The DIY LOS Era Is Here

In today’s edition of The Mortgage Scoop: UWM & Rocket are preparing to spend years fighting over the alleged bounty on Mr. Cooper’s servicing book.

Plus, some lenders are trying to build their own LOSs, having presumably decided that building a POS wasn’t painful enough 🤣 . And two engineers behind Friday Harbor’s underwriting AI became borrowers themselves, only to discover just how messy even a “simple” mortgage file can get.

UWM previewed its legal defense in the $100M lawsuit over whether it illegally tried to win back loans in Mr. Cooper’s book.

There’s a lot of legalese here & I don’t want to bore you, so let’s just touch on a couple of items.

Per NMN’s Andrew Martinez, attorneys for UWM noted that Mr. Cooper did not sue UWM for its conduct before it was acquired by Rocket. Attorneys also disputed whether UWM’s ad campaigns were generalized or took aim at the Mr. Cooper loans in question.

For example, while UWM claims Refi75 was advertised through general press releases, Rocket says UWM made no effort to exclude ✂ Mr. Cooper loans. Rocket also says UWM didn’t try to exclude Mr. Cooper mortgages from KEEP. UWM says the complaint doesn’t spit facts. Finally, UWM says Rocket didn’t ID any affected loans under Refi Shield 100. 

This one is going to drag on —> Rocket says it is expecting a ‘28 trial ⏱️, per NMN.

I've covered housing & mortgage for more than a decade. (Some people even refer to me as “Mortgage Schefter,” which is both funny & disconcerting.) Yet I can confidently tell you, dear reader, that doing this type of work does not prepare oneself for the vagaries of full doc lending. I've probably learned more about lending by trying to secure an NYC co-op loan for myself over the past 2 weeks than I have from 10+ years of covering the biz.

It is astonishing how many edge cases there are in mortgage. Building efficient workflows & good technology for mortgage is hugely challenging simply b/c there are so many variables.

I'm not the only one who’s had to live it to get it. Two of the architects behind AI underwriting shop Friday Harbor recently ‘dogfooded’ 🐶 their own product by becoming borrowers. Founding engineer Yannis Katsaros & lead software architect Paddy Quinn both worked w/ Jim Carroll, head of home lending at Partners Bank, on home purchases earlier this year. What they got was a firsthand look at what the tech does well, where it falls a bit short & why the human in the loop isn’t going anywhere.

Katsaros went in w/ what should've been a relatively easy conforming loan file. Except he also had multiple W-2s, a pending home sale, & discrepancies across his down payment, his bank accounts & how the property was being treated. Friday Harbor kicked out 15 conditions, which sounds like…a lot?

Katsaros’s first reaction was that the system had to be wrong. But every flag turned out to be legit. The difference-maker was timing: b/c Carroll saw all of it in week one, they cleared it well before closing instead of discovering problems in the 11th hour.

Quinn's file was the hairy one. He had unseasoned bitcoin, restricted stock, a planned rental conversion & a possible jumbo.

"I knew going in that my situation wasn't straightforward,” Quinn said. “That was kind of the point. If the system could handle my file, I'd know something real about what we'd built." 

The platform surfaced the seasoning & documentation requirements fast & then surfaced something v useful: Partners didn't have the right product for him. Carroll referred him out to a broker & the loan closed elsewhere. Nobody wasted like, three weeks finding out they couldn’t do it in-house. With the doc review, the manual calcs & the guideline checks handled, Carroll's time instead went to advising & strategizing.

Being borrowers also exposed their own gaps. Quinn came away wanting broader non-agency guideline support. Katsaros wanted the system to do a better job explaining why it flagged what it flagged (which, having received a 15-condition list myself in spirit, I'd co-sign).

Around this time last year, sources across the industry told me that over a dozen lenders were taking a shot at building their POSs. Why spend the money on Blend, nCino, Floify, Consumer Connect, BigPOS, BeSmartee, etc. if you can do it yourself? (The answer: It’s way harder than it looks haha.) 

The new thing, according to multiple sources on the tech side, is lenders attempting to build their own LOSs. One source said the list runs from top-25 lenders to mortgage shops doing 200 loans a month. “A good chunk of them are hoping to sell it to other lenders once they have it built,” the source said. 

More to come in a future edition of The Scoop…

The share of mortgages in active foreclosure reached 0.53% in June, the highest level since ‘20 years, per ICE. Foreclosures are up 16% y-o-y, but still historically low. Foreclosure sales in June were 46% below pre-pandemic levels. The DQ rate also rose slightly, increasing 5 bps on a month-over-month basis in June. (It’s still 60 bps below the June ‘19 pre-pandemic benchmark.)

A few weeks ago my wife & I made an offer on a two-bedroom co-op in Ditmas Park, Brooklyn. The sellers selected another offer. So we are back on the hunt! Over the weekend, we checked out: 

800 Cortelyou is huge (1,300 sf) & has multiple exposures, but a very similar unit in the building sold for $900K in January & was newly renovated. This unit needs a lot of work & is $50K more expensive. Is it a good idea to spend $100K to make it the nicest unit in a so-so building? It’s also not in the school district we prefer, so that’s out. My wife found 355 Stratford to be too small, even though I love the location. The condo at 11 Terrace Place is sunny, has a great inside-outside thing going & benefits from the 421a tax abatement (monthlies are like $400). But the living space is very small & it’s not in a prime location. Can we make 893 sf work w/ two kids when we both WFH often? I don’t think we can. We’re tabling it for now. Here’s what we are seeing this week:

We’re branching out by looking at a few multifamilies (809 Friel, 1268 Prospect) & potentially renting out one unit & living in the other. I’m most drawn to 391 3rd St. & 310 Windsor, but the former might be too small & the latter might be too expensive & arduous a reno process.

Email me w/ what you think is best. Reminder: We have two kids (4, 2) & my wife & I both work from home, so a decent amount of space is important!

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