Trust starts with
showing the work.

What is available today

The private marketplace is a working preview. Consumers build a financing scenario, choose up to five originators who receive it, compare entered test responses, and preview contact consent and follow-up. Enrique Pelayo Jr is the first pilot profile. Two explicitly fictional test profiles let visitors exercise multiple selections in preview only. No live requests, email notifications, or public outcome scores are active. The free Market Desk, calculators, resources, and widgets remain available.

Private requests, chosen recipients

The request contains structured financing details and a random ID. It excludes name, email, phone, exact property address, Social Security number, and documents. Originators must be selected by the consumer and activated for the property state before receiving a live request. Identity remains private from the originator until the consumer authorizes contact sharing with that specific recipient. Platform operators still process the records to run the service.

Scenario details are fixed after submission. A changed scenario requires a new request. The pilot response window is 24 hours, and an originator response is not guaranteed. A quote is preliminary, carries an expiration, and does not constitute a rate lock or approval.

How offers are compared

Each response must match the request’s product, term, property, credit assumption, occupancy, amount, and requested lock duration. The originator supplies the rate, APR, points, lender fees, credits, and underlying lender. Quoted fees must include applicable compensation. Current comparisons use only the latest submitted version per originator, excluding expired versions and quotes without an available fee worksheet. A new version awaiting a worksheet does not revive an older version. An expired replacement never revives an older quote.

Consumers choose lowest upfront lender cost, monthly principal and interest, or modeled cost over their chosen timeframe. The timeframe calculation combines scheduled interest through that period and upfront lender cost, excluding principal repayment. It excludes third-party costs, taxes, insurance, mortgage insurance, and tax effects. Figures are estimates, not complete Loan Estimate comparisons.

Sorting uses unrounded values; ties use submission time and then record ID. The three highlights represent cost priorities, not necessarily three different originators. All submitted versions remain available in the private request history. A single response is shown as one response, not manufactured competition. Advertising and platform ownership do not influence ordering.

Worksheets first. Quality checks later.

Each originator attaches an itemized fee worksheet to complete a preliminary quote. It is labeled as a worksheet, not an official Loan Estimate. PDF presence does not verify the contents. Uploaded worksheets are private to the requester and submitting originator, subject to necessary platform administration.

After an introduction, the borrower can optionally share a redacted Loan Estimate for a private platform quality check. This file is not shared with originators. The borrower can report whether the loan closed, flag a problem, and request assistance without uploading a document. A received document or concern is not a completed review or evidence of misconduct.

For covered loans, a lender must provide a Loan Estimate after receiving the required application information. That information is supplied directly to the lender through its secure process. A shopper may request estimates from multiple finalists before choosing one. See the CFPB application-information guidance and Loan Estimate explainer.

Quality review compares matching scenarios, dated fees, lock status, and contemporaneous pricing. Market movement and changed circumstances require context. No automated penalty, document extraction, staffed assistance, or scheduled follow-up is running in preview mode.

The payment calculation

P is the loan amount, r is the annual note interest rate divided by 100 and then 12, and n is the term in years multiplied by 12. In this formula, the input is the note interest rate—not the disclosed APR. At 0% interest, monthly principal is P ÷ n. Results are calculated before display rounding; dollar amounts may be rounded to the nearest dollar.

This assumes a fully amortizing fixed-rate loan with equal monthly payments. It excludes taxes, insurance, HOA dues, mortgage insurance, and closing costs. Scheduled interest totals assume the loan runs its full term without extra payments or refinancing. Widgets do not model adjustable rates or interest-only payments.

Points, credits, and upfront cost

One point is 1% of the loan amount. Net upfront lender cost excludes third-party charges, prepaid expenses, escrow deposits, and other closing costs. A negative value is a credit toward eligible closing costs, not cash back. APR in a submitted response is supplied by the originator; the platform does not independently calculate or certify it. Compare real offers using official Loan Estimates.

