How loan stacking actually shows up in a bank statement

The textbook case

Two deposits arrive in a single week. The descriptions contain the word "FUNDING" or carry a known funder's name (QUICKFUND CAPITAL, VELOCITY LENDING, KAPITUS, ON DECK). The amounts are round, the timing is suspicious, and any underwriter who has seen ten files knows what they are looking at.

This case still happens. It accounts for the easy half of loan-stacking detection, the half that ten-year-old keyword rules still catch. The harder half is the rest of this post.

Disguise pattern 1: platform-mimicking descriptions

An MCA funder who wires money to the merchant has incentive to make the deposit look ordinary. The description on the ACH or wire is something the originating bank does not validate. A funder operating under "BLUE OAK FINANCIAL" can describe the deposit as "BLUEOAK PAYOUT," and to a keyword-based detector that reads more like a delivery-platform settlement than a lender wire.

The give-away is not the description. It is the counterparty entity. A real platform settles thousands of merchants every week. A disguised lender settles a handful. Counterparty resolution against historical platform signatures separates the two.

Disguise pattern 2: fragmentation across accounts

A merchant with three operating accounts can split a single funding round across all three. Each account shows one deposit, well-spaced from the next. None of the three individual statements trip a same-window-multiple-lender flag.

This is the pattern most often missed by single-account analysis. It requires reading the merchant's full bank picture, including any account the merchant did not voluntarily submit. The detection signal is consistency of amount and timing across accounts, not concentration within any one account.

Disguise pattern 3: factor company channels

Freight factoring is a legitimate financing channel for trucking businesses. Factor companies (Triumph, RTS, Apex, OTR) buy receivables and pay merchants up front. The deposits look superficially like lender wires.

Sophisticated stackers use this cover. A non-factor lender describes its wire to look like a factor settlement, and a keyword-based stacking detector trained on trucking files passes it through.

Industry-aware detection helps here. The transportation segment has a known set of factor entities; deposits from outside that set, described to mimic factor entries, are surfaced as candidates for stacking, not for factoring.

Disguise pattern 4: split-tranche funding

A 30,000-dollar funding round delivered as three 10,000 wires across a week, each described differently. Each wire individually fits within a benign deposit pattern; in aggregate they are a stacking event.

Detection here is statistical, not keyword-based. The signal is the joint distribution of amount, timing, and counterparty across the entire statement.

What Vyaso does

The loan-stacking layer reads the entire statement set, applies counterparty resolution against a maintained set of funder, fintech-lender, factor, and platform entities, and evaluates timing and amount jointly across accounts. Every flag carries a confidence figure and the specific deposits that contributed.

The underwriter sees a single line: "Loan stacking suspected (confidence 0.83). 2 lender deposits, 5 days apart." With one click into the flag, the deposits and the entity resolution are exposed.

What committee should ask

When the loan-stacking flag fires, the questions that should be asked at committee are not "is this fraud" but the following.

1. What confidence does the engine assign?
2. Which counterparties were resolved as lenders, and on what evidence?
3. Are the deposits across multiple accounts, or concentrated in one?
4. Has the merchant disclosed any active facilities?

The answers should be in the analysis. They are.