Reported vs. Calculated Multiples in Precedent Transactions

Every multiple in a precedent set has a provenance: either a disclosing party stated it in a public document, or you derived it by dividing two figures from separate disclosures. The first is a reported multiple. The second is a calculated multiple. Both are legitimate; an unlabelled mix of the two is a source of hidden inconsistency that is difficult to detect and easy to propagate.

What a reported multiple is

A reported multiple appears when a banker, adviser, acquirer or target explicitly states a transaction multiple in a public filing or announcement. The most common sources are:

• Fairness opinion selected-transaction tables in proxy statements (Form DEF 14A, Form DEFM14A) and target solicitation/recommendation statements (Schedule 14D-9, Schedule 13E-3). SEC Regulation M-A, Item 1015 requires any party soliciting shareholder approval in a merger to disclose the basis of any fairness opinion, including the procedures used to select transactions and the multiples applied. As a result, investment banks disclose their selected-transaction tables, usually in the form “Selected Precedent Transactions — EV/LTM EBITDA” with a range across the set.
• Deal press releases and investor presentations that state an implied multiple as part of the announcement narrative (e.g., “the transaction implies an EV/EBITDA multiple of 12.4×”).
• Acquirer earnings calls and investor day slides that describe a deal in multiple terms.

What makes a multiple “reported” is that the disclosing party made a methodological choice on your behalf: they selected the financial period, defined the metric (as-reported or normalized), and determined how to construct enterprise value. That choice is embedded in the stated figure and may not be fully transparent.

What a calculated multiple is

A calculated multiple is one you construct from two or more separately disclosed inputs. The typical steps:

1. Find the deal price. The Form 8-K filed at signing or at announcement discloses the offer price per share, total equity value, or total implied enterprise value. For tender offers, the Schedule TO states the offer price.
2. Construct enterprise value. Take the equity value from step one, add disclosed net debt (from the target’s most recent Form 10-K or Form 10-Q balance sheet), and add any other bridge items (see the equity value versus enterprise value guide and the net debt calculation guide).
3. Find the financial metric. Extract LTM EBITDA, LTM revenue, or another metric from the target’s public filings, typically the most recent annual or quarterly report filed before announcement.
4. Divide. EV ÷ EBITDA = the calculated EV/EBITDA multiple.

Because each input comes from a different document with its own as-of date, preparation method and disclosure convention, the calculation requires careful matching and explicit documentation.

How they differ in practice

Dimension	Reported multiple	Calculated multiple
Source	Stated by banker, adviser, or party	Derived by the analyst
EBITDA definition	Often normalized; adjustments may be undisclosed	As-reported (unless analyst adjusts and documents)
Financial period	Chosen by the disclosing party	Chosen by the analyst; must match the stated period
EV construction	Embedded in the stated figure	Constructed by the analyst from balance-sheet inputs
Reproducibility	Stated result is visible; methodology may not be	Fully reproducible if inputs and formula are recorded
Where it appears	DEF 14A, 14D-9, 13E-3, press releases	Analyst’s own workpapers

Neither type is inherently more reliable. A reported multiple from a regulated fairness opinion has the credibility of a disclosed adviser’s view; a carefully calculated multiple from two independently sourced filings is reproducible and auditable. The risk arises when the two are placed side by side in a precedent set without labels.

Why labelling matters for a precedent set

Suppose a precedent set includes three transactions. For Transaction A the analyst uses the EV/EBITDA from the target’s DEF 14A fairness opinion table, where the banker applied normalized EBITDA that excluded restructuring costs and a management fee add-back. For Transaction B the analyst calculates EV ÷ as-reported LTM EBITDA from the 10-K. For Transaction C a press release states the multiple.

The three figures will sit in a column labelled “EV/LTM EBITDA.” A reader who does not know their provenance cannot tell that one uses normalized EBITDA, one uses as-reported EBITDA, and one uses the acquirer’s own framing. If all three are then summarised as a median, the median conflates three different definitions of EBITDA into one number.

The solution is not to avoid one type of multiple. It is to:

• label every row with its source type (reported / calculated);
• for reported multiples, record the source document, filing date and disclosed period;
• for calculated multiples, record the formula, each input, its source document and its as-of date;
• normalize wherever possible — if you can reconstruct the metric on a consistent basis across all transactions, that is preferable to a mixed set, provided the reconstruction is documented; and
• flag where normalization is not possible and the comparison must be treated as approximate.

Consistency within a set

The EBITDA normalization guide covers the mechanics of removing non-recurring items and applying consistent lease and pension treatment across a set. That discipline applies regardless of whether you start from a reported multiple or a calculated one.

