer group at all.

Building the Group in Practice

A credible core peer group has 5-12 companies. Most banks require at least five for the statistics to mean anything, and many prefer seven or more; a group of three is a list, not a sample, because any one company's anomaly dominates the median. The standard approach starts wide, with 15-20 candidates, and narrows.

Candidates come from a handful of sources:

• •SEC filings: competitors named in the 10-K business description and risk factors, plus the compensation-benchmarking peer group in the proxy statement
• •Equity research coverage universes, which reflect how the market already groups these businesses
• •Screening tools in Bloomberg, FactSet, and Capital IQ, filtered by industry, size, and financial profile
• •Internal peer groups from prior engagements, already vetted by senior bankers (but refreshed, never reused blindly)
• •Client management's view of who its real competitors are

Narrowing follows the dimensions above: business model and size first, then growth and profitability, then a company-by-company check for distortions such as pending M&A (the stock price reflects deal speculation rather than standalone value), major one-time events, or management upheaval. The final group is documented: a written rationale for every inclusion and every significant exclusion.

Two structural concepts help organize the result. A pure-play comparable is a company whose business is substantially the same activity as the target's, with minimal unrelated revenue; these produce the cleanest multiples, whereas diversified conglomerates produce blended multiples that obscure segment economics. Secondary or reference comparables are companies that match on some dimensions but miss on others (size, geography, business model); they are shown separately for context and excluded from the benchmark statistics, for example three or four European chemicals producers displayed beneath a core group of US specialty chemicals peers.

When Perfect Comparables Do Not Exist

Many targets occupy niches with no clean public analog. The methodology does not collapse; it adapts:

• •Broaden to adjacent industries that share the key economics (growth, margins, customer dynamics) even if the product differs
• •Add international comparables, adjusting for currency, tax, and accounting differences
• •Segment the group into closest comparables (2-3 names) and broader comparables (5-8 names), presenting both
• •Shift weight toward the DCF and precedent transactions in the overall triangulation
• •Regress multiples against growth or margins across a wider universe to estimate where the target should trade, a technique more common in equity research but useful when direct comps are thin

Weak comparability should widen the output range, not narrow it. Eight closely matched peers can support a tight 10-11x range; a loose group supports something more like 8-14x. Presenting a tight range from a weak peer group is false precision, and a wide comps bar honestly signals that other methodologies should carry more weight.

Pitfalls That Invalidate the Analysis

Four failure modes recur. A too-broad group (15+ loosely related names) buries the signal in noise: the median describes an average of a diverse portfolio, not any real company. A too-narrow group (fewer than five) leaves the statistics hostage to any single company's temporary distortion. A stale peer group recycled from a prior engagement may still contain companies that have since been acquired, changed business model, or hit company-specific trouble. And strategic bias is the most dangerous: research has documented that banks systematically select higher-multiple peers for sell-side clients, and peer selection is scrutinized in fairness opinion litigation. The defense is a principled, documented process: "we excluded Company X because its revenue is 10x the target's" survives cross-examination; "we excluded it because it lowered our range" does not.

Gathering and Normalizing the Data

For each company in the finalized group, the analyst collects four families of data:

• •Market data: current share price, diluted shares outstanding, equity value, net debt, and enterprise value built through the EV bridge
• •Historical financials: revenue, EBITDA, EBIT, and net income for the last twelve months and most recent fiscal year
• •Projected financials: consensus analyst estimates for the current and next fiscal year, from Bloomberg, FactSet, or Capital IQ
• •Operating metrics: growth rates, gross and EBITDA margins, leverage (net debt to EBITDA), and return measures such as ROIC

Raw reported numbers are rarely comparable as filed. Normalization (scrubbing) adjusts each company's financials to reflect sustainable earning power: adding back non-recurring items (restructuring charges, litigation settlements, impairments, acquisition costs), correcting for accounting differences (depreciation methods, revenue recognition, stock-based compensation treatment), and fixing M&A distortions such as a partial-year contribution from a mid-year acquisition. The goal is clean, recurring metrics, because a multiple is only as meaningful as its denominator.

