Stock Valuation Methods Compared
DCF, multiples, and earnings power — when each method dominates and when it deceives
Valuation methodology
Comparable Company Analysis
How to select peers, pick the right multiples, and derive a defensible valuation range — without cherry-picking the comps that make your number look right.
EV/EBITDAP/EEV/RevenueP/FCFLTM / NTMPercentile ranges
Comparable company analysis — "comps" — estimates a company's value by observing what the market currently pays for similar businesses. You take a set of peers, calculate their valuation multiples (EV/EBITDA, P/E, EV/Revenue, P/FCF), then apply that multiple range to your subject company's financial metrics to derive an implied price range. Comps are the standard sanity check alongside a DCF model: the DCF tells you what the business is worth on its own merits; comps tell you where the market is pricing the category. When they diverge significantly, you need a reason. See stock valuation methods compared for when each approach dominates.
Step 1
How to select peer companies
The peer group is the most consequential decision in a comps analysis. An analyst who cherry-picks peers gets any answer they want, so the selection criteria must be defined before you start screening — not after you see which companies produce flattering multiples.
Four concrete filters in order of strictness:
| Criterion | Target range | Why it matters |
|---|---|---|
| GICS subsector | Same 6-digit subsector; expand to industry if peers are scarce | Different end markets price differently even within the same sector |
| Revenue scale | ±50% of subject company LTM revenue | Micro-cap multiples and mega-cap multiples carry fundamentally different liquidity and scale premiums |
| Growth profile | Revenue CAGR within 5–8 pp over last 3 years | A 30% grower will always command a premium to an 8% grower in the same subsector — mixing them distorts the median |
| Capital structure | Net Debt/EBITDA within 2x of subject; exclude negative EBITDA peers if using EV/EBITDA | Leverage changes EV-based multiples materially; a highly levered peer is priced differently for bankruptcy risk, not just ops |
Hard exclusions, no exceptions: companies in active M&A (target or acquirer) and companies in financial distress (interest coverage below 1.5x or covenant violations disclosed). M&A targets trade at a control premium that has nothing to do with standalone value. Distressed companies trade on recovery value, not earnings power.
Aim for 5–10 peers. Fewer than 5 gives you a percentile range with no statistical meaning. More than 12 usually means you have relaxed the criteria until the peer group is so heterogeneous it tells you nothing. If your subject is genuinely rare — a niche industrial or a specialist financial — say so explicitly and widen only one criterion at a time.
Step 2
Which multiples to use and when
The multiple you choose should be determined by the business type, not by which one produces the most convenient answer. Each multiple has a specific structural reason it works well for certain companies and lies for others.
EV/EBITDA — capital-intensive mature businesses
EV/EBITDA is the workhorse multiple for M&A and for comparing businesses across different capital structures. Because it uses enterprise value (market cap plus net debt) in the numerator and strips interest, taxes, and D&A from the denominator, it neutralizes three sources of noise: financing choices, tax jurisdiction, and accounting depreciation schedules. An industrial conglomerate with a 30% debt-to-cap ratio compares cleanly against a peer that is debt-free. A company running accelerated depreciation on its equipment compares cleanly against one using straight-line. Use EV/EBITDA for manufacturers, cable companies, utilities, and any business where D&A differences across peers are large enough to distort a P/E comparison.
Calculate EV/EBITDA for any stock →P/E — mature earners with stable earnings
Price-to-earnings is the most widely quoted multiple and the most sensitive to distortions. It works well when peers have similar leverage (so interest expense is roughly proportional), similar tax rates, and earnings that are genuinely representative — not inflated by one-time gains or distorted by large non-cash items. P/E breaks down when you are comparing a levered company against an unlevered peer: the levered company will have lower net income from interest expense, making it look more expensive on P/E even if EV/EBITDA is identical. Normalize earnings before comparing — strip asset sale gains, restructuring charges, and any “adjusted” add-backs that conveniently show up every quarter. Reserve P/E for large-cap consumer staples, financials with low leverage variance, and established software businesses with predictable earnings streams.
