ARC Comply
Collateral Screening

One asset.Two lenders.Neither one knows.

Double pledging is invisible from inside a single tape. We reconcile the whole book, across every warehouse, SPV and term deal, and measure the risk.

And it doesn’t stop there. The same run surfaces shared collateral, re-cut tapes and origination fraud. Patterns no single lender can see alone.

ARC Comply Collateral Screening run: £14.3m of balance exposed, 6.0% of a £238m pool, 31 findings to review on 29 assets, 14,860 assets screened. Pair-score distributions for duplicate assets and shared collateral, with the findings that clear a movable 0.85 risk threshold shaded at the top of each, and findings broken out by type and split between issues sitting across two books and issues inside one book (illustrative, synthetic data)
In practice

From raw tapes to a signed-off report

Four steps, from the files you already produce to a report a human can sign.

01 · IngestA raw tape mapped into the canonical field model

Any facility, any layout

Upload the tapes as they were exported. Mapping to the canonical model is automatic and analyst-correctable.

02 · ScreenEvery loan compared with every other, one pair matching

Every loan against every other

Every comparison surface, both directions, the full population. On every tape cut, not twice a year.

03 · ReviewLinked findings collapsed into a single case

One case per situation

Per-field evidence, the matched records side by side, and a recorded disposition.

04 · ReportA signed report carrying its fingerprint, config hash and engine version

The bundle a human signs

Findings, cleared population, limitations and provenance. Re-running the report on the same inputs reproduces it exactly.

What it does

Every pledge tested. Every alert evidenced.

Duplicate pledges surfaced

  • One asset, two vehicles: the same loan pledged into a second warehouse line, SPV or term deal — including collateral left behind in a warehouse after the sale.
  • Matched on substance, not labels: re-cut the tape and renumber the loans, the pair still surfaces.

Segregated by design

  • Scoped reporting: the book is screened as a whole; each lender sees evidence for its own facility only.
  • No new legal machinery: runs under the audit rights already in the facility documents.

Continuous surveillance

  • By tape or by API: screen at every borrowing base cut, or as the book changes.
  • No waiting for the annual exam: a duplicate introduced after closing surfaces at the next cut.

Audit-ready evidence

  • Evidence on every alert: the matched records, the fields that drove the match, the recorded disposition.
  • A report you can hand over: dated, loan-level, sized as balance exposed, reproducible on the same inputs.
Coverage

Every loan in the tape. Not a sample.

We screen the whole tape: every loan you send, matched against every other loan, in both directions, on every run. Nothing is sampled, nothing is set aside. That matters because a double pledge has two legs, sitting in different places on the book — a sample finds it only when the draw happens to pull both.

A sampled audit checks a scattering of loans across the book; ARC Comply checks every row of the tape (illustrative)
What we compare

Four comparison surfaces, one screen

Each book on its own comes back clean. The finding sits in the space between them.

Surface 01Two rows of one loan tape matched to each other

Within one tape

One loan booked twice, or two loans written independently against the same asset, inside a single pool.

Surface 02The same row matched between two tape cut-offs

Across cut-offs

A loan re-aged or churned to a new reference between two snapshots of the same book.

Surface 03A pool row with no matching row in the origination record

Pool against origination

A loan in the pool that the originator's own record does not support, or supports at a different size.

Surface 04A row in one book matched to a row in another book

Pool against another book

One asset, two funding vehicles, unrelated references. Neither side can see it from its own tape.

What it detects

Four categories of finding

Findings join into one case per situation, so the same event is never reviewed twice.

Duplicated loans

The same loan presented twice, re-papered under a new reference, or re-aged between snapshots so it reads as new business.

Shared collateral

Two loans, independently negotiated, secured on one asset.

Performance and value

Balances that never pay down the way the schedule says, and advances that sit above what the asset is worth.

Unsupported loans

Loans in the pool that the originator's own system of record does not account for.

False positives

Coincidence sits below the line

Economic neighbours exist in every book: loans that look alike because the market wrote them alike. The engine is built to keep them out of the review queue.

Score distribution of every compared pair on a log scale: the bulk of pairs score low, a band of coincidental economic neighbours sits just below the review threshold, and the findings that reach the queue sit above it (illustrative, synthetic data)

Weak fields never flag

Rate, term and region agree by chance all the time. They corroborate a finding; on their own they never raise one.

Severity needs an anchor

A high score with nothing hard agreeing behind it is not a critical finding, whatever the arithmetic says.

The dials are yours

Review threshold, field weights and tolerances are analyst-controlled and moved live, and every run records the configuration it used.

Why ARC Comply

Collateral integrity in structured finance

Every loan, not a sample

Every asset across every vehicle, at every borrowing base cut. Whole-of-book reconciliation as a standing control, not an annual exercise.

vs two agreed-upon-procedures engagements a year on a sample

Beyond loan IDs

Similarity resolution on the loan's real characteristics. Re-cut the tape, change the identifiers: the duplicate still surfaces.

vs identifier matching a spreadsheet edit defeats

No registry, no network

Works on one originator's book from day one. No shared platform, no industry adoption curve, no waiting.

vs central registries that only catch what participants upload

Explainable

A detailed, audit-ready rationale on every alert, defensible to lenders, trustees and regulators.

vs a clean / not-clean flag no one can stand behind

Who it's for

Evidence for every side of the deal

Originators

Evidence a clean book to your lenders and investors.

Lenders

Independent evidence behind every borrowing base, across the originator's whole book.

Investors

Comfort that the loan pool is screened for double pledging.

Arrangers & structurers

Screen the pool before pricing. Protect your investors, and market the deal with evidence behind it.

Book a demo

See a clearer path to compliance

We'll show you explainable screening tuned to your policies.

Book a demo