Skip to content

Deal and Tranche Identity / Data product

Deal and Tranche Identity: Making changing documents persist

Separating enduring economic identity from preliminary, final, and amended report occurrences, with inspectable matching and human confirmation.

Product concept

Persistent lineage and matching design; deployment is not claimed.

The problem

Renamed facilities and fresh report-level IDs can fragment the history of one economic facility. Borrower names alone cannot establish identity.

My responsibility

  • — Persistent identity model
  • — Deal and tranche matching requirements
  • — Candidate narrowing and weighted features
  • — Tie handling and human confirmation
  • — Historical validation and ingestion maturity

What we’re connecting

  1. 1Extract tranches
  2. 2Find candidate deal
  3. 3Confirm deal
  4. 4Suggest tranche matches
  5. 5Confirm lineage
  6. 6Persist relationships
  7. 7Continue to scoring

Sanitized representative artifact

Decisions, alternatives, and trade-offs

04 / Data model

Separate identity from occurrence

Report-level TrancheIDs change even when the economic facility persists.

DealID → TrancheLineageID → report-level TrancheID occurrences.

05 / AI & trust

Automate candidate discovery before automating judgment

Borrower name alone cannot determine whether an agreement belongs to an existing deal.

Automate candidate discovery before judgment.

06 / AI & trust

Make human review become exception handling over time

Default human review limits scale, but unsupported autonomy creates risk.

Capture corrections, measure errors, calibrate, then automate high-confidence cases.

13 / Workflow

Deterministic fact outranks probabilistic similarity

Exact duplicates and similar precedents answer different questions.

Deterministic fact outranks probabilistic similarity.

Persistent identity, changing occurrences

Deal 481 → persistent economic tranche
├ Preliminary → report occurrence 101
├ Final → report occurrence 177
└ Amendment → report occurrence 245

Sanitized representative artifact · Illustrative data

91% likely same deal · 68% review required · 42% no reliable match. Illustrative confidence only; thresholds require historical calibration and expert review. Ties are escalated rather than silently resolved.

What I learned / what comes next

  • Identity and occurrence are different concepts.
  • Automate candidate discovery before automating consequential judgment.

Validate against historical groupings, ties, renamed facilities, closed tranches, and bulk-ingestion cases.

Examples are recreated or sanitized to protect confidential company, client, and legal information. They preserve the product problem, decision logic, and working method without reproducing proprietary content.

Inspect product artifacts