What Zillow Can’t Show You lays out the core structural argument: two data gaps — non-agency RMBS delinquency and private-listing inventory — are widest in the same Sunbelt markets where prime-cohort stress is building. This companion focuses on what that implies for portfolios.
The thesis
The housing market’s two most-cited data sources — the MBA delinquency rate and MLS-reported inventory — are both structurally incomplete, and both gaps are widest in the same markets: the Sunbelt metros where remote-work migration drove peak-cycle price appreciation. Here’s what that implies if you’re watching markets.
What the thesis suggests watching
Before naming sectors, the mechanism deserves a sentence. This stress event, if it materializes, won’t look like 2008 — and that matters more than it sounds. The institutional infrastructure built to catch housing stress was designed after 2008 for 2008-style problems: subprime borrowers, adjustable-rate mortgages, synthetic leverage distributed through the global financial system. None of those conditions apply now. What’s moving instead is the prime cohort: fixed-rate borrowers with excellent credit who bought at peak prices in markets that have since seen employment reverse. The policy tools and data systems calibrated for the 2008 crisis will take longer to recognize stress that originates here. When the playbook doesn’t match the problem, markets price it late.
Fitch’s quarterly RMBS Performance Monitor is the most direct public instrument for tracking whether the non-agency universe is confirming or contradicting the thesis. The MBA’s delinquency survey covers the agency universe — Fannie, Freddie, Ginnie. Fitch’s monitor covers non-agency RMBS 2.0, which now carries $1.7 trillion in outstanding securities and is on pace for record 2026 issuance. The Prime Jumbo 30-day delinquency rate in that monitor currently stands at 1.09 percent, up 22 basis points year-over-year. That number is publicly available, released quarterly, and tracks exactly the borrower cohort — prime, jumbo, fixed-rate — that the main piece identifies as the leading stress signal. Watch whether it continues climbing toward 1.5 percent or reverses. The direction of that line is the thesis’s most direct market confirmation.
Homebuilder incentive ratios in Sunbelt-heavy reporters are a proxy for two things simultaneously: underlying demand softness and the shadow inventory risk that the Zillow litigation could force into the open. D.R. Horton (NYSE: DHI) and Lennar (NYSE: LEN) have the deepest Sunbelt exposure of any publicly traded builders. Both entered 2026 with incentives running near 5 percent of order value — near the high end of the non-recessionary historical range — and both report order volumes, cancellation rates, and gross margins quarterly. The Compass/MRED litigation matters here directly: if a court ruling restores private listings to public IDX feeds, the competitive supply picture for builders in those markets changes without any macro shift. Watch cancellation rates specifically — they tend to move before orders, and they would be the first signal of a buyer confronting supply they didn’t know existed.
Title insurer order volumes are the most granular real-time read on actual housing transaction activity, and they resolve an ambiguity the thesis doesn’t. First American Financial (NYSE: FAF) reported Q1 2026 title revenue up 21 percent, driven by a 9 percent increase in direct orders closed. Title revenue goes up when closings increase. In a scenario where the Zillow litigation unlocks shadow inventory and distressed sellers start transacting, that’s more closings — which is positive for title, not negative. But if private-network listings sell without ever entering the public MLS, they may close without generating a title insurance event through the standard agency channel. Watch FAF’s direct order count quarterly. If it stays flat or falls while overall market commentary describes rising supply, that divergence would suggest transactions are clearing through channels that bypass the standard title pipeline.
The bull case
If both gaps begin closing — the Fitch Prime Jumbo delinquency trend stabilizes and the Zillow litigation restores listing transparency — the market gets a more accurate picture of housing stress simultaneously in both data dimensions. Stress that gets priced early and accurately produces orderly adjustment, not crisis. Builders reprice land, non-agency RMBS investors demand spreads that reflect the actual delinquency picture, and the market finds a clearing price without a forced-liquidation event.
In that scenario, the structural winners are companies with the least commitment to peak-cycle assumptions. NVR (NYSE: NVR) controls lots through options contracts rather than owned land. If market conditions deteriorate in a Sunbelt market, NVR walks away from an option at a small cost. D.R. Horton and Lennar, holding owned land, face the full cost of being wrong. That model distinction isn’t new — it’s visible in every cycle. What’s new is that the geography where the distinction matters most is now also the geography where both data gaps are widest.
The bear case
Two things would break the market expression of this thesis. The first is that the data gaps turn out not to matter: non-agency RMBS delinquencies stabilize before migrating into 90-plus-day readings, Compass’s private listings clear quickly enough that the inventory shadow is temporary, and the Superprime consumer delinquency signal is a one-quarter anomaly driven by student loan reporting catch-up. If all three normalize by Q3 2026, the thesis has a direction correct but no durable catalyst.
The second risk is more structural: policymakers recognize the stress and reach for the 2008 toolkit anyway. Broad forbearance programs, GSE balance sheet expansion, renewed NPL purchase activity. Those interventions didn’t resolve 2008-style subprime stress — they delayed its recognition. Applied to prime-cohort employment-driven stress in Sunbelt markets, they would extend the period during which the non-agency data gap understates the problem. That’s the thesis’s core concern: not a crash, but a prolonged period of mispriced risk during which buyers in the most stressed markets keep making decisions based on incomplete information.
The specific signal to watch: Moody’s or Fitch issuing surveillance notices on 2022–2024 vintage Prime Jumbo RMBS pools, citing Superprime cohort delinquency trends specifically rather than aggregate non-QM performance. Rating agency surveillance actions typically precede spread movement and investor repositioning by two to four quarters. They are publicly announced and appear in both agencies’ press release archives. If those notices start appearing on prime — not non-QM, not non-prime — pools, the data gap the main piece describes has been formally acknowledged by the entities whose job it is to price it.
The tracked call
Tracked Call: The Prime Jumbo 30-day delinquency rate in Fitch’s quarterly U.S. RMBS Performance Monitor will exceed 1.5 percent by Q4 2026 — a 37 basis-point increase from its current level of 1.09 percent — as Sunbelt-concentrated prime cohort employment stress migrates from early-stage consumer delinquency into RMBS collateral performance.
Falsification window: Q4 2026 Fitch U.S. RMBS Performance Monitor, expected January 2027. Source is publicly available and released on a defined quarterly schedule.
Confidence: Low-Medium. The directional argument is supported by three converging signals — Fitch’s current upward trend, VantageScore’s Superprime acceleration, and Sunbelt employment deterioration. The timing is genuinely uncertain: policy intervention, rate relief, or a one-quarter statistical anomaly could delay the migration from consumer delinquency to RMBS performance data.
What would change my mind: If the Q3 2026 Fitch report shows the Prime Jumbo 30-day rate stabilizing at or below 1.10 percent — flat to slightly improved versus current — the upward trend has broken and the prime cohort stress signal is not migrating into RMBS collateral the way the thesis expects.
This is not investment advice. The thesis, the tracked call, and prior companions are at working-hypothesis.com.
If you found this useful, the best thing you can do is forward it to one person who would push back on it. I’d rather be wrong in public than right in private.