Built on the data you have.

Vintage takes loan-level files straight out of your core system, in whatever shape they come. There is no format to match and no template to fill in first.

Portfolio snapshots

The book as it stood on a date. Monthly, quarterly, or yearly.

Origination files

The terms each loan started with.

Transaction files

Charge-offs, recoveries, payoffs, and prepayments.

Bring any combination of the three; none is mandatory, and both CSV and Excel are read as they are. If you want a starting point, the upload guide helps you lay out an upload plan and can hand you a template for each file, though nothing requires you to use one.

Before anything uploads, a preflight in your browser strips personal information out of the files, so Vintage's servers never receive it. The security overview covers how that works.

Vintage recognizes many fields by name or a known synonym and suggests a mapping for each column it can. You review every suggestion and decide the rest, so a column never becomes a number you did not agree to.

You do this once

You tell Vintage which of your status codes and flags mean paid off or charged off. It remembers that vocabulary for your institution, so the next upload in the same format asks nothing at all. Every core encodes these differently, which is exactly why Vintage asks instead of assuming.

A bad match cannot slip through

When a column's values do not fit the field it was mapped to, Vintage stops and shows you, rather than coercing them quietly into shape and carrying the result into a curve.

When you join an upload to your portfolio, Vintage first shows a Review. It reads out, output by output and loan by loan, what can be modeled from the combined data today, by which method, and exactly which fields would widen coverage.

Partial data is accepted. An upload does not have to make every loan model-ready to be useful. The verdict reflects the data you have, not optimism about it, so you find out what your book can support before you have built anything on top of it.

Every figure is measured by historical cohort averaging: a life-table of per-age rates weighted by exposure, with no machine learning, no fitted curves, and no hidden parameters. A banker or an examiner can re-derive any number on the screen by hand.

Credit loss

Cumulative loss curves by loan age, built vintage by vintage from your own charge-offs, with a lifetime loss rate. Confidence bands over the observed range, and the projected tail marked apart from it.

Prepayment

Monthly SMM and annualized CPR from your own payoffs and prepayments, and a remaining-balance curve read against the contractual schedule.

Pricing

A required note rate built up line by line: funding at expected life from your own FTP curve, expected credit loss, operating cost, a credit for fees, and your ROA or ROE target.

Profitability

The same loan followed month by month: net income, the break-even month, lifetime profit, and NPV.

Slice the portfolio by any field you uploaded, from origination year to product to FICO, and every curve and price follows the slice. Loans a figure cannot use are counted and explained on the screen, never dropped in silence.

Every chart and table downloads as CSV, or as Excel with the chart image embedded, so a figure can go straight into a board packet, an ALCO deck, or a CECL workpaper.