Multiples & comparables

Definition

Multiples value a business by ratio to a fundamental — P/E, EV/EBITDA, EV/Sales, P/B, P/FCF — benchmarked against comparable companies or the firm’s own history. Every multiple is a compressed DCF: it embeds growth, margin, risk, and reinvestment assumptions without stating them — which makes multiples fast, comparable, and systematically abused. The platform’s rule: a multiple comparison is only as good as the comparability of what’s embedded.

How it works / structure

• The ladder: EV/Sales (works when earnings don’t exist; assumes margins converge), EV/EBITDA (capital-structure- neutral operations), P/E (equity earnings, leverage embedded), P/FCF (the cash discipline), P/B (balance-sheet anchored — financials’ native multiple — fa-sector-metrics).
• What ranks multiples: Liu-Nissim-Thomas (2002) measured valuation accuracy across multiples — FORWARD earnings multiples performed best, sales multiples worst; the documented ordering behind the platform’s preference for forward-earnings bases where estimates exist (fa-guidance-estimates).
• Comparability adjustments (engine-executable): same fiscal basis, lease/pension treatment aligned, one-time items removed symmetrically (fa-earnings-quality), growth and margin percentile context attached — a “cheap” multiple with worse growth/margins is usually correctly priced, not cheap.
• Own-history form: multiple vs its own 5/10-year distribution — regime-aware (the rate regime moves ALL duration-heavy multiples — regime-rate-environments).

When it applies

Cross-sectional value screens (strategy-factor-investing value definitions ARE multiples); relative theses within sectors (the honest habitat — same embedded assumptions); sanity-checking DCF outputs; event repricings (post-derating entries measured against the multiple’s history).

Risk profile & failure modes

• Cheap-for-a-reason selection: the low-multiple tail concentrates value traps (declining businesses correctly discounted); multiple screens without quality/trend filters select for them (fa-earnings-quality).
• Denominator games: adjusted-EBITDA multiples using the company’s own exclusions; peak-earnings P/Es at cycle tops (cyclicals look cheapest exactly when earnings are about to halve — the classic error, inverted at troughs).
• Comps-set shopping: choosing flattering comparables is the sell-side’s oldest trick; the platform pins comps sets by mechanical criteria (sector, size band) before looking.
• Regime blindness: comparing today’s multiple to a different rate regime’s average without adjustment.

Evidence & limits

Liu-Nissim-Thomas (2002) is the peer-reviewed accuracy ordering; the value-factor literature carries the systematic return evidence with its drought history (strategy-factor-investing). Multiples’ descriptive function is arithmetic; every “cheap therefore buy” inference imports unstated assumptions the entry’s discipline exists to surface.

Falsifiable-thesis examples

Illustrations only, not signals:

• “X, at the 10th percentile of its own 10-year EV/EBITDA with stable margins, will re-rate above its median within two years” — falsified by the multiple path.
• “The cheapest quartile of sector S by forward P/E (quality- filtered) will outperform the richest over the next year” — falsified by the cohort returns.

Cross-references

• The uncompressed form: fa-dcf-valuation
• Inputs and hygiene: fa-ratio-analysis, fa-earnings-quality, fa-guidance-estimates
• Sector-native multiples: fa-sector-metrics
• Systematic use: strategy-factor-investing (value)