Sample report
The actual report example. Synthetic data, detailed findings.
A replay of 2,006 funded invoices from the synthetic test tape, with seven context findings. The fixed run date is 13 October 2026; some reference data has later dates.
Download the sample report as text ↗ · Read the format guide ↗
Sample findings
4. WHAT CONTEXT CHECKS WOULD HAVE FLAGGED -- CHECK THESE
We show what these checks would have flagged on your past invoices.
Judged against their own history: 1,606 of the 2,006 funded invoices. Flagged: 14 invoices, in 7 findings.
Each invoice is judged only against invoices funded before it. The thresholds are stated priors, not yet tuned on outcomes, and a flag is worth a look, not a finding of fraud.
Amounts are compared after adjusting for general price changes, using the U.S. Bureau of Labor Statistics producer price index for final demand (as fetched on 2026-10-17).
Compared with your price list: 23 of the 25 funded invoices that name an item; 2 name an item not on it; 0 were billed below half its price.
Compared with public market prices: 1 of the 1 funded invoices whose item you mapped to a reference series; 24 name an item you did not map; 0 were not compared because their unit is not the reference's; 0 are dated before the reference starts. The reference runs to the week of 2026-10-12 (Source: U.S. Energy Information Administration).
| Invoices | Why |
|---|---|
| SAMPLE-INV-03-2-20 | this invoice was 9.8 times the usual amount for this borrower and debtor (funded 2026-07-19) |
| SAMPLE-INV-04-1-06 | billed at 2.5× the EIA weekly U.S. average retail diesel price for the week of 2025-12-15 (Source: U.S. Energy Information Administration) |
| SAMPLE-INV-04-1-12 | billed at 2.4 times the price on your declared price list for item API-CALLS (per 1K-tokens) |
| SAMPLE-INV-19-3-15 and SAMPLE-INV-19-3-16 and SAMPLE-INV-19-3-17 and SAMPLE-INV-19-3-18 and SAMPLE-INV-19-3-19 | from 2026-05-06, 5 invoices for this borrower and debtor ran at about 5.8 times the usual amount |
| SAMPLE-INV-D15-1 and SAMPLE-INV-D15-2 | possible duplicate: same borrower, debtor and amount, each invoice dated under 7 days after the one before; 2 invoices to the same debtor within 2 days, each just below the approval limit and together above it (from 2026-06-19); 2 invoices to the same payee within 2 days, each just below the approval limit and together above it (from 2026-06-19) |
| SAMPLE-INV-P11-1 | the first funded invoice paid to a payee this borrower had not used before was 5.5 times the usual amount for this borrower (funded 2026-08-21); this invoice was 5.9 times the usual amount for this borrower and debtor (funded 2026-08-21) |
| SAMPLE-INV-S7-1 and SAMPLE-INV-S7-2 and SAMPLE-INV-S7-3 | 3 invoices to the same debtor within 4 days, each just below the approval limit and together above it (from 2026-08-02); 3 invoices to the same payee within 4 days, each just below the approval limit and together above it (from 2026-08-02) |
By chance alone, the amount checks would be expected to flag about 3 invoices on a tape with no fraud, if amounts behave as the model assumes.
400 invoices were not judged against their own borrower-debtor history because there was too little of it. These include the first invoices of every new relationship, where some fraud patterns concentrate.
Outcomes: of the 1,440 funded invoices due at least 90 days before 2026-10-13 (from 20 borrowers), 5 were flagged, from 2 borrowers; 0 of those ended failed or written off. That is too few to compare with the invoices not flagged: a comparison needs at least 30 flagged invoices, at least 5 of them failed or written off, from at least 10 borrowers.
NOT EVALUATED (so not scored as clean)
- cumulative and outstanding caps (they are enforced against a live book under a lock, not a historical tape)
- checks needing `cure_date`, which your tape did not include
1. WHAT WOULD NOT HAVE LEFT THE BUILDING
would have been held for review before money moved, by control profile
of which would have counted as money stopped, by control profile
A hold is a referral, not a refusal: the engine holds the payment before it moves and a person approves or rejects it.
1. WHAT WOULD NOT HAVE LEFT THE BUILDING
SAMPLE ONLY: every amount in this section is invented money from a tape built for this demo. None of it is money stopped for anyone. Strict caps one payment at 50% of the borrower's largest funded invoice in the tape, and most invoices in this invented tape are above that, so most are held.
the new-payee check is UNSEEDED: no known-payee list was supplied, so the only history it has is this tape. A payee's first appearance in the window is therefore a first appearance and NOT evidence that the borrower had not paid it before -- a payee of nine years' standing looks identical. These holds are not counted as money stopped. The example has no lender-supplied known-payee list.
STRICT
MIDDLE
PERMISSIVE
2. POSSIBLE DUPLICATES WITHIN YOUR OWN TAPE -- CHECK THESE
2. POSSIBLE DUPLICATES WITHIN YOUR OWN TAPE -- CHECK THESE
| Invoices | Why |
|---|---|
| SAMPLE-INV-D15-1 and SAMPLE-INV-D15-2 | possible duplicate -- same borrower, same debtor, same amount (9,200.00), each invoice dated under 7 days after the one before, each funded; check these |
Cross-lender check: not evaluated. This report does not compare real lenders or their records.
