Yahoo's EV/EBITDA can be computed off EBIT, inflating the multiple by the full D&A load — the tell is EV/EBITDA landing near a plausible EV/EBIT

Claim

EV/EBITDA and EV/EBIT differ by exactly the D&A load. On a capital-light name the two are close and a swapped field is invisible. On a goodwill-heavy roll-up amortising acquired intangibles, EBIT can be little more than half of EBITDA, so the same swap doubles the multiple and makes a normally-priced business look untouchable.

Evidence — LHX, 2026-08-19

L3Harris carries $26.5B of goodwill and acquired intangibles against a $42.9B asset base, and amortises roughly $720M/yr of it. TTM EBIT $2.31B against TTM EBITDA $4.00B — EBIT is 58% of EBITDA.

Source	Field	Value
Yahoo (fin.py snapshot)	EV/EBITDA	32.09
roic.ai FY2025	ev_to_ttm_ebitda	19.47
roic.ai FY2025	ev_to_ttm_ebit	30.76
Computed by hand (TTM, EV $62.0B ÷ EBITDA $4.00B)	EV/EBITDA	15.5x

Yahoo's "EV/EBITDA" of 32.09 tracks the EV/EBIT series, not the EBITDA series. The two roic.ai figures on the same fiscal year are the clean proof: 32.09 sits next to EV/EBIT of 30.76 and nowhere near EV/EBITDA of 19.47.

Consequence if taken at face value: a defense prime at 32x EV/EBITDA is priced like a distressed compounder and screens as an automatic AVOID. At the true 15.5x it is an ordinary sector multiple. The field alone flips the valuation section's conclusion.

Note the second error compounding it on the same name — Yahoo's EV of $68.87B against a hand-computed ~$62.0B, tied to a float (219.82M) exceeding sharesOutstanding (186.21M). Both the numerator and the denominator of this multiple were wrong, in the same direction.

How to apply

1. Never quote a vendor EV/EBITDA on an acquisitive company. Rebuild it: sum four quarters of EBITDA from the statements, divide into a hand-computed EV.
2. The one-line check: compute EBIT/EBITDA from the statements. If the ratio is well below ~0.8, the two multiples are far enough apart that a swapped field is material — and worth testing before use.
3. Cross-check against roic.ai, which returns ev_to_ttm_ebitda and ev_to_ttm_ebit as a pair on the same period. A vendor EV/EBITDA sitting next to the EBIT figure rather than the EBITDA figure settles it in one call.
4. The distortion is proportional to goodwill, so suspect it hardest on exactly the names where the multiple matters most: post-merger primes, serial acquirers, LBO'd issuers.
5. This trap runs against the company — it makes the business look more expensive than it is. Correcting it belongs to building the bull case honestly, and it will not surface on its own, because nobody re-checks a number that already argued for caution.

What would falsify this

A vendor whose enterpriseToEbitda reconciles to hand-built EBITDA on a high-amortisation issuer. Test per vendor with the EBIT/EBITDA ratio check above; one clean reconciliation on a goodwill-heavy name is enough to clear that feed.

Related

• [[pitfall-vendor-ev-inverts-net-cash]] — the other half of the same multiple. On LHX both the EV numerator and the EBITDA denominator were wrong at once, and the float > sharesOutstanding tell fired on this issuer too.
• [[pitfall-entg-margins-corrupted-in-vendor-feeds]] — same genre: a vendor margin field carrying the wrong line item.
• [[pitfall-multiple-trim-inherits-the-broken-vendor-field]] — what happens downstream when a broken multiple is written into a watchlist trim.
• [[principle-primary-source-beats-vendor]]

History

• 2026-08-19 — Found during /analyze LHX. The Yahoo snapshot's 32.09 was implausible for a defense prime with a $42B backlog; roic.ai's paired EBITDA/EBIT multiples on the same fiscal year identified the swap rather than merely disagreeing with it.