Where market numbers come from

We use Treasury data for recent yields and FRED for historical Treasury and mortgage series. Treasury values take precedence on overlapping Treasury dates. We do not create a proprietary daily mortgage index from bond yields. A Treasury yield is not a borrower’s mortgage rate, and a weekly average is not today’s available lender offer.

Source checks use separate hourly caches. An update request after the cache expires triggers revalidation; the response may initially retain the previously cached observation while that refresh completes. Visible market pages and market widgets check our server every 15 minutes. Refreshing a widget does not make a weekly source update daily.

Dates, gaps, and changes

Every benchmark carries its actual reporting date. Changes compare the latest observation with the previous available observation from the same series. One basis point is 0.01 percentage point. Calendar lookbacks use the last observation on or before the target date, not a count of trading days.

Charts share a percent scale and calendar axis when comparing two series. Lines connect reported values; missing daily or weekly values are not invented. Exact tables leave unreported dates blank. Invalid, nonnumeric, and future-dated observations are excluded. The dashboard includes up to 400 calendar days of data.

We flag Treasury observations older than five calendar days and mortgage observations older than ten. These are display thresholds, not guarantees of freshness. Failed source checks are disclosed; missing values are not replaced with a made-up rate. Source revisions appear on a later successful refresh. The “What changed?” brief describes numerical changes only; it does not infer a cause or predict a rate move.

Our proposed Transparency Score

Proposed, not active. The goal is to measure how consistently a lender publishes complete, current pricing that matches independently checked quotes. It would not measure approval odds, creditworthiness, service quality, or whether a loan is right for you.

Proposed score = 0.35 × consistency + 0.30 × completeness + 0.20 × freshness + 0.15 × corrections, with each component normalized to 0–100. This is a design proposal, not a calibrated scoring system. Component rubrics, minimum sample requirements, and a documented dispute process must be established before any lender receives a public score.

The pilot proposal uses a rolling 90-day window and a 30-day evidence half-life. Coverage and sample counts would accompany the score. Insufficient evidence would show “Not yet rated”; missing evidence would never receive assumed positive marks. Ads, subscriptions, and referral compensation would be excluded.

Verification would combine identity checks, source records, and independently selected matching quotes. Market benchmarks could flag an outlier for review but could not prove a lender’s quote wrong. Any future badge must link to dated evidence and support expiration and correction. No Transparency Score badge is offered in the current widget builder.

Our widgets follow the same rules

The payment widget uses the same amortization calculation as our calculators. The market widget uses our existing benchmark pipeline, source credits, dates, and stale-data thresholds. Neither collects contact information or submits financial inputs. The default calculator rate is an example. Branding on a host website does not change the numbers or imply an endorsement.

Recent reviews and outcome evidence

External reviews will show the latest five dated reviews from a connected source within the past 12 months, in date order regardless of rating, with a link to the full source history. No review count or minimum star rating is required. No external source is connected yet. Source-specific terms govern what may be displayed and how it can be refreshed.

Private marketplace follow-up records service feedback, outcome stage, whether the scenario changed, lock context, and optional later rate and fees. These are unverified reports, not automatic public penalties. Later pricing must be compared with matching contemporaneous originator evidence. Treasury and MBS movements cannot be mechanically translated into an allowed mortgage-rate change.

The automation plan includes scheduled freshness checks, dated source ingestion, evidence validation, score recomputation, and audit logs. Disputes, identity mismatches, and ambiguous pricing changes require review. No automated score or review-harvesting job is active in this preview.

Updates and corrections

Material methodology changes will receive a new version and explanation on this page. Version 1.2 adds multiple recipients, required fee worksheets, private borrower quality-check documents, assistance requests, and profile photos. Version 1.1 replaces the fictional lender table with the private request pilot and adds profile approval, recent-review rules, and outcome collection. Version 1.0 documented the educational comparison, market pipeline, widgets, and proposed scoring framework. If a source revises an observation, we use its updated published value when retrieved. Our independent lender audit and appeal program has not launched.

For a benchmark discrepancy, compare the linked source and reporting date first. For an actual loan discrepancy, use the lender’s official disclosures; our sample comparison is not a quote. See our editorial standards and advertising policy.