3. THE BOOK, AND WHAT THE LADDER WOULD HAVE DONE
3. THE BOOK, AND WHAT THE LADDER WOULD HAVE DONE
Book grade: not rated. The sample tape does not show the controls and model facts a grade would need. Ratings remain a research goal.
concentration: 20 borrowers, 80 debtors, 4,461,496.30 funded
STRICT - what the ladder would have done
MIDDLE - what the ladder would have done
PERMISSIVE - what the ladder would have done
5. BORROWING BASE -- RECONCILIATION QUESTIONS
These are RECONCILIATION QUESTIONS about your files, not findings and not exceptions. Each says what would answer it. One becomes an exception only when your own records do.
aging-roll breaks: NOT EVALUATED -- it needs a SECOND tape -- the next period's aging, so that an invoice which left the book can be seen to have left -- together with your declared ineligible age and basis. None was supplied.
AR falling without collections: NOT EVALUATED -- it needs a SECOND tape, to compare one period's open balance with the next. None was supplied.
eligibility against your terms: NOT EVALUATED -- it needs the terms from your own agreement: an advance rate, a per-debtor concentration cap, an ineligible age and basis, and any excluded statuses. None was supplied.
Plain-text report
Read the sample text +
REIN SHADOW TEST - what our controls would have done with your tape
===================================================================
Synthetic report. No money moved. No real lender records are represented here.
Rows read: 2006. Funded: 2006.
1. WHAT WOULD NOT HAVE LEFT THE BUILDING
SAMPLE ONLY: every amount in this section is invented money from a tape built for this demo. None of it is money stopped for anyone. Strict caps one payment at 50% of the borrower's largest funded invoice in the tape, and most invoices in this invented tape are above that, so most are held.
the new-payee check is UNSEEDED: no known-payee list was supplied, so the only history it has is this tape. A payee's first appearance in the window is therefore a first appearance and NOT evidence that the borrower had not paid it before -- a payee of nine years' standing looks identical. These holds are not counted as money stopped. The sample has no lender-supplied known-payee list.
STRICT
assumes: strict: one payment may be up to 50% of the borrower's largest funded invoice, outstanding up to 40% of everything it funded, and a payee it has not paid before is held for a person
held before money moved: 1522 invoices, 3,399,415.00
- a payee this borrower had not paid before: 21 (50,220.00)
- over the per-payment approval threshold: 1501 (3,349,195.00)
of which countable as MONEY STOPPED: 1517 invoices, 3,389,575.00 -- holds that rest on a cap, or on a payee the lender's own known-payee list does not contain
NOT counted as money stopped: 5 invoices, 9,840.00 held only because the payee first appears in this tape, with no known-payee list to check it against: SAMPLE-INV-03-0-00, SAMPLE-INV-07-0-00, SAMPLE-INV-11-0-00, SAMPLE-INV-15-0-00, SAMPLE-INV-19-0-00
MIDDLE
assumes: middle: one payment may be up to 100% of the borrower's largest funded invoice, outstanding up to 70% of everything it funded, and a payee it has not paid before is held for a person
held before money moved: 21 invoices, 50,220.00
- a payee this borrower had not paid before: 21 (50,220.00)
of which countable as MONEY STOPPED: 0 invoices, 0.00 -- holds that rest on a cap, or on a payee the lender's own known-payee list does not contain
NOT counted as money stopped: 21 invoices, 50,220.00 held only because the payee first appears in this tape, with no known-payee list to check it against: SAMPLE-INV-00-0-00, SAMPLE-INV-01-0-00, SAMPLE-INV-02-0-00, SAMPLE-INV-03-0-00, SAMPLE-INV-04-0-00, SAMPLE-INV-05-0-00, SAMPLE-INV-06-0-00, SAMPLE-INV-07-0-00, SAMPLE-INV-08-0-00, SAMPLE-INV-09-0-00, SAMPLE-INV-10-0-00, SAMPLE-INV-11-0-00, SAMPLE-INV-12-0-00, SAMPLE-INV-13-0-00, SAMPLE-INV-14-0-00, SAMPLE-INV-15-0-00, SAMPLE-INV-16-0-00, SAMPLE-INV-17-0-00, SAMPLE-INV-18-0-00, SAMPLE-INV-19-0-00, SAMPLE-INV-P11-1
PERMISSIVE
assumes: permissive: one payment may be up to 150% of the borrower's largest funded invoice, outstanding up to 100% of everything it funded, and any payee is allowed -- only the caps bind
held before money moved: 0 invoices, 0.00
- nothing, under these caps
of which countable as MONEY STOPPED: 0 invoices, 0.00 -- holds that rest on a cap, or on a payee the lender's own known-payee list does not contain
2. POSSIBLE DUPLICATES WITHIN YOUR OWN TAPE -- CHECK THESE
- SAMPLE-INV-D15-1 and SAMPLE-INV-D15-2: possible duplicate -- same borrower, same debtor, same amount (9,200.00), each invoice dated under 7 days after the one before, each funded; check these
cross-lender check: not evaluated in this sample
3. THE BOOK, AND WHAT THE LADDER WOULD HAVE DONE
book grade: not rated. A tape doesn't show the controls and model facts a grade needs. Ratings remain a research goal.
