Mortgage Pricing Engine Software

Buy for the loan that breaks the engine

Pick the pricing engine that gets your hardest loan right, and weigh everything else after that. A rate sheet that misses one adjustment on a jumbo cash-out costs real basis points. The loan still closes. The margin does not survive. That is why depth of function carries the heaviest weight we assign in this category. An engine that is pleasant and wrong is worse than one that is clunky and right.

Production impact comes next, and for a good reason. Every lock in the shop passes through this system. Slow scenario runs and hand-keyed locks tax the desk all day, every day. Adoption and friendliness get the lightest weight we use anywhere on the site. A secondary desk will learn a hard tool. It will not forgive a bad price.

Who maintains the rate sheets is the real fork

Two models exist here, and they lead to very different lives. In the vendor-maintained model, the provider ingests investor sheets and codes the adjustments for you. Optimal Blue states it imports more than 45,000 investor rate sheets daily. Its PPE page claims pricing accuracy of 99.5 percent across 150-plus investors. Those are the vendor's own published figures. They set the bar every competitor gets measured against.

In the configured model, you own the rules. LoanPASS gives lenders direct control over pricing structures, margins, eligibility criteria, conditions, and exceptions. The company says most customers go live in 30 to 60 days. Lender Price publishes a catalog of more than 130,000 loan products. It builds granular rules using natural language input rather than raw code.

The trade is simple to state and hard to live with. Vendor-maintained content means you wait on someone else for a mid-day reprice. It also means waiting when you add an investor. Configured content means the errors belong to you. Ask each vendor who fixes a wrong adjustment at two in the afternoon. Ask how long that fix takes to reach production.

Rule sprawl is the hidden cost of the configured model. A lender with six investors and four channels ends up maintaining the same margin logic many times over. Optimal Blue shipped a rules release in May 2025 aimed at exactly this. A single rule now applies across multiple investor relationships instead of being cloned. Ask any vendor to show you the same change applied across every investor at once.

Lock desk mechanics decide your Tuesday

Pricing is half this purchase. The lock desk is the other half, and buyers underweight it. Polly builds lock workflow into the engine itself. That covers locks, extensions, re-locks, re-prices, price exceptions, and float downs. Polly also shipped a post-lock product change function back in 2022. Swapping product type and term after lock no longer means rebuilding the loan by hand.

Delivery to the investor is the other half of the lock. In July 2025, Optimal Blue automated best efforts locks directly with participating investors over API. The company says the manual version averaged about 15 minutes per loan. It offers the automation at no added cost and says setup is minimal. Before that, a lock desk user logged into an investor portal and rekeyed the loan.

Push a real loan through the ugly path during the demo. Run a product change, then an extension past expiration, then a re-lock, then a price exception. Watch what price comes back at each step. Check whether worst-case pricing fires the way your policy is written. Then read the audit trail and see if a repurchase reviewer would accept it.

Integration claims you can actually check

"We integrate with the agencies" is a claim with a verifiable answer. Fannie Mae's Loan Pricing API returns asset price, servicing released premiums, loan-level price adjustments, and further Fannie adjustments together. The API issues a quote ID with it. That quote ID feeds the Loan Committing API for loan-level best efforts commitments. Ask whether your engine holds and uses that quote ID, or only displays a number on screen.

Freddie Mac gates vendor access, and that gate has a schedule. A vendor needs an executed Form 478 Secondary Market Advisor Selling Agent Agreement or an approved technology vendor agreement. Freddie describes an eight-step onboarding with a test environment and formal approval before production. Nobody flips that switch during your implementation week.

On the LOS side, the shape of the integration matters more than the logo. MeridianLink runs PriceMyLoan as a native engine inside its own origination system. That removes a hop, and it also removes your choice. ICE PPE publishes endpoints for programs and rates, eligible rates, rate selection, adjustments, guidelines, and custom fields. Polly says its APIs map standard and custom fields into any LOS. Ask which of your custom fields survive the round trip back into the loan file.

Non-QM is where engines quietly fail

The common failure is not a missing product. It is a guideline the engine cannot express. The tool returns a clean price on a loan no investor will buy. You find out at purchase review, weeks later, at your expense.