The same skepticism applies with extra force to the target itself in a deal context: a seller's adjusted EBITDA is almost always higher than its reported EBITDA, because management adds back everything it can characterize as one-time. Scrutinizing those add-backs is the buyer's and the banker's job. One further mechanical issue, differing fiscal year-ends across the peer group, is solved by calendarization, covered in the time-dimension section below.

Choosing the Multiples

The Matching Principle

Every multiple must pair a numerator and denominator that belong to the same claimholders. Enterprise value pairs with unlevered metrics available to all capital providers: revenue, EBITDA, EBIT. Equity value pairs with levered metrics that belong to shareholders after debt service: net income, book equity. Pairing EV with net income, or market cap with EBITDA, is the classic matching error. The comps table shows several multiples, but the analyst designates one primary valuation multiple by industry: EV/EBITDA for most operating businesses, EV/Revenue for high-growth technology, P/E and P/B for financial institutions.

EV/EBITDA: The Workhorse

EV/EBITDA is the default multiple in trading comps, the standard in precedent transactions, and the shared language of M&A and private equity: when a sponsor says "we paid 11x," everyone knows what that means. The pairing works because enterprise value is the value of the business to all capital providers, and EBITDA is the cash flow proxy available to all of them before interest, taxes, and depreciation. Its dominance rests on a triple neutrality:

• •Capital-structure neutral. EBITDA sits above interest expense, so two identical operators with different leverage show the same EBITDA and the same EV/EBITDA. Their P/E ratios would diverge because the levered company pays more interest.
• •Tax-neutral, approximately. EBITDA is pre-tax, so jurisdiction, NOLs, and tax planning do not distort the comparison. It is not perfectly neutral (enterprise value implicitly prices debt tax shields), but the distortion is far smaller than in after-tax metrics.
• •Depreciation-neutral. Adding back D&A removes differences from depreciation policy, asset age, and acquisition-related intangible amortization.

Read the multiple as the number of years of current EBITDA the market pays for the business: higher multiples price faster growth, more durable margins, or lower risk. Absolute levels differ enormously by sector (software trades far above utilities or energy) and move with interest rates, credit availability, and risk appetite, so a multiple is never cheap or expensive in isolation; it is cheap or expensive relative to comparable companies in the same market environment.

EV/EBITDA works best when EBITDA is positive, capex is moderate and predictable, the corporate structure is standard, and the industry has enough public peers. It breaks down in four situations. For financial institutions, debt is an operating asset, not financing: deposits fund lending, interest is an operating cost, and both EV and EBITDA lose meaning, so banks and insurers are valued on P/E and price-to-book. For pre-profit companies, negative EBITDA makes the multiple undefined; the analysis switches to EV/Revenue or a forward year in which EBITDA turns positive. For capital-intensive businesses with uneven capex, EBITDA has a capex blind spot: a miner generating $500 million of EBITDA that must spend $400 million a year to maintain operations has only $100 million of free operating cash flow, and comparing its multiple to a software company with $50 million of capex is misleading; EV/EBIT or EV/(EBITDA minus capex) serves better. Finally, EBITDA adds back D&A but not stock-based compensation; some technology analysts exclude SBC as non-cash while others insist it is real economic dilution, and the choice must at least be consistent across the peer group.

EV/Revenue and the Rule of 40

EV/Revenue is the fallback when EBITDA is negative or not yet meaningful (clinical-stage biotech, early-stage SaaS), when cost structures diverge so much across the group that EBITDA multiples are incomparable, and for high-growth companies whose current earnings understate future earning power. Its weakness is the revenue quality problem: a dollar of 80 percent gross margin recurring subscription revenue is worth far more than a dollar of 30 percent margin hardware revenue, so two companies both at 8x revenue are not similarly valued unless their margin profiles match.

For software specifically, the Rule of 40 compresses the growth-profitability tradeoff into one score: revenue growth rate plus EBITDA (or free cash flow) margin should reach at least 40. A company growing 30 percent at a 15 percent margin scores 45; one growing 10 percent at a 20 percent margin scores 30. Companies above the threshold consistently trade at a large premium on EV/Revenue, historically around two to three times the multiple of those below it, which makes the rule a fast way to compare businesses making different tradeoffs between growth and margin.

