P/E, P/B, EV/EBITDA

Three multiples that describe the whole market

A market price without context is meaningless. $180/share says nothing; $180/share at 30× earnings vs $180/share at 8× earnings are entirely different businesses at entirely different valuations. Multiples are how the market compresses a company's financials into a comparable number.

The three you'll see most often:

• P/E — Price / Earnings per share — "how many years of current earnings does today's price represent?"
• P/B — Price / Book value per share — "how much more than the company's accounting net worth am I paying?"
• EV/EBITDA — Enterprise Value / EBITDA — "how many years of operating cash flow does the full business cost?"

Each answers a different question, hides different sins, and suits different industries. Using the wrong one on the wrong company is the single most common valuation mistake.

Play with it

Toggle through the presets and notice: the growth-tech company returns N/A on P/E because earnings are negative. This is the everyday reason analysts reach for EV/EBITDA or price-to-sales: a growing but unprofitable company has no meaningful P/E.

P/E — the default multiple

Two common flavors:

• Trailing P/E (TTM): based on the last 12 months of realized EPS. Most conservative.
• Forward P/E: based on analyst-estimated next-12-months EPS. Tends to be lower (because estimates assume growth) — and anchors on potentially-optimistic analyst forecasts.

What it does well: single-number comparison of earnings yields. A P/E of 15 means a 1/15 ≈ 6.7% earnings yield.

What it hides:

• Non-cash earnings distortions: depreciation methods, intangible amortization, stock-based comp treatment. Two companies with identical cash generation can report very different earnings.
• Capital structure: a company that grows earnings by piling on debt looks identical to one that grows via operations when you only look at P/E.
• One-off items: write-downs, gains on asset sales, settlement payments — these get included in reported earnings but don't repeat. Many pros adjust to "operating EPS" that strips these out.
• Negative earnings: a startup with −$2 EPS has no meaningful P/E. Reporting P/E as negative is not useful; most screens just show "N/M" (not meaningful).

Rule of thumb: mature industrials and financials trade 10–18×; growth tech 25–60×; regulated utilities 12–20×; deep cyclicals (steel, airlines) can trade at 4× at the top of the cycle and 60× at the trough — which is exactly backwards from intuition. P/E for cyclicals is dangerous.

P/B — for asset-heavy businesses

Book value = Shareholders' Equity = Total Assets − Total Liabilities.

Where it shines: financial companies. A bank's balance sheet is its business — loans, deposits, trading assets. Book value approximates liquidation value more closely than for operating businesses. Historical bank P/B ranges: distressed banks trade below 1.0× (market says book is overstated); healthy banks 1.2–2.0×; premium franchises (JPM peaks, Bank of America in good times) can reach 2.5×.

Where it misleads:

• Asset-light businesses: Apple's book value per share is a fraction of its market price not because Apple is overvalued, but because most of Apple's value (brand, ecosystem, engineering talent, operating leverage) doesn't sit on the balance sheet. Apple-like P/B of 45× is meaningless.
• Intangibles-heavy businesses: software, pharma, and media companies have the same issue. The balance sheet understates the productive assets.
• Goodwill from acquisitions inflates book value without adding productive capacity. Companies that have grown by acquisition often show P/B that looks reasonable but is misleading because the goodwill will be written down at some point.
• Share buybacks reduce book value numerically (equity falls as cash leaves) while the business is unchanged. P/B can look elevated purely from a buyback program.

Rule of thumb: P/B is useful for banks, insurance, REITs, industrial conglomerates; mostly useless for tech, software, pharma, brand-driven consumer staples.

EV/EBITDA — the capital-structure-neutral multiple

Read this as: "how many years of operating cash generation would it take to buy the entire enterprise — equity and debt alike, net of cash on hand?"

Why this is the analyst's favorite:

• Capital-structure-neutral. Two otherwise-identical companies, one with $50B of debt and one with zero, have very different P/Es but very similar EV/EBITDA. EV/EBITDA tells you about operating economics, stripping out financing choices.
• Depreciation-neutral. EBITDA = Earnings Before Interest, Taxes, Depreciation, Amortization. This strips out non-cash accounting choices that plague P/E. Two mining companies with identical cash generation but different depreciation schedules will show identical EBITDA and different EPS.
• Cross-border comparable. Different countries have different tax regimes, but EBITDA is pre-tax. You can compare a French industrial to a Japanese one without first having to match accounting norms.

Where it falls down:

• Heavy-capex businesses: EBITDA ignores capex, but capex is real money spent to maintain the business. Utilities, telecom, heavy manufacturing — EBITDA systematically overstates free cash flow. Use EV/EBIT or EV/FCF instead for these.
• Bank / insurance: EV/EBITDA is meaningless for financials. Their "operating" cash flow looks more like net interest income, and their debt isn't financing debt — it's the product. Use P/B and P/E for these.
• Unprofitable companies: if EBITDA is negative (some early-stage SaaS, biotech), EV/EBITDA goes negative or infinite. Analysts switch to EV/Revenue or EV/ARR.

Rule of thumb: EV/EBITDA is the default for M&A analysis, cross-company comparisons in the same industry, and any company with meaningful debt. Historical ranges: mature industrials 7–10×; consumer staples 12–16×; growth tech 20–40×.

When each multiple is the right tool

The growth-tech preset — why analysts reach for alternatives

The calculator's growth-tech preset shows EPS = −$1.50. The P/E is literally undefined — the formula breaks. This is common: Uber didn't have positive EPS until 2023; Tesla's first full profitable year was 2020. Yet these companies had tens of billions in enterprise value during the negative-earnings years.

For these, analysts use:

• EV/Revenue — simple, unburdened by profitability
• EV/ARR (annual recurring revenue) — for subscription businesses
• EV/Gross profit — when revenue has wildly different unit economics across peers
• Rule-of-40 for SaaS — revenue growth % + operating margin % ≥ 40%

Kaufman and Murphy don't cover these (they're technical-analysis-focused). For the grounding: Graham's Intelligent Investor defines P/B and P/E extensively; Schilit's Financial Shenanigans covers how each metric gets gamed (revenue recognition tricks inflate P/E upward; goodwill and intangibles inflate P/B downward).

Quick check

What you now know

• P/E — simplest multiple; distorted by non-cash charges, one-offs, and capital structure; useless for negative earnings
• P/B — best for asset-heavy businesses (banks, insurance, REITs); meaningless for asset-light (tech, brands, software)
• EV/EBITDA — capital-structure-neutral, depreciation-neutral; best default for most operating businesses; wrong for banks and capex-heavy
• Different industries demand different multiples; using P/E on a cyclical or P/B on Apple will mislead you
• Growth companies with negative earnings break P/E entirely — analysts substitute EV/Revenue or Rule-of-40
• Low multiples aren't automatic buys; they're the starting point for "why is the market pricing this cheap?"

Next: Key Metrics & Shenanigans — Howard Schilit's catalog of how managements distort each of the metrics we just covered.