Calculate P/E fair value →EV/Revenue — high-growth or pre-profit companies
When a company is unprofitable or in the early phases of scaling, EBITDA and earnings are either negative or too small to anchor a meaningful multiple. EV/Revenue sidesteps this by using the top line, where values are positive and meaningful even for loss-making businesses. The tradeoff is that revenue multiples embed an implicit assumption about future margins — a 10x EV/Revenue company needs to eventually generate significant profit margin to justify that price. EV/Revenue is appropriate for pre-profitability SaaS, early-stage biotech (by sector analogy), and any hypergrowth company where the peer set is priced on ARR or GMV. When using it, flag the implied margin expectation explicitly in your output.
P/FCF — capex-light models
Price-to-free-cash-flow is the cleanest multiple for businesses where CapEx is minimal and earnings convert directly to cash — SaaS companies, marketplace platforms, asset-light franchisors. FCF is harder to manipulate than GAAP earnings and strips out the D&A add-back EBITDA uses, keeping maintenance CapEx visible. For software or platform businesses where peers have vastly different D&A loads from acquisitions, P/FCF often gives a cleaner cross-company comparison than P/E. Use it when the peer set has high stock-based comp variability — subtract SBC from FCF before comparing, or you end up ranking companies by how generously they compensate employees, not how much cash they generate.
How to use this — multiple selection decision tree
- Is the subject company profitable on an EBITDA basis? If no → use EV/Revenue.
- Does the peer group have materially different leverage levels or D&A schedules? If yes → use EV/EBITDA, not P/E.
- Is CapEx under 5% of revenue for most peers? If yes → consider adding P/FCF alongside EV/EBITDA.
- Are earnings stable and leverage uniform across peers? If yes → P/E is acceptable as the primary multiple.
- Use at least two multiples and note when they disagree — the disagreement is information.
Step 3
Building the comps table
A comps table is only as useful as the consistency of its inputs. The two most common ways analysts corrupt their own comps: mixing LTM and NTM figures within the same column, and carrying through one-time items that distort the denominator.
LTM vs NTM: Last-twelve-months (LTM) multiples use reported financials and are directly verifiable from filings. Next-twelve-months (NTM) multiples use consensus estimates and forward-price the stock against expected performance. Both are valid — but you must use one consistently across every peer in the table. A mixed table where you use LTM EBITDA for companies that have already reported Q1 and NTM EBITDA for those that haven't is comparing different time horizons. For mature businesses in stable sectors, LTM is fine. For high-growth companies where the trailing year understates the current trajectory, NTM is often more representative. Pick one and apply it uniformly.
Stripping one-time items: For every peer, adjust the denominator to remove items that will not recur and that management would exclude from their own “adjusted” presentation — but verify each add-back is genuinely non-recurring. Common adjustments: restructuring charges, gain/loss on asset sales, impairments, acquisition-related amortization (for P/E comparisons), and COVID-era anomalies for businesses that had unusual years. Do not add back stock-based comp unless you are computing an adjusted EBITDA specifically to match how the sector is conventionally priced (e.g., some software comps universally exclude SBC).
How to use this — comps table construction checklist
- Pull LTM financials from the most recent 10-K/10-Q for all peers. Calculate NTM from consensus if needed — document the source and date.
- Adjust each peer's EBITDA/EPS/FCF for identified one-time items. Note every adjustment in a footnote — your table should survive a colleague questioning any single line.
- Calculate EV for each peer: market cap + total debt − cash and equivalents. Use fully diluted shares (include RSUs, options in-the-money at current price).
- Compute multiples for each peer. Flag outliers (multiples more than 2 standard deviations from the group) and decide whether to keep them, exclude them with a note, or trim them.
