Trading Comps Analysis: EV/EBITDA, Peer Selection, and Outlier Handling

Trading comps (comparable company analysis) is the most commonly used valuation methodology in investment banking — and one of the most commonly botched in interviews. It's not hard to divide enterprise value by EBITDA. The skill is in selecting the right peers, normalizing the multiples, handling outliers, and presenting a range that a client or VP will actually trust. This guide covers the full workflow, from peer selection to the football field.

What trading comps measure

Trading comps value a company based on what the public market currently pays for similar businesses. The core formula:

Implied EV = Target EBITDA × Peer Median EV/EBITDA
Implied Equity Value = Implied EV − Net Debt
Implied Share Price = Implied Equity Value ÷ Diluted Shares

The logic: if comparable companies trade at 12x EV/EBITDA, and our target generates $100M of EBITDA, the market would value it at ~$1,200M enterprise value — subject to adjustments for size, growth, and quality differences.

Trading comps reflect a minority, non-control stake valuation — the price for buying a small slice of stock on the open market, with no premium for control or synergies. That's why precedent transactions (which include control premiums) typically show higher multiples.

Step 1: Select comparable companies

Good comps share these characteristics with the target:

| Criterion | Why It Matters | |-----------|---------------| | Same industry/sector | Revenue drivers, margin structure, and risk profile align | | Similar business model | SaaS vs. hardware vs. services have different multiple ranges | | Similar size (revenue/EBITDA) | Smaller companies often trade at a discount (liquidity, risk) | | Similar growth profile | High-growth companies command premium multiples | | Similar geography/end markets | Regulatory, currency, and demand exposure align | | Similar capital structure (for P/E) | Leverage affects net income and P/E comparability |

In practice, perfect comps don't exist. The skill is picking the closest available set and explicitly flagging where they diverge:

"These four comps skew larger ($2B+ revenue vs. our target at $400M), so we'd apply a 10%–15% size discount to the median multiple."

Where to find comps

• Capital IQ / FactSet / Bloomberg: screen by industry, size, geography
• Target's 10-K: "Competition" section often names direct peers
• Recent sell-side research: analysts covering the target usually list comps
• Precedent transactions: targets of recent deals in the same space

A typical comp set is 5–10 companies. Fewer than 5 lacks statistical credibility; more than 10 dilutes relevance.

Step 2: Calculate enterprise value for each comp

For each comparable company:

Market Cap = Share Price × Diluted Shares Outstanding
Total Debt = Short-Term Debt + Long-Term Debt (from latest balance sheet)
Preferred Stock = Preferred equity (if any)
Minority Interest = Non-controlling interests (if any)
Cash = Cash & Equivalents (and sometimes short-term investments)
Enterprise Value = Market Cap + Total Debt + Preferred + Minority − Cash

Use market values, not book values. Share price is live; debt is usually close to par for investment-grade issuers.

LTM EBITDA

LTM EBITDA = Sum of last 4 quarters' EBITDA

Use LTM (last twelve months) for consistency — not forward EBITDA (which embeds analyst estimates that vary) and not annual EBITDA from a stale fiscal year. Some practitioners show both LTM and NTM (next twelve months) multiples; LTM is the standard starting point.

The multiple

EV/EBITDA = Enterprise Value ÷ LTM EBITDA

Also commonly calculated: EV/Revenue (for high-growth, pre-profit companies), P/E (for profitable companies with similar capital structures).

Step 3: Build the comps table

A standard trading comps table looks like this:

| Company | Market Cap | Net Debt | EV | LTM EBITDA | EV/EBITDA | EV/Revenue | |---------|-----------|----------|-----|-----------|-----------|------------| | Comp A | $5,200M | $800M | $6,000M | $480M | 12.5x | 3.2x | | Comp B | $3,100M | $1,200M | $4,300M | $310M | 13.9x | 4.1x | | Comp C | $8,400M | $2,100M | $10,500M | $950M | 11.1x | 2.8x | | Comp D | $1,800M | $400M | $2,200M | $165M | 13.3x | 3.5x | | Comp E | $4,600M | $900M | $5,500M | $520M | 10.6x | 2.5x | | Median | | | | | 12.5x | 3.2x | | Mean | | | | | 12.3x | 3.2x |

Use median, not mean, as the primary reference. Median is less distorted by outliers. Mean is shown for completeness but shouldn't drive the implied valuation alone.

