The shadow-test report
See the report before discussing a use case.
The historical replay asks what specified controls would have flagged on an invoice tape. The public sample uses invented data; nothing moved money. It is not proof of performance on a lender's book.
Control profiles
For strict, middle and permissive profiles, the report shows which invoices would have been held in the simulated replay, the reasons, and which holds depend on a cap rather than an unseeded payee history. A hold is a review referral, not a refusal.
Possible duplicates
Invoices with the same borrower, debtor and amount, dated close together. They are questions to check: a tape cannot by itself tell a re-raised invoice from two genuine deliveries.
The book and the stop ladder
Concentration by borrower and debtor, plus what the ladder would have done. The sample book is not rated.
Context checks
Amounts against each borrower and debtor's earlier history; invoices split below a declared approval limit; large first invoices to new parties; prices compared with a declared price list or mapped market series. The report states coverage, missing inputs and why each invoice was flagged.
Borrowing-base reconciliation
Aging-roll breaks, receivables falling without collections and eligibility against declared terms. These are reconciliation questions, not findings. Missing second-period tapes or agreement terms are explicitly NOT EVALUATED.
What a flag means
A flag is worth a look, not a finding of fraud. Thresholds in this sample are stated priors, not tuned on real outcomes. Planted cases show the output format, never a catch rate.
Discuss a use case
Read the format guide and data-sharing boundaries. Contact gaurav@reincredit.com before sharing any business records. This page does not invite a new upload or promise key issuance.