A near three-year low in home sales is the news a lender paid for

In the four weeks to September 6th a little more than 309,000 American homes went under contract, the fewest in nearly three years on a seasonally adjusted basis. Yet the median sale price rose 2.2% year on year, to roughly $398,600. Weak volume, firm prices: the signature of a market in which supply and demand have not met but evaporated. The oddity is who is delivering the bad news. Redfin, the most-visited real-estate brokerage website in America, no longer trades as a stock. In July 2025 Rocket, the country's largest mortgage lender, folded it into the group's accounts in exchange for $1.75bn in shares. The report of collapsing home buying is therefore not an industry bulletin from a disinterested observer. It is the internal market intelligence of the company that paid for Redfin precisely to feed on the home-purchase market that is now at or near a three-year low.

The mechanism producing this stuck market is the famous lock-in effect, and it binds on both sides. Homeowners who borrowed at 2.7% in 2020-21, when prices were rising far faster than rents, will not sell if doing so means financing a replacement at a current rate above 6%. So sellers stay put, and the listings that do appear crowd into fewer, costlier transactions. This autumn the brief thaw several forecasters promised has not come: the weekly average 30-year fixed rate was 6.71% in early September and the daily average touched 6.97% days later, pushing the typical homebuyer's monthly payment to a 14-month high of about $2,641. Buyers, meanwhile, have gained negotiating power that they largely cannot spend. New listings are up, active inventory has grown and months of supply has inched toward the four-to-five range that marks a balanced market; a fifth of listings now carry price cuts. But a discount on an unaffordable home is still an unaffordable home.

Now take the parent. Rocket did not buy Redfin for brokerage commissions, a slender business dependent on transaction volume. It bought roughly 50m monthly visitors and a network of more than 2,200 agents to drive the one thing mortgage lenders need most and have been losing for years: purchase originations, the loans written when someone actually buys a house rather than refinances one they already own. The merger documents promised more than $60m in revenue synergies from pointing home-searchers at Rocket's mortgages, title and servicing, and said the deal would add to adjusted earnings per share by the end of 2026. The entire bet rests on turnover — on the number of homes that change hands each quarter.

That is the number refusing to cooperate. Rocket's own chief financial officer conceded in August that "the expected housing recovery in 2026 has not materialized," and that the group's real-time data had shown a tougher market than industry forecasts implied. Craning for comfort, the company counts its execution: record purchase and refinance market share in the second quarter, and clear progress on the integration — 47% of Redfin's buy-side clients now finance with Rocket, nearing a 50% target, and mortgage leads from the website doubled year on year in June. This is the right defence of a bad hand. Executing well and growing share inside a market that is itself contracting is worth something, but it converts a falling market into a steadily more modest slice of it. A purchase business cannot recapture what the market refuses to produce.

The market's verdict reflects the tension. Rocket shares trade around $14, capitalising the company at roughly $37bn and carrying a three-year return near 50%; independent models put their fair value close to, or slightly below, the market price. Investors have paid up for a story — the homeownership platform that turns a national search habit into a self-feeding purchase machine — that the housing data keep failing to reward. That is the uncomfortable place this stock now stands: not a broken company, but a well-integrated one whose central thesis is hostage to a variable its own subsidiary reports each week.

The lever that would change the judgment is a fall in mortgage rates large enough to unlock the pent-up supply rather than merely to shuffle it. A move toward, say, 5.5% would persuade legions of locked-in owners to list, widen competition for buyers' scarce dollars and finally put volume back into the purchase pipeline that Rocket acquired a search engine to fill. Until the numbers show that — and Redfin is the most direct place to watch it — the messenger of the three-year low is also its proof. Rocket bought Redfin to capture the housing market's recovery. For now Redfin keeps reporting, week after week, that there is less to capture.