Step 4: Handle outliers

Outliers are the most common source of bad comps analysis. Handle them explicitly:

Identifying outliers

• Visual scan: a comp at 25x when the rest cluster at 10x–14x
• Statistical: anything beyond 1.5× the interquartile range
• Fundamental: a comp with a one-time EBITDA spike, recent M&A distortion, or pending bankruptcy

What to do with outliers

1. Investigate first — don't automatically exclude. Understand why the multiple is high or low:

   • Recent acquisition premium still in the stock price
   • One-time cost cuts inflated EBITDA
   • Pending litigation or regulatory action
   • Hyper-growth phase (NTM EBITDA much higher than LTM)

2. Normalize if possible — adjust EBITDA for one-time items, use NTM instead of LTM for high-growth comps

3. Exclude with justification — if the outlier reflects a fundamentally different business or temporary distortion, remove it and note why: "Comp B excluded — pending acquisition at 18x, not representative of standalone trading value"

4. Show the impact — present valuation with and without the outlier so the reader sees the sensitivity

The "only show comps that support our price" trap

This is the most dangerous mistake in comps analysis — and the one interviewers test for. Cherry-picking comps to justify a predetermined valuation destroys credibility. The correct approach: select comps based on business similarity, show the full range (including outliers with explanation), and let the median speak.

Step 5: Apply to the target

Target LTM EBITDA: $100M
Peer Median EV/EBITDA: 12.5x
Implied EV: $100M × 12.5x = $1,250M
Target Net Debt: $200M
Implied Equity Value: $1,250M − $200M = $1,050M
Diluted Shares: 50M
Implied Share Price: $1,050M ÷ 50M = $21.00

Adjustments to consider

• Size discount: target is smaller than comps → apply 10%–20% discount to median
• Growth premium/discount: target growing faster/slower than comps → adjust
• Margin premium/discount: target has better/worse margins → adjust
• Liquidity discount: target is private/non-traded → 15%–30% discount

Always state your adjustments explicitly: "We apply a 15% size discount to the 12.5x median, yielding 10.6x, implying $1,060M EV."

Step 6: Present as a range, not a point

No single multiple is "correct." Present a range:

Implied EV Range:
  25th percentile (10.6x): $1,060M
  Median (12.5x):        $1,250M
  75th percentile (13.9x): $1,390M

This range goes on the football field alongside DCF and precedent transaction ranges. The overlap across methodologies is where the "fair value" conversation happens.

Common interview questions

"Why EV/EBITDA instead of P/E?" EV/EBITDA is capital-structure-neutral — both EV and EBITDA are pre-financing. P/E is affected by leverage (interest expense reduces net income). Use EV/EBITDA to compare companies with different debt levels; use P/E when comparing similar capital structures (e.g., within banks).

"Why subtract cash from EV?" Cash is a non-operating asset the acquirer effectively gets back. EV represents the cost of acquiring the operating business net of cash already on the balance sheet.

"Trading comps vs. precedent transactions — which is higher?" Precedent transactions are typically higher because they include a control premium (acquirer pays extra for control and synergies). Trading comps reflect minority stake pricing. Rule of thumb: precedents > trading comps > DCF (though DCF can go either way).

"What if there are no good comps?" Flag it explicitly. Widen the range, use precedent transactions or DCF as primary, and apply larger adjustment discounts. Showing intellectual honesty about comp quality is better than forcing a clean-looking table from bad peers.

The workflow to practice

1. Select 5–8 comps by business similarity
2. Calculate EV and LTM EBITDA for each
3. Compute EV/EBITDA; identify and explain outliers
4. Use median (not mean) as the primary reference
5. Apply size/growth/quality adjustments to the target
6. Present as a range on the football field

Trading comps look simple on the surface. The discipline is in peer selection, outlier handling, and honest adjustment — exactly the skills that separate a good analyst from one who just divides two numbers.