Methodology & KPI Definitions

All KPIs are computed in the browser at page load from the full set of Report records — there is no hand-maintained total. Each dossier carries its own peak_drawdown_pct and market_value_corrected_bn fields and its own outcomes list. The headline figures are simply aggregates of those per-report fields. Click method on any KPI to jump to its definition here; click events to see the underlying records.

Formula: count of all Report records.

Each record is one published short report by a tracked forensic firm. The count includes reports regardless of whether they produced a material outcome. Underlying events: The Archive.

Formula: count of distinct values in the firm field across all Report records.

This is the number of separate forensic firms whose work is archived, not a fixed editorial number — it grows as new desks are added. Underlying events: The Archive (filter by firm).

Formula: sum(peak_drawdown_pct) / count(all reports), rounded to the nearest whole percent.

Per-report drawdown definition: the peak-to-trough decline in the share price measured from the closing price on the trading day before the report was published to the lowest closing price within 90 calendar days after publication, expressed as a percentage. Prices are split-adjusted for any stock splits occurring within the window.

Missing data: reports with no verified drawdown are treated as 0% in the average (i.e., no decline recorded), not excluded — so the average is conservative and includes misses.

FX: for non-USD listings the drawdown is computed in the local listing currency; only the market-value correction (below) is converted to USD.

Formula: sum(market_value_corrected_bn) across all Report records, in USD billions.

Per-report definition: the reduction in equity market capitalization from the closing price the trading day before publication to the closing price 90 calendar days after publication — or to the delisting/bankruptcy date if earlier — split-adjusted, stated in USD billions.

FX: non-USD market caps are converted to USD at the period-end exchange rate on the measurement date.

Limitation: where the same company was covered by more than one firm, its correction may be counted more than once. We flag this rather than silently de-duplicate; see Limitations.

Formula: count of Report records whose outcomes array includes Executive Resignation.

One count per report, regardless of how many executives resigned. Underlying events: Accountability Ledger.

Formula: count of Report records whose outcomes array includes Regulatory Investigation.

Covers formal inquiries, enforcement actions, and trading halts initiated by a regulator. Underlying events: Accountability Ledger.

Formula: count of Report records whose outcomes array includes Delisting or Bankruptcy.

The two terminal outcomes are combined into a single KPI. Underlying events: Accountability Ledger.

Formula: total count of outcome tags across all reports, excluding No Material Outcome.

Because a single report can carry several outcomes (e.g., a resignation and an investigation), this figure can exceed the report count. Underlying events: Accountability Ledger.

We do not assert strict causation. An outcome is attributed to a report when it is (a) documented in the public record and (b) temporally follows publication within a reasonable attribution window. We present these as association, not proven causation — markets move for many reasons, and a drawdown or resignation after a report is not proof the report caused it.

Each dossier includes a timeline of dated events so readers can judge proximity for themselves, and links to the original report and primary sources. Readers are explicitly encouraged to compile their own track records and draw their own conclusions.

Per-report figures are stored on each Report record at the time the dossier is written, sourced from the original published report and public market data. Headline aggregates recompute live on every page load from the current Report set, so they update as soon as a dossier is added or corrected. The effective “last updated” date is the most recent report published_date.

• Double-counting: companies covered by multiple firms may contribute their correction more than once to the total.
• Survivorship: the archive reflects reports we have chosen to track; it is not a census of all short research ever published.
• Stale figures: per-report drawdowns and corrections are point-in-time measurements and are not re-marked to current prices.
• Attribution: outcomes are associated, not proven to be caused (see above).

Found an error in a figure? See our Corrections policy.