268 data points, across eight underwriting domains.

Open the explorer to see what sits in each domain, how much of the file it carries, and where the published taxonomy stops and the proprietary model begins.

268Data points
8Domains
45Largest domain
0Points inferred
Why 268
The input layer

A mortgage file is not a rate shopping exercise. It is a data problem.

Most declines and most overpayments trace back to something that was already sitting in the application on day one — unread, or read too late.

Read once, scored many ways

The same file is scored under each lender’s own income calculation, qualifying rate and property policy. One borrower produces dozens of different qualifying pictures depending on who is looking at it.

Missing data is a first-class result

A blank point is recorded as blank, never inferred. The engine returns a ranked list of exactly which missing points would change the outcome, so an agent chases three documents instead of fourteen.

Every derived figure carries a confidence flag

Income reconstructed from T1 lines, rental offsets and add-backs is marked as derived, with the source it came from, so nothing enters a submission that the agent has not been able to see the basis for.

The taxonomy is published; the weighting is not

You can tell a client exactly what informed their recommendation. You cannot hand a competitor the model — and neither can we.

The inputs
What the engine reads

268 data points, across eight underwriting domains.

We publish the domains and how many points sit in each, because you should be able to tell a client what informed their recommendation. What each point is worth, and how they interact, is the part we don't publish.

268Total data points

See the engine read one of your files.

Bring a real deal to the demo and we will show you every point it populated and every one it could not.

Book a 20-minute demo