P/E, PEG, and Price-to-Book

The price-to-earnings ratio divides equity value by net income. It is a levered multiple: capital structure, tax rates, and D&A all flow through the denominator, which is exactly why EV/EBITDA displaced it for cross-company comparison. P/E remains primary in two places. For financial institutions it is the right measure, because earnings after funding costs are the true bottom line of a bank. And in public equity markets it is the convention: research price targets, index valuation debates, and EPS-driven narratives all run on forward P/E.

The two multiples are related but not convertible without additional information. The bridge runs:

Each arrow demands an assumption: D&A as a share of EBITDA, then interest expense, then the tax rate, then net debt to step from enterprise to equity value. The EV/Net Income rung is deliberately mismatched: it exists only as an intermediate step in the algebra, never as a multiple you would quote. Identical EV/EBITDA multiples can therefore map to very different P/E ratios; a company at 10x EV/EBITDA but 30x P/E is carrying heavy D&A, heavy leverage, or both.

Two related equity-side tools complete the kit. The PEG ratio divides P/E by the expected EPS growth rate: 1.0x reads as fairly valued relative to growth, below 1.0x as cheap, though the output is only as good as the growth estimate chosen. Price-to-book (and price-to-tangible-book) anchors bank, insurance, and REIT valuation, because those businesses are priced off their net asset base. For banks the driver is return on equity relative to cost of equity: a bank earning 15 percent ROE against a 10 percent cost of equity deserves a premium to tangible book, while one earning 6 percent destroys value and trades at a discount. That is why a top-return franchise can trade above 2x tangible book while structurally low-return European banks have historically sat at 0.5-0.8x.

EV/EBIT and Sector-Specific Multiples

EV/EBIT keeps depreciation in the denominator, making D&A a rough proxy for the maintenance capex a business must keep spending. It is preferred when capital intensity varies across the group or the industry is asset-heavy (manufacturing, telecom, transportation), and it captures acquisition-related amortization that EBITDA ignores. Its limitation is that D&A is an accounting estimate, not a cash cost.

Some industries price on metrics built for their specific economics: EV/EBITDAR (retail, airlines, restaurants) adds back rent so owned and leased real estate compare cleanly; EV/Reserves and NAV value oil, gas, and mining companies on their resource base rather than cyclical earnings; NAV plus FFO/AFFO replace earnings for REITs, where real estate depreciation distorts net income; EV/Subscribers and EV/ARPU capture media and telecom customer economics; and Price/AUM values asset managers on the driver of their fees.

Using the wrong multiple is a fast credibility killer: EV/EBITDA for a bank, EV/Revenue for a mature cash-cow industrial, or P/E for a pre-revenue biotech each signals a misunderstanding of what the multiple measures. Knowing which multiple fits the situation, and why, matters more than reciting any formula.

The Time Dimension: LTM, NTM, and Calendarization

Every multiple's denominator carries a time stamp: trailing actuals, forward estimates, or a specific calendar period. Applying periods inconsistently across the peer group manufactures phantom over- or undervaluation, so bankers calculate both trailing and forward multiples and present them side by side.

LTM Multiples and the Stub Period

LTM (last twelve months) multiples, also called trailing or TTM, use the most recent twelve months of reported results. Because fiscal years and filing dates rarely line up with today, LTM figures are assembled with stub period arithmetic: most recent full fiscal year, plus the current-year interim (stub) period, minus the prior year's corresponding stub. As of Q3 2025 for a December fiscal-year company: LTM EBITDA = FY2024 EBITDA + Q1-Q3 2025 EBITDA - Q1-Q3 2024 EBITDA. That construction always captures exactly twelve months.

LTM's strength is that the numbers are facts: reported, audited, and verifiable by every party to a negotiation, which makes them the defensible basis in formal settings like fairness opinions. Its weaknesses mirror that. It is backward-looking, so fast growers look artificially expensive: a SaaS company that grew revenue from $100 million to $150 million and is heading to $210 million has a much higher LTM revenue multiple than NTM, purely because the denominator lags. And LTM inherits every one-time item in the trailing window; a large Q2 restructuring charge depresses LTM EBITDA and inflates the multiple unless the figure is normalized.