LoanPASS aims straight at this problem. Its rules handle bank statement income, asset depletion, DSCR ratios, and rent schedules. It supports layered adjustments, channel-specific margins, investor-specific rules, and product-level incentives. Conditions and underwriting outcomes generate in real time. Loans falling outside the defined rules get flagged for manual review instead of quietly pricing. Lender Price sells FLEX for non-agency lenders. It allows custom fields and an interface configured by channel, product, pricing tier, and branch.

Optimal Blue approached the same market from the investor side. It gives investor users a way to benchmark non-QM rate sheet pricing against other investors in real time. That helps the desk setting the sheet more than the desk consuming it. If you originate non-QM, test the engine on a loan with two compensating factors and one exception.

Eligibility deserves its own look during the demo. A good engine tells you why a loan failed, in words a processor understands. A weak one returns an empty product list and no reason at all. ICE PPE separates eligible rates from the full program list in its published endpoints. Ask any vendor to show the ineligibility reason text on a real declined scenario. That text is what your originators will read fifty times a week.

Price the whole loan, not the note rate

Mortgage insurance belongs inside the comparison, and many engines treat it as a side trip. LoanPASS bought PMI Rate Pro to close that gap. It quotes Arch MI, National MI, MGIC, Essent, Radian, and Enact. Coverage spans the five standard MI products, sorted by best execution or by a risk allocation you define. Monthly, single, split, and lender-paid structures each shift the ranking. An engine that ranks products without MI is ranking on partial math.

Government product carries its own traps

FHA and VA pricing is not only investor pricing. FHA mortgage insurance premiums come from HUD, not from the investor sheet. The VA funding fee changes with the down payment and with prior use of the benefit. Exempt borrowers pay none of it. High balance limits move by county every year.

Put a VA IRRRL and a funding-fee-exempt purchase into every demo. Add an FHA purchase sitting right at the county high balance limit. Then add a VA loan with a second-time user and minimal down payment. If the engine prices the note correctly but botches the fee, your comparison screen lies to the originator.

Brokers are buying a different product

A broker-owner does not hedge a pipeline. Breadth of wholesale coverage and speed to a quote are close to the whole game. Loansifter, an Optimal Blue product, states coverage of more than 120 wholesale investors. Its page reports 800,000 monthly searches across more than 6,000 users. Lender Price runs a broker marketplace and reports over $300 billion in locked volume across its customer base. Mortech, owned by Zillow, sells investor pricing and product eligibility data, plus historical pricing and LOS connections.

Pay attention to what the broker tools stop short of. Lock submission back to the wholesaler is not always part of the package. The Loansifter page describes comparison and workflow connection without spelling out lock submission. Get that confirmed in writing before you assume the round trip exists. A broker shop that still rekeys locks into four wholesale portals has not bought much.

Know who owns your vendor

Ownership shapes roadmaps here more than in most categories. ICE completed the sale of Optimal Blue and the Empower LOS to Constellation Software's Perseus Group in September 2023. The FTC required that divestiture as a condition of the Black Knight acquisition. ICE kept its own pricing engine and still sells it. So a lender running Encompass with Optimal Blue buys from two firms that compete directly.

That is not a reason to avoid either one. It is a reason to put integration commitments in the contract. Roadmap goodwill between competitors gets thin when volume drops.

Originator-facing tools are worth something, just not much

Side-by-side scenario comparison has become standard equipment. Optimal Blue added an originator tool in October 2024 that generates side-by-side loan comparisons automatically. Polly delivers pricing on any device for scenarios run away from the desk. These features move retail conversion, and a good sales team will demo them first.

Hold the line anyway. A prettier comparison screen showing a wrong price is a faster path to a bad loan. Score these tools after you have satisfied yourself on rate sheet accuracy and rule depth. The order matters more than the score.

How to run the bake-off

Pull ten real loans from the last ninety days. Include your two worst pricing exceptions and one government loan. Hand every vendor the same file and the same investor set. Price each loan yourself first, by hand, straight from the raw sheets. Then compare line by line, not on the bottom-line rate.

Refuse a demo run entirely by a sales engineer. Ask for the person who writes rules and watch them build one live. Have them change a margin by channel while you time it on your phone. If that change needs a support ticket, price the friction into the total cost.

Then ask the question vendors dislike. What happens when the engine prices wrong and the loan is already locked? Who eats the difference, and at what threshold? Get that answer written into the agreement, not spoken in the demo. Every other question in this category is smaller than that one.