- Calculate 25th percentile, median, and 75th percentile across the peer group. Do not report just the mean — a single outlier can shift it materially.
The 25th/median/75th percentile structure is important: it brackets the peer distribution without being distorted by outliers, and it forces the analysis to acknowledge that your subject company's appropriate multiple depends on where it ranks within the peer group — not just on the average.
Step 4
Deriving a valuation range
Apply the 25th percentile multiple to your subject company's metrics for the low end of the range, and the 75th percentile for the high end. The median is your central estimate. This produces an implied enterprise value range; convert to equity value by subtracting net debt, then divide by fully diluted shares to get an implied price range.
The output should always be a two-column table: low and high. A single-point estimate from comps is false precision — the spread in peer multiples already tells you the market assigns a wide range of values to businesses in this subsector. Honor that uncertainty.
| Step | Low (25th pctile) | High (75th pctile) |
|---|---|---|
| Peer EV/EBITDA multiple | 8.5x | 12.2x |
| Subject LTM EBITDA | $480M | $480M |
| Implied enterprise value | $4,080M | $5,856M |
| Less: net debt | ($620M) | ($620M) |
| Implied equity value | $3,460M | $5,236M |
| Fully diluted shares | 210M | 210M |
| Implied share price | $16.48 | $24.93 |
When the comps-derived range and your DCF range overlap, the thesis is more robust. When they diverge by more than 30%, you are either modeling a different set of assumptions than the market is or your peer group has a structural issue. Triangulate rather than average — understand why the methods disagree.
Failure modes
Common mistakes that invalidate a comps analysis
Comps are the valuation method most susceptible to motivated reasoning, because the answer is directly determined by which peers you pick and which time period you use. These are the four mistakes that most often produce a comps table that looks rigorous but is quietly working backwards from a predetermined conclusion.
Cherry-picking flattering comps
Including only the peers with the highest multiples produces an inflated valuation range; including only the lowest produces a discount. The tell: if your peer group consists of companies that all happen to trade at the high end of the sector range, or if you quietly dropped two peers after seeing their multiples, the analysis has been reverse-engineered. Define selection criteria before pulling data, document every excluded peer and why, and require someone else to verify the list.
Ignoring leverage differences
A peer trading at 6x EV/EBITDA with 4x Net Debt/EBITDA and a peer trading at 9x EV/EBITDA with 0.5x leverage are not directly comparable on equity value. The levered peer is priced in part for its financial risk. If you apply the 9x EV/EBITDA multiple from the unlevered peer to your highly levered subject company, you are attributing financial-risk premium to operating quality. Always check capital structure before including a peer. When leverage varies significantly across the group, use enterprise value multiples rather than equity multiples, and note the dispersion in the table.
Mixing LTM and NTM figures
Five peers on LTM multiples and three peers on NTM estimates in the same column means your percentile range is comparing different time periods. A company that has already reported a strong Q1 will have a lower LTM multiple than its NTM multiple if the market expects continued growth — the periods are not equivalent. Choose one time period standard and apply it universally, even if it means using consensus estimates for companies that have not yet reported.
Using stale data
A comps table built on multiples from three months ago is a comps table that ignores any earnings prints, guidance revisions, or macro moves that occurred in the interval. For equity fair value work, rebuild the multiples fresh from current market prices and the most recent reported financials. For a buy-side model being updated at earnings, rebuild the table after every quarterly report cycle. Data that is more than six weeks old is unreliable for pricing decisions in a normally active market.
Go deeper
Related valuation guides
How to Value a Stock
Implied expectations framework — decode what the price already believes before building a model
FCF vs EBITDA
Which metric is right for each situation and why divergence is a red flag
Discounted Cash Flow Guide
DCF mechanics — projecting cash flows, terminal values, and sensitivity analysis
Apply it
Run a full fundamental analysis in minutes.
Basis Report generates valuation, earnings quality, and risk factors for any ticker — comps context included. Start with the data before you build the model.