The Mortgage Data Arms Race

Model Match was born out of the wreckage of the Great Financial Crisis, when the industry was bleeding talent & lenders were scrambling to rebuild teams. 

But the first version of Model Match wasn’t a data product at all. It was a recruiting CRM for mortgage lenders. Only after the SAFE Act made data far more accessible did the vendor layer in loan-level data to help lenders answer a bigger question: 'Who should we be having conversations with?'

Today, lenders demand far more than just accurate data from vendors like Model Match & its many competitors. In today’s special edition featuring Vendor Wars, we break down the nuances of the space, what makes Model Match unique, & what’s next.

Also in today’s special weekend edition: a new mortgage tech firm named after “The Black Dread” in Game of Thrones🐉 , how CrossCountry Mortgage won the bidding war for Two Harbors, Pennymac sued by a CRM shop & more.

(🙏 If you like what you’re reading, tell a fellow mortgage junkie to sign up here.)

"Model Match was founded in 2014. We were born in this industry," Thomas Seelbinder, Model Match’s head of product, told me in a recent interview. 

The company traces back to Hammer House, a mortgage recruiting firm that Drew Waterhouse & Eric Levin launched out of the ‘08 crisis. "They wanted to start a software company using the methodologies that they were using to help their lender partners to recruit," Seelbinder said. The first version of Model Match was essentially a purpose-built CRM for mortgage recruiting, a gap he says no one else was filling.

The data piece came around ‘17. "I believe we were No. 2 to that space. MMI was number one," he said. Today, RETR, iEmergent, Polygon Research, Modex, InGenius are all vying for a piece of the pie, pitching lenders on clean data that can be operationalized to win originators, capture more refis & more.

The data companies are typically collecting data from the same public sources, so you won’t find Seelinbder making bold claims that Model Match’s data is “the best” or is more accurate than others. "I learn something new about this data almost every single day,” he said. “I've been looking at it constantly since 2017 — it only expands. It only gets deeper. Sometimes it only gets messier."

He said the various data vendors have all carved out their own niche. Some are SaaS-focused, others are geared toward the LO & some have a capital markets focus. Others still offer a services/big-report-type model.

But importantly, the data is most useful if you can do something with it w/o leaving the tool, Seelbinder said. "We didn't come into this space thinking, 'Hey, we're going to create this data platform & go find users that want to sign up & look at numbers on our screen,'" he said. "What makes Model Match unique is that before the data, we have had those CRM-type of features or abilities built into the platform. So not only visualizing the data, but being able to take action on the data within a single environment — reducing the amount of tech bloat required by our partners."

Model Match, he noted, runs a proprietary database structure built in-house, versus "other firms maybe using something that's kind of off the shelf to connect all these pieces together."

The old stat is that only 20% of LOs use any of the tech tools their lenders give them. So I asked Seelbinder what the power users are doing differently. His answer is the spider web. 

"It's not just 'here's an agent, here's how much volume they've done, and these are the other originators they work with.' That's the surface." The real leverage, he said, is in the layers underneath: loan-type patterns, borrower-side data, & the relationships that connect them.

The LOs connecting all the dots & using the data to reach new referral sources & clients are "just full steam ahead. It's quite unbelievable."

Running a mortgage data platform is not cheap, & Seelbinder is candid about why the field isn't more crowded than it already is. "We just breached six billion records total in our database. That's a massive database… There have been others in the past that have attempted and have come & gone, because it is so incredibly expensive."

AI isn't bailing anyone out, either. "I don't think it's getting any cheaper. It only gets more expensive as you go & as you get bigger."

I asked if there’s enough business to go around or if will we soon see a culling as expenses mount. He says yes. "We're sitting at what, 200K loan officers — and that's just the loan officer space." Layer in capital markets, secondary, executive leadership use cases, plus adjacent verticals like title, MI, real estate, & the TAM is far bigger than the recruiting-only framing suggests.