NTM Multiples and Consensus Estimates

NTM (next twelve months) multiples use forward estimates, calibrated to the twelve months from the analysis date. The denominator is the consensus estimate: the median of sell-side analyst forecasts, aggregated by Bloomberg, FactSet, Capital IQ, or Visible Alpha. NTM aligns with how markets actually price assets (expectations, not history), is largely free of trailing one-time items because analysts project normalized earnings, and is essential for high-growth companies whose forward earnings are the real basis of their price.

The cost is that estimates are forecasts, not facts. If consensus expects $500 million of EBITDA and the company delivers $400 million, every NTM multiple built on that estimate was wrong. Consensus can also go stale after an earnings miss or guidance change, in the lag before analysts update their models.

Calendarization

Peer companies with different fiscal year-ends report "annual" results covering different periods, so their figures must be calendarized to a common window before multiples are compared. The mechanic is a weighted blend of two fiscal years, weighted by the months each contributes to the target period:

• •December year-end: calendar 2026 estimate is simply the fiscal year figure, say $400 million of EBITDA
• •June year-end: calendar 2026 = 50 percent of FY ending June 2026 plus 50 percent of FY ending June 2027, so estimates of $350 million and $380 million blend to $365 million
• •March year-end: calendar 2026 = 25 percent of FY ending March 2026 plus 75 percent of FY ending March 2027, so $290 million and $320 million blend to $312.5 million

Data providers produce calendarized estimates automatically, but the analyst should know the mechanics to verify the output and to patch gaps when provider data is incomplete or stale. Fiscal-year variety is common: retailers often end in January or February, and technology year-ends are scattered.

Using LTM and NTM Together

The two bases are complements, not competitors:

In the comps table both appear, with NTM usually carrying more weight in the conclusion and LTM serving as the reality check. In precedent transactions the convention flips: transaction multiples are quoted on LTM EBITDA at announcement, because that was the information the price was set on. In live negotiations, each side quotes the earnings basis that frames the price most favorably to its argument, which is exactly why an analyst must be fluent in both.

The LTM-NTM spread is itself information. A company at 15x LTM but 10x NTM EBITDA is expected to grow earnings roughly 50 percent over the next year; near-identical multiples signal a mature business; and an NTM multiple above LTM (rare) implies expected earnings decline. Scanning the spread down the peer group instantly separates the growers from the mature names.

Spreading the Comps Table

Spreading comps means calculating each peer's multiples and organizing them into a structured table ready for a pitchbook, board deck, or fairness opinion. It is one of the most common first-year analyst tasks, and the skill is less in the arithmetic than in the discipline: consistent structure, current data, clean formatting. The standard layout flows left to right in five sections:

1. 1.Company identifiers: name and ticker, with the target shown separately (shaded or set apart) from the peer group
2. 2.Market data: share price, diluted shares outstanding, equity value, net debt, and enterprise value, sometimes with the 52-week range
3. 3.Financial metrics: LTM and NTM revenue, EBITDA, EBIT, and net income, plus growth rates, margins, and ROIC
4. 4.Valuation multiples: EV/EBITDA, EV/Revenue, EV/EBIT, and P/E for each period, with headers naming both metric and period ("NTM EV/EBITDA," never just "EV/EBITDA")
5. 5.Summary statistics: mean, median, high, low, and often the 25th and 75th percentiles

The operating metrics are not decoration; they are what turns a data display into an analytical tool. A peer at 15x NTM EBITDA against an 11x median stops looking anomalous when the table shows it growing revenue at 20 percent while the group grows at 5.

A comps table is a living model, not a one-time deliverable. Share prices refresh daily, moving every EV-based multiple; quarterly earnings require updated LTM figures and a check on consensus shifts; and events force decisions, such as removing a peer that has been acquired or footnoting a major one-time item. Presenting stale comps, a table missing a peer's earnings from two weeks ago or still listing a company that was taken private, is one of the fastest ways for a junior analyst to lose credibility.

Formatting follows bank-standard conventions: blue font for hard-coded inputs, black for formulas, green for links to other worksheets; multiples to one decimal place; footnotes for any adjustment. The test is scannability: an MD should find the median NTM EV/EBITDA in seconds. For the pitchbook itself, the full spread is condensed into a summary version, usually the primary multiple plus key operating stats and the benchmark statistics, presented alongside the football field.

Interpreting the Output

A finished comps table does not interpret itself. Two analysts holding the same spread can reach different valuation conclusions, because interpretation involves choices: which statistic anchors the analysis, what to do with outliers, and where the target belongs relative to the group.

