Two engines, one solution

Eligibility first.
Then true best execution.

Every loan runs the same path: decide what qualifies, then price what passed and rank it. Eligibility and price are kept strictly separate, the flow never reverses, and every step is recorded so the result can be defended field by field.

Program Engine
Does it qualify?

Narrows from every program the lender subscribes to down to the exact program-and-product combinations the borrower is eligible for. It decides eligibility only — it never touches price.

Pricing Engine
At what price?

Takes only what already qualified, builds each option's final rate and price, and ranks the results by best execution. It prices what passed — it never re-decides eligibility.

This clean separation means each engine has a single responsibility, the eligibility decision is fully auditable on its own, and the same priced results can be re-ranked without re-running guideline logic.

A loan's journey

Six stages, from scenario to defensible answer

1

Submit

A guided form where an officer builds one valid pricing scenario and submits it with a single click.

Originator, loan category, property, and borrower profile are entered across validated tabs. Cascading dropdowns ensure only programs the client is subscribed to appear, and fields auto-calculate and auto-geocode — so a user can only build a request that's valid to price. One Submit posts the scenario to the Program Engine.

Cascade
Transaction → Channel → Program Type → Pricing Type, each filtering the next
Geocoding
Address auto-resolves census tract, county FIPS, AMI, and DTS eligibility
Next
A single Submit runs eligibility →
2

Eligibility — Program Engine

A pipeline that narrows every subscribed program to the exact program-and-product combinations the loan is eligible for.

Built as a chain of responsibility, each stage hands an accumulating scenario to the next: setup and enrich, categorical match, numeric match, product match, overlay application, and outcome. Overlays can only tighten a program's published guideline, never loosen it. The output is program×product pairs — not a flat program list — and the engine decides eligibility only.

Architecture
PreGuideline → Header → Finance → Product → Overlay → PostGuideline
Output
Program × product pairs that qualified
Principle
Overlays only tighten the published guideline
3

Best-Ex — Pricing Engine

Takes each eligible pair, builds its final rate and price, and ranks the results by best execution.

A named-stage flow looks up base pricing, applies loan-level and servicing adjustments, applies post-adjustment ratesheet overlays, resolves the lock window, then ranks every result so the strongest options surface first. It prices only what already qualified, and runs full-matrix for interactive quoting or single-rate for high-volume batches.

Stages
Base pricing → LLPA/SRP adjustments → ratesheet overlays → lock resolution → BestEx ranking
Run modes
Full-matrix (interactive) vs. single-rate (batch, note rate only)
Scale
The base rate/price grid is the platform's largest dataset; the layer evaluates only records that can apply
4

Results — True BestEx

The payoff screen: a rate-by-product grid showing the best available price across every qualifying program.

Rates run down the side and products across the top; each cell holds the best price any qualifying program offers at that rate-and-product. Click any cell to see its full breakdown, compare alternative programs at the same rate, change the lock window to re-price, and request a rate lock that snapshots the selected price. A display-only price-format toggle never changes the computed number.

Grid
Rates (high → low) × products (Fixed / ARM / Interest-Only), best price per cell
Lock
A rate lock captures a pricing snapshot; Combo loans render two paired grids
Display
Price-format toggle is display-only — never changes the price
5

Counter-Offer — best achievable terms

A non-fit is a starting point, not a stop. Instead of a financing failure, the engine automatically returns the best achievable terms — ranked.

When a scenario doesn't qualify as entered, the engine relaxes one constraint and re-runs once to surface the best terms actually available — turning a dead-end decline into a real counter-offer the officer can take back to the borrower. (If the officer left a field blank to shop a maximum, this is that requested exploration instead.) Options arrive ranked best-first; the officer dials in the borrower's actual values, which recompute the tiers live, then continues to pricing.

Two flavors
Requested (officer-shopped maximum) vs. CounterOffer (engine-relaxed alternative)
Bounded
Single-shot — the engine relaxes once, never recurses
Dynamic
Only the fields the search optimized are shown
6

Audit — the why

Every program listed Pass or Fail, with a field-by-field reconstruction of exactly why.

Because the engine evaluates dozens of programs against thousands of guideline and overlay rules, the Loan Audit is how a user finds the cause of any unexpected result. Clicking a program replays the eligibility pipeline in a no-write audit mode off the loan's saved execution log, reporting the first field that disqualified it and whether the limit came from a guideline or an overlay.

Replay
Re-executes the Program Engine in audit mode (no DB writes) from the saved log
Panels
Pass / Fail / per-program Audit Details, plus a Non-QM credit pre-screen
Reports
The first failing field and its source (guideline vs. overlay)
Search modes

Let the engine solve for the borrower's maximum.

The feature officers reach for every day. Leave a financial field blank and check Max or Min — instead of pricing a number you typed, the engine searches for the number itself: the largest loan amount, the highest LTV, or the most cash-out the borrower actually qualifies for. Shop two dimensions at once, and the counter-offer stage returns the answer ranked best-first.

It turns "what will this borrower qualify for?" from a guessing game into a single priced run.

  • Leave a field blank, check Max or Min — the engine solves for the value
  • Shop up to two dimensions in a single run
  • Results arrive ranked best-first, recomputing live as you refine
The modes

Shop the loan from every angle

One scenario, many questions answered — each mode runs the full Program Engine, so the answer is always what actually qualifies.

📑

Multi-product

Price several products side by side for a direct best-execution comparison — Fixed, ARM, and Interest-Only weighed at a glance.

📈

Max Financing

Solve for every qualifying LTV / CLTV combination — the most leverage the borrower and the programs actually allow.

💰

Max Cash-Back

Solve for the most cash-out the borrower qualifies for — no hand-iterating the loan amount to find the ceiling.

📊

Min FICO

Find the lowest credit score that still qualifies — know exactly how much score headroom a scenario has before it breaks.

🔗

1st / 2nd Combos

Price a first and second lien together as one structured scenario — the combo rendered as paired best-ex grids.

🔍

Min / Max any field

The general form: leave any financial field blank, check Min or Max, and let the engine solve for the number — up to two at once.

🔍

"Is it the code, or the data?" — answered.

The single hardest question about owning a pricing engine. For any loan, even one priced months ago, the engine replays its evaluation field by field off a saved execution log and attributes every outcome to a specific cause: a published guideline, an overlay rule (with the exact condition that fired), or the input data. "The loan did something unexpected" becomes "this field failed because of this rule" — a defensible, repeatable answer instead of a guess.

Price a real scenario through it.

Qualified parties get private, hands-on access to the live engine — to price a loan and interrogate the result field by field.