When you start from a reported multiple, the normalization has already happened inside the adviser’s workpapers, but you cannot always verify what adjustments were made. The fairness opinion text sometimes describes them (“we normalized for one-time restructuring charges of $12 million”) and sometimes does not. When the adjustments are opaque, the reported multiple is less useful for building a consistent set than a calculated multiple where you control the metric.

When you start from a calculated multiple, you apply normalization yourself. You can match the period (see the LTM/NTM guide), adjust for IFRS 16 lease capitalization if the industry is lease-intensive, and decide how to treat pensions and non-recurring items on the same basis across every transaction.

A precedent set made entirely of carefully calculated multiples, where every metric is derived on the same basis from each target’s public financials, is often more consistent than a set assembled from reported multiples sourced across multiple different advisers’ fairness opinions.

How the distinction affects AI-assisted analysis

If an LLM receives a precedent set without provenance labels, it cannot detect that some multiples embed normalized EBITDA and others use as-reported EBITDA. The model can calculate a median, rank transactions by multiple, and explain the range — and every step of the arithmetic can be technically correct while the conclusion conflates incompatible inputs.

The NIST AI Risk Management Framework Generative AI Profile (NIST AI 600-1) identifies confabulation and data provenance as primary risk categories for generative AI in high-stakes analysis. In transaction valuation, provenance means more than just “where did the number come from” — it means “what methodological choices does this number already embed.”

The practical solution is a provenance tag on every multiple in the data model:

• source_type: reported or calculated
• source_document: the specific filing (e.g., DEF 14A filed 2024-03-15 for Company X)
• metric_definition: as-reported / normalized / adjusted (with description)
• financial_period: LTM as of what date, or NTM for which fiscal year
• formula: for calculated multiples, numerator / denominator with source of each

The AI-ready precedent transaction data schema defines a field-level structure that captures these attributes. The verified deal data for AI valuation guide explains why a single wrong field propagates through an entire multiple calculation.

Common errors

Using a reported multiple without recording the period. A fairness opinion may say “EV/EBITDA of 11.5×” in the summary without specifying whether it is LTM or NTM EBITDA. The full selected-transaction table usually discloses the period; record it rather than assuming.

Comparing a reported normalized multiple with a calculated as-reported multiple. The normalization gap can easily be one to two turns of EBITDA for a target with material non-recurring items or IFRS 16 exposure. See the EBITDA normalization guide.

Using a press-release implied multiple without checking its basis. Acquirers sometimes state an implied multiple in a press release using forward EBITDA or consensus estimates. That is a different figure from LTM EBITDA and belongs in a different column.

Treating a calculated multiple as a reported one. If you derive a multiple yourself, label it as calculated, even if the figure appears in your output exactly as you found it in a third- party source. The provenance of your inputs, not the face of the number, determines the type.

Changing the type mid-analysis without notation. Some practitioners switch from reported to calculated when a fairness opinion is unavailable. That is acceptable; it must be flagged.

Documentation checklist per transaction

Item	Reported multiple	Calculated multiple
Source document	Filing type, filer, date, URL	n/a
Formula	n/a	Numerator source, denominator source, each filing and date
EV construction	As stated; note any disclosed bridge items	Own bridge: equity value + net debt, each input sourced
Financial metric	As stated; record disclosed period and any noted adjustments	As-reported or normalized; state adjustments applied
Period	As disclosed	LTM as of what date; or NTM for which fiscal year
Normalization	Unknown unless disclosed; flag if opaque	Document each adjustment and its basis
Currency	Record as disclosed	Record; convert only if needed, note rate and date

---

A well-sourced precedent set is one where a reader—or an AI system—can trace every multiple to its inputs, understand what methodological choices are embedded in it, and judge whether those choices are consistent with the rest of the set. Labelling each figure as reported or calculated, and recording the provenance either way, is the minimum standard that makes that review possible.

Access the PrivyLogic precedent transaction database to find sourced precedent deals with reported and calculated multiples distinguished and documented. Enterprise and custom enquiries: contact us.

Sources

1. SEC Regulation M-A — Item 1015: Fairness Opinions — U.S. Securities and Exchange Commission; accessed September 21, 2026
2. Regulation M-A Adopting Release (Release No. 33-7760) — U.S. Securities and Exchange Commission; accessed September 21, 2026
3. Artificial Intelligence Risk Management Framework — Generative Artificial Intelligence Profile — National Institute of Standards and Technology; accessed September 21, 2026