Median vs. Mean

The median is the default anchor. Its virtue is outlier resistance: in a group with multiples of 8x, 9x, 10x, 10.5x, 11x, 11.5x, 12x, and 25x, the median of 10.75x reflects the group's central tendency while the mean of 12.1x is dragged upward by one extreme value. The median also describes what a typical comparable trades at and stays stable when a single peer is added or dropped. The mean earns preference only in small groups (fewer than five companies) with no clear outliers, where a median can be dominated by one name; and even then a single anomaly wrecks it, which is one more reason banks insist on five-plus peers. Whichever anchors the work, no single statistic is ever the output: comps produce a range, and a point estimate pretends to a precision the method cannot deliver.

Percentiles and the Interquartile Range

Percentile statistics convert the peer group's dispersion into a usable range. The 25th percentile marks the low end, the median the center, and the 75th percentile the high end. The span between the quartiles, the interquartile range (IQR), captures the middle 50 percent of the group while trimming both extremes, and it is the range most commonly plotted as the comps bar on the football field. A narrow IQR says the market prices these peers similarly and raises confidence in the benchmark; a wide IQR flags a heterogeneous group or heavy company-specific differentiation.

A worked pass: ten sorted NTM EV/EBITDA multiples of 7.8x, 8.5x, 9.2x, 9.8x, 10.3x, 10.8x, 11.2x, 11.9x, 13.1x, and 16.5x give roughly 9.0x at the 25th percentile, a 10.55x median, roughly 12.2x at the 75th, and a 10.9x mean pulled up by the 16.5x outlier. Against target NTM EBITDA of $200 million, the interquartile range implies enterprise value of $1.8 billion at the low end, $2.1 billion at the median, and $2.4 billion at the high end.

Handling Outliers

An outlier is a prompt to investigate, never a reflex to delete. A high outlier may be a genuine growth disruptor, the subject of acquisition speculation (its price embeds a rumored deal premium, not standalone value), or a company whose temporarily depressed EBITDA inflates the multiple. A low outlier may face company-specific trouble (litigation, product failure, management crisis), sit in a different phase of its cycle, or be enjoying a one-time EBITDA benefit. The disposition follows the diagnosis:

• •Genuinely comparable with a temporary distortion: keep it, with a footnote explaining the issue
• •Different business model or growth phase: move it to the secondary comparables
• •EBITDA near zero or negative: mark the multiple NM (not meaningful) and exclude it from the summary statistics
• •Price inflated by acquisition speculation: exclude or footnote
• •Already acquired: remove it; it is no longer a trading comparable

The one forbidden move is excluding a genuinely comparable company because its multiple widens the range or drags the median the wrong way. Every exclusion must stand on the merits, because peer exclusions are exactly what courts and counterparties dissect when a fairness opinion is challenged.

Explaining the Dispersion

Summary statistics say what the multiples are; the analyst's job is to explain why they differ. A peer at 14x against a 10.5x median is usually earning that premium with superior growth (20 percent versus 8), higher margins (35 percent versus 22), or lower risk; a peer at 7.5x is usually paying for cyclical headwinds, margin pressure, or concentration risk. That mapping is what positions the target within the range: above-median growth and margins argue for a multiple above the median, and the reverse argues for a discount.

In practice this is often done visually: plot each peer with revenue growth on one axis and EV/EBITDA on the other, confirm the expected upward relationship, and read the target's implied multiple off the trendline given its own growth rate. Even informally, the chart reveals which peers the target genuinely resembles. A quick numerical version divides the multiple by the growth rate, a growth-adjusted comparison in the spirit of the PEG ratio: a company at 15x growing 25 percent (0.6x per point of growth) is cheaper on a growth-adjusted basis than one at 10x growing 10 percent (1.0x). The strongest comps work never stops at a range; it explains why the target belongs at a specific point inside it.

