PollyOS Loan Margin Tracking: What to Verify

The test for PollyOS loan margin tracking is whether a lender can reconcile its original lock economics with final loan-sale proceeds and explain each difference. A margin figure is useful only if finance can reproduce the calculation. Traceability can reveal why a loan missed its target; it cannot guarantee profitability.

What the Polly Hedge launch combines

In its October 6, 2026 announcement, Polly said PollyOS combines its product, pricing and eligibility engine, new Polly Hedge system and relaunched Loan Trading Exchange on shared data. The company claims intraday pipeline updates, loan-level attribution and alignment with generally accepted accounting principles (GAAP). The release provides no independently measured profitability results.

For mortgage technology buyers, the opportunity is to compare decisions made at different stages without rebuilding the loan’s history. An evaluation should follow a loan through pricing, risk management and sale, with finance defining the margin measure before the demonstration starts.

Reconcile the original lock with settlement

Start with the original lock economics and preserve that baseline. Here, settlement means the lender’s sale to an investor, so the reconciliation must reach beyond the borrower’s closing.

For a proposed acceptance test, request four connected sets of records:

Specify how the margin calculation treats origination fees, interest earned and funding costs while awaiting sale, servicing value and operating expenses. Keep cash proceeds and accounting income in separate columns, and reconcile them. Have the controller or accounting adviser approve the definitions.

For Fannie Mae whole-loan sales, the agency’s purchasing guidance provides an external check. Its Purchase Advice identifies loans, price adjustments and proceeds; the guidance calls for lenders to reconcile the advice to their books and confirm the bank credit. Test the software’s total against those records.

Where a hedge covers several loans, ask how shared gains and losses are allocated. Reconcile those allocations, plus any unallocated residual, to the desk’s total. The OCC’s mortgage-banking handbook explains that effective pipeline hedging depends on estimating pull-through and fallout—the shares of commitments that close or fail to close as expected. Include canceled locks in the test population.

A hypothetical loan: explain an $800 shortfall

Suppose a lender expects at lock to fund a $400,000 loan at par and sell it for $408,000. Define contribution here as sale premium plus allocated hedge results. Assume no principal change or accrued-interest adjustment; exclude origination income, carrying costs, servicing value, operating expenses and taxes. Each change below is incremental and counted once.

Under these assumptions, the investor pays $406,800. The resulting $6,800 sale premium plus the $400 allocated hedge gain yields a $7,200 contribution. The $800 shortfall from the initial expectation equals 20 basis points of the assumed balance. This is hypothetical arithmetic, not a Polly customer result or net profit.

The repricing should link to its approval, the investor adjustment to settlement detail, and the hedge gain to a documented allocation. A balancing “other” entry could make the totals agree while leaving the cause unresolved. At settlement, also check how prior estimates are reversed or replaced so the reconciliation counts each gain or loss once.

Preserve the history behind every override

Ask the provider to demonstrate a correction entered after a loan has settled. The record should preserve the previous value, replacement value, person or process responsible, reason, approval and resulting margin change. Request both the date the event took effect and the date the system received it.

That distinction lets reviewers reconstruct what the desk knew when it acted. A corrected final record alone cannot answer whether an earlier decision used stale data.

Apply the same requirements to manual price overrides and changes in hedge-allocation assumptions. Request the applicable rule or model version, an exportable history and a way to reproduce prior reports. A later calculation should not silently rewrite the explanation of yesterday’s margin.

For missing or conflicting inputs, agree on an exception owner and resolution deadline. Keep unresolved differences visible in the reconciliation, including after month-end, until someone documents their disposition. These are proposed evaluation controls, not verified PollyOS features.

Verify availability before replacing a system

Polly said its initial customer-partner cohort had been onboarded, with broader availability to follow MBA Annual, October 11–14 in Chicago. As of October 7, the announcement did not establish universal availability or a firm general-release date.

Ask which functions are production-ready for the lender’s own investors and loan products, and put delivery commitments in writing. Then run a representative cohort of loans through the proposed platform and existing process across settlement and month-end.

For leaders responsible for lender operations, useful acceptance measures include unexplained dollar differences, manual accounting adjustments, unresolved exceptions and elapsed time between a source event and its appearance in the system. Measure reconciliation effort separately from profitability, and include implementation and ongoing platform costs in the business case.

Before retiring the existing process, require finance and capital markets to reproduce the same loan histories and agree on any remaining differences.