Bankruptcy vs Corporate Default: Building the Right Event Dataset

Separate missed payments, covenant breaches, acceleration, restructurings, and court bankruptcy filings before calculating distressed-equity outcomes.

Bankruptcy and corporate default are related but different event concepts. A company can miss a payment or breach a covenant without filing for bankruptcy, enter a restructuring before court proceedings, cure a default, or file for bankruptcy after a longer distress sequence. Alphanume's Corporate Default Events dataset identifies default events from SEC filing text and supplies the obligation, amounts, dates, acceleration status, evidence, and source filing when explicitly supported.

The Corporate Default Events guide provides the dataset workflow. The feed should be treated as a default-event source, not a bankruptcy docket and not a claim that every distressed equity is worth zero. Bankruptcy filings and later resolutions require separately defined evidence if they are part of the research question.

Define the event taxonomy first

The event table should distinguish a missed interest or principal payment, covenant breach, declared acceleration, restructuring agreement, and court filing. These stages can occur in sequence, on the same day, or not at all. Assign each observation to the narrowest supported category and keep the original evidence rather than forcing every row into a terminal bankruptcy label.

The served event_type describes the labeled default event. acceleration_declared is a separate three-state field, so null means the filing did not explicitly support the value. Preserve nulls across principal, accelerated amount, missed payment, default date, and grace-period date rather than filling them with zeros.

Retrieve and preserve the filing evidence

Use exact dates or a completed range to build the event panel. Each returned row includes an event date, ticker, filing URL, confidence, and evidence quote, plus enriched obligation fields. The event date is the feed's ordering axis, while default_date is the date stated in the filing when available. Keep both because the economic event and public filing can differ.

GET /v1/corporate-default-events
date_gte=2024-01-01
date_lte=2025-12-31

retain event_date, ticker, event_type, obligation_name,
       obligation_type, creditor_name, principal_outstanding_usd,
       amount_accelerated_usd, missed_payment_amount_usd,
       default_date, grace_period_end_date,
       acceleration_declared, evidence_quote, filing_url

research_event = taxonomy(event_type, evidence_quote)

Do not calculate returns until the public-event rule is explicit. If only a date is available, map it consistently to a session and describe the limitation. The filing URL is the primary audit trail for whether a row represents the first public disclosure, an update, or a repeated description of an earlier default.

Build sequences instead of isolated labels

A distress study becomes more informative when events are linked into issuer sequences. Sort by a durable issuer mapping and public event date, then identify first default, grace-period end, acceleration, restructuring, bankruptcy, cure, and other later resolution where those events are separately evidenced. Do not let a later filing overwrite what was known at the first observation.

• Use the first supported default disclosure as the starting event for a default study.
• Keep repeated updates in the sequence but outside the first-event denominator.
• Right-censor issuers with no observed later resolution at the dataset cutoff.
• Record ticker changes, delistings, and missing-price outcomes explicitly.

The Corporate Default Events feed does not itself provide a complete legal resolution chain. If bankruptcy petitions, court dates, or recoveries are required, join a separately sourced and dated table. Keep provenance per milestone so a filing-derived event is not confused with a court-derived event.

Amounts also require separate columns. Principal outstanding, missed payment, and accelerated amount answer different questions and can be null when the filing does not state them. Do not sum those fields into a generic loss estimate. Retain the obligation type and creditor name so repeated disclosures about the same debt can be reviewed before event counts are finalized.

Avoid terminal-outcome and timing bias

The main failure mode is treating every default as immediate bankruptcy and every bankruptcy as zero equity value. Defaults may be cured, waived, refinanced, restructured, or followed by court proceedings. Equity can keep trading, be halted, move venues, or eventually cancel. Preserve those outcome states rather than dropping securities when conventional price histories end.

A second failure is conditioning the first-event sample on future resolution. Selecting only defaults that later filed bankruptcy creates lookahead. Define the default cohort at the public event, then measure subsequent paths. Report how many events lack explicit dates, amounts, prices, or later milestones.

Create a staged distress ledger

As a concrete next action, retrieve a fixed historical window and audit a small sample of evidence quotes against the source filings. Build one row per first default with its narrow event category, obligation details, public date, stated default date, grace-period end, and acceleration status. Add later restructuring or bankruptcy milestones only from separately documented sources. Then calculate outcomes by starting category and publish the missing-outcome table. This design compares bankruptcy with earlier default stages without assuming that every distress disclosure reaches the same terminal state.