From Multiples to Implied Valuation

The final step converts benchmarks into a valuation. Step one applies the benchmark multiple to the target's corresponding metric:

Suppose the target has NTM EBITDA of $300 million and the peer group's NTM EV/EBITDA quartiles are 9.8x, 11.5x, and 13.2x. The implied enterprise value runs from $2.94 billion (9.8x) through $3.45 billion at the median to $3.96 billion (13.2x). One discipline governs the whole step: match the periods. An NTM multiple applies to the target's NTM metric, an LTM multiple to LTM; crossing them mixes forward pricing with trailing performance and corrupts the answer.

Step two converts enterprise value to equity value by running the EV bridge in reverse:

If the target carries $500 million of debt, $50 million of preferred equity, no minority interests, and $200 million of cash, the midpoint case resolves to:

Step three divides by diluted shares outstanding to reach per-share value:

With 150 million diluted shares:

Running the low and high cases through the same bridge gives $17.27 and $24.07, so the comps-implied range is $17.27-24.07 per share. The word "implied" is doing real work: the number is conditional on the peer group chosen, the metric used, and the period applied, which is why implied valuations from different methodologies are compared rather than averaged.

Four errors repeatedly corrupt this step:

• •Applying peer multiples (built on normalized figures) to the target's unadjusted reported EBITDA
• •Mixing periods, such as an NTM multiple on an LTM metric
• •Using an outdated share count after issuance or buybacks
• •Presenting implied enterprise value as if it were equity value, skipping the bridge entirely

The range lands as one bar on the football field chart, alongside precedent transactions, DCF, LBO, and the 52-week trading range. The comps bar typically sits below precedent transactions, because deal multiples embed control premiums, and the zones where methodologies overlap mark the highest-confidence values. Where the target belongs inside the comps bar is an argument, not a calculation: sell-side bankers press the case for a premium to the median on growth, margins, and positioning, while buy-side bankers emphasize risks and integration challenges to argue a discount.

Strengths, Weaknesses, and When to Trust Comps

Comps earn their default status honestly. They are market-based: built on observable prices and public filings that both sides of a negotiation can verify. They are fast to build and update, which makes them the first analysis on any time-pressured engagement. They are intuitive enough for any board member to follow. And they embed the market's real-time consensus, the aggregated judgment of thousands of investors, which no single analyst's model replicates.

The central caveat is that comps capture what the market currently thinks, not what a company is intrinsically worth, and markets are periodically, systematically wrong. During the 2021 growth rally many SaaS companies traded at 30-50x revenue, and comps built then produced valuations that collapsed with the 2022-2023 correction; the dot-com bubble and the 2008 crisis cut the same lesson in opposite directions. The structural weaknesses follow: comps inherit any market-wide distortion, they are only as good as the peer group, their time bases are either backward-looking (LTM) or forecast-dependent (NTM), and they price minority stakes, not control.

Trading vs. Transaction Multiples

That last point deserves precision. A trading multiple comes from a company's current public market valuation and reflects what a passive investor pays for a small stake. A transaction multiple comes from an actual M&A deal price, which includes a control premium, typically 20-40 percent above the undisturbed trading price. For the same company the transaction multiple almost always exceeds the trading multiple, which is why the comps bar sits below the precedent transactions bar on the football field, and why a comps-based value understates what an acquirer should expect to pay. The gap between the two bars approximates the market's going rate for control.

When Comps Deserve Weight

Trust comps most when the peer group is well matched and the market is stable, when the target sits in a well-covered sector with many public peers, and when the output is triangulated against other methods. Trust them least during bubbles and panics (sentiment, not fundamentals, is setting the multiples), for unique business models where the peer group is forced, and for targets undergoing transformational change that historical comparables cannot capture. By situation: comps carry the most weight in IPO pricing, where the company will trade against public peers from day one, and the least in restructuring, where the target's price reflects distress while its peers are healthy.

Triangulating with DCF and Precedent Transactions

No banking deliverable presents comps alone, and not merely by convention: each pillar covers the others' blind spots. The DCF supplies a fundamental anchor independent of market pricing; when comps and DCF converge, confidence is high, and when they diverge the gap itself is diagnostic (a DCF above comps suggests the market undervalues the target or the projections are optimistic; a DCF below suggests sector-wide overpricing or conservative assumptions). Precedent transactions supply what buyers have actually paid, control premium included, which sets sell-side price expectations and benchmarks buy-side offers. Comps tell you where the market prices similar companies today. The other two methods tell you whether to believe it, and what someone might actually pay.