As for selling to them, Seelbinder told me Model Match is unique in that it doesn’t have a sales team (it’s a bit Valon-like in that respect) b/c they’re focused on investing fully back in the product. On that front, Model Match has been active. The headline news from Seelbinder is a six-month, foundation-level rebuild that Model Match just began rolling out.

"We were going to appear to look slow. We were going to take a step back," he said. The team acquired a new mortgage-activity data set (pre-close all the way through originated), restructured the database to handle it, & rebuilt the platform itself. "We went all the way down beyond the studs. We maybe took down the studs & went down to the foundation." Phase one is focused on the data layer & a new market signals report, launched just before our interview.

As for what's coming next, Seelbinder wouldn't give away the roadmap, but he didn't hide his conviction: "The future of Model Match is one that is 100% pure innovation, & something that I don't think the data space in this industry, has seen — but certainly will soon."

Flávia Furlan Nunes has an updated account of what went wrong for UWM in its pursuit of Two Harbors. Per Nunes, UWM revised its original proposal on two occasions to save the deal as the stock price descended. UWM’s first revised proposal guaranteed TWO shareholders up to $10.71 a share, w/ a cap at $2 per share. Still, The TWO board leaned toward accepting the CCM offer. At that point, a third undisclosed bidder, “Company A,” proposed acquiring TWO for $10.75 per share in cash or a stock-for-stock reverse merger, where Company A would merge into TWO, giving TWO stockholders a stake of roughly 16.1% in the combined company. 

CCM then raised its offer to $10.80. Company A countered w/ a cash election option allowing up to 25% of TWO’s common shares to be cashed out at $11.09 per share. “But Company A lacked draft deal agreements, required significant due diligence, faced a lengthy path to closing and failed to provide enough financial information for TWO to properly value the bid, the seller said,” Nunes reported. “Meanwhile, UWM submitted a second revised bid, raising its offer to provide a cash-equivalent value of $10.95 per share through a mix of stock and cash. This version removed the cap on total cash consideration.”

UWM accused TWO of breaching merger agreement obligations & warned it could pursue a hostile bid or sue. The wholesale lender sent TWO a document preservation notice, signaling potential legal claims and requiring records related to the dispute to be retained, Nunes reported.

A small CRM vendor is suing Pennymac for allegedly stealing its technology, saying it is reminiscent of a trade secrets case Pennymac had w/ Black Knight, in which the wholesaler paid $155M.

Surge, which claims to work w/ large wholesale lenders, sued Pennymac for breach of contract last week in Michigan, NMN reported. Surge wants over $176K it says Pennymac hasn't paid since ‘23, for Surge's due diligence & sandbox services. 

"Pennymac's theft is no accident but a deliberate attempt to convert Surge's software, terminate its relationship with Surge, and continue to use that software without reimbursing Surge," wrote attorneys for the company. The vendor asked the court for an injunction to stop Pennymac from using its data, & for unspecified damages to be tripled upon a ruling.

A mortgage tech startup in the AI underwriting space just raised $6M from Kleiner Perkins, w/ participation from Formation & BoxGroup. Balerion AI, founded by Naren Krishna & I think named after the Targaryen’s biggest dragon in “Game of Thrones,” says it is already in use from lenders, including FM Home Loans, run by David Brecher.

Its flagship product (only product?), Balerion Loan Intelligence, is an “agentic reasoning engine” embedded directly into the origination workflow. It automates the full loan manufacturing lifecycle & ensures full compliance w/ Fannie Mae, Freddie Mac, FHA, & custom non-QM overlays.”

This “AI pre-underwriting” space is rapidly becoming very competitive within mortgage tech. Friday Harbor, Candor, Orcolus & others are present, but the vast majority of lenders still haven’t adopted the tech. I think we’ll see way more ‘pre-underwriting’ entrants in ‘26 & ‘27.

What have you seen in your market? Email me at [email protected].

(🙏 If you like what you’re reading, tell a fellow mortgage junkie to sign up here.)