Every record in Dealert is intended to be defensible in a client deliverable. The principles below describe the rules a transaction must satisfy to enter the database, how missing values are estimated, and how the comparable-set engine selects precedents.
A transaction is admitted to Dealert only when at least one primary disclosure exists: a buyer or seller press release, regulatory filing, prospectus, or audited financial statement. Secondary references — Bloomberg, Reuters, S&P, industry trade press — are stored alongside the record but never serve as the basis for a reported value.
Where two primary sources conflict, the more recent and the more detailed is preferred; both are retained and time-stamped. The record carries an audit trail that any subscriber can inspect from the deal page.
For every monetary field — transaction value, enterprise value, target revenue, EBITDA, EBIT — Dealert distinguishes disclosed figures from estimated bands. Disclosed figures come directly from the primary source. Estimated bands are derived from the comparable-set median multiplied by the disclosed metric (e.g. EV from EV/EBITDA × EBITDA).
Estimates render as a range with a clear visual treatment ("$120–180M"). They are never blended into the disclosed column, and they are excluded from any aggregate statistic published as a single point estimate.
Every buyer, seller and sponsor is resolved to a single canonical entity. "KKR", "Kohlberg Kravis Roberts & Co. L.P.", and "KKR & Co. Inc." appear as one entity with one platform history. Each entity is classified as strategic or financial, with the subtype recorded (PE, VC, family office, hedge fund, infrastructure, sovereign wealth, asset manager, corporate, or other).
The classification is explicit per record and reviewable. Where strong heuristic signals exist but a primary source is silent, the field is left as unclassified rather than guessed.
Targets are coded against the Refinitiv Business Classification (TRBC) hierarchy, levels 1 through 5. Free-text sector descriptions are kept for human readability but never used for filtering or aggregation. Comparable sets and sector statistics are therefore reproducible because they reduce to a deterministic set of codes.
Where a target straddles two L5 codes, the primary code is chosen by revenue contribution if disclosed; otherwise by the activity the deal rationale references.
For each (industry × geography × size band × deal year) cell, Dealert computes the median EV/Revenue and EV/EBITDA across deals with disclosed figures. Estimates apply the median to the target's known metric. If fewer than four disclosed observations exist for the cell, the band widens to the parent L3 code; if still under-served, to L1.
Bands therefore carry implicit confidence: the narrower the band, the deeper the underlying comparable set. The methodology page in the live product surfaces this count per record.
The comparable-set engine ranks candidates against the subject deal on four dimensions: industry overlap (TRBC code distance), geographic overlap (country and region), size proximity (EV or revenue band), and recency. Each axis has an explicit weight, defaulted to the values shown on the comparable-set page and adjustable by the user.
The engine does not learn from clicks. It is a deterministic ranking function. Re-running the same query with the same weights returns the same ordered set.
Dealert publishes a quarterly review log of corrections, restatements and re-classifications. Every figure carries a date of last review. Subscribers can flag a record for re-review from the deal page; the methodology team responds within five business days.
Questions about a specific record or principle? Contact the methodology team at [email protected].