concentration: 20 borrowers, 80 debtors, 4,461,496.30 funded
largest borrower 9.1% of funded value; top five 31.9%
largest debtor 4.7% of funded value; top five 11.2%
STRICT - what the ladder would have done
- no borrower would have been paused or frozen
MIDDLE - what the ladder would have done
- no borrower would have been paused or frozen
PERMISSIVE - what the ladder would have done
- no borrower would have been paused or frozen
4. WHAT CONTEXT CHECKS WOULD HAVE FLAGGED -- CHECK THESE
We show what these checks would have flagged on your past invoices.
Judged against their own history: 1,606 of the 2,006 funded invoices. Flagged: 14 invoices, in 7 findings.
Each invoice is judged only against invoices funded before it. The thresholds are stated priors, not yet tuned on outcomes, and a flag is worth a look, not a finding of fraud.
Amounts are compared after adjusting for general price changes, using the U.S. Bureau of Labor Statistics producer price index for final demand (as fetched on 2026-10-17).
Compared with your price list: 23 of the 25 funded invoices that name an item; 2 name an item not on it; 0 were billed below half its price.
Compared with public market prices: 1 of the 1 funded invoices whose item you mapped to a reference series; 24 name an item you did not map; 0 were not compared because their unit is not the reference's; 0 are dated before the reference starts. The reference runs to the week of 2026-10-12 (Source: U.S. Energy Information Administration).
- SAMPLE-INV-03-2-20: this invoice was 9.8 times the usual amount for this borrower and debtor (funded 2026-07-19)
- SAMPLE-INV-04-1-06: billed at 2.5× the EIA weekly U.S. average retail diesel price for the week of 2025-12-15 (Source: U.S. Energy Information Administration)
- SAMPLE-INV-04-1-12: billed at 2.4 times the price on your declared price list for item API-CALLS (per 1K-tokens)
- SAMPLE-INV-19-3-15 and SAMPLE-INV-19-3-16 and SAMPLE-INV-19-3-17 and SAMPLE-INV-19-3-18 and SAMPLE-INV-19-3-19: from 2026-05-06, 5 invoices for this borrower and debtor ran at about 5.8 times the usual amount
- SAMPLE-INV-D15-1 and SAMPLE-INV-D15-2: possible duplicate: same borrower, debtor and amount, each invoice dated under 7 days after the one before; 2 invoices to the same debtor within 2 days, each just below the approval limit and together above it (from 2026-06-19); 2 invoices to the same payee within 2 days, each just below the approval limit and together above it (from 2026-06-19)
- SAMPLE-INV-P11-1: the first funded invoice paid to a payee this borrower had not used before was 5.5 times the usual amount for this borrower (funded 2026-08-21); this invoice was 5.9 times the usual amount for this borrower and debtor (funded 2026-08-21)
- SAMPLE-INV-S7-1 and SAMPLE-INV-S7-2 and SAMPLE-INV-S7-3: 3 invoices to the same debtor within 4 days, each just below the approval limit and together above it (from 2026-08-02); 3 invoices to the same payee within 4 days, each just below the approval limit and together above it (from 2026-08-02)
By chance alone, the amount checks would be expected to flag about 3 invoices on a tape with no fraud, if amounts behave as the model assumes.
400 invoices were not judged against their own borrower-debtor history because there was too little of it. These include the first invoices of every new relationship, where some fraud patterns concentrate.
Outcomes: of the 1,440 funded invoices due at least 90 days before 2026-10-13 (from 20 borrowers), 5 were flagged, from 2 borrowers; 0 of those ended failed or written off. That is too few to compare with the invoices not flagged: a comparison needs at least 30 flagged invoices, at least 5 of them failed or written off, from at least 10 borrowers.
5. BORROWING BASE -- RECONCILIATION QUESTIONS
These are RECONCILIATION QUESTIONS about your files, not findings and not exceptions. Each says what would answer it. One becomes an exception only when your own records do.
aging-roll breaks: NOT EVALUATED -- it needs a SECOND tape -- the next period's aging, so that an invoice which left the book can be seen to have left -- together with your declared ineligible age and basis. None was supplied.
AR falling without collections: NOT EVALUATED -- it needs a SECOND tape, to compare one period's open balance with the next. None was supplied.
eligibility against your terms: NOT EVALUATED -- it needs the terms from your own agreement: an advance rate, a per-debtor concentration cap, an ineligible age and basis, and any excluded statuses. None was supplied.
NOT EVALUATED (so not scored as clean)
- cumulative and outstanding caps (they are enforced against a live book under a lock, not a historical tape)
- checks needing `cure_date`, which your tape did not include
-------------------------------------------------------------------
tape: 2006 rows, 20 borrowers, sha256 a83cfdf8b6f12434…
columns absent: cure_date
run at 2026-10-13T12:00:00+00:00 (a fixed sample date)