Radian Bought Its Own Book: A New CEO, a $379 Million Buyback, and Two Risks the Market Hasn't Priced

Q3 was a strong quarter that moved the stock zero percent since July — because the real story is now capital return, a credit-scoring grid change, and an MI cancellation pilot.

RDN · Earnings Call · 2026-11-05

The debutante's real message

Mike Weinbach's first call as CEO was framed as an integration update: revenue more than doubled to $624 million, adjusted pretax operating income hit $254 million, adjusted EPS rose 25% to $1.44, and operating ROE reached 16%, up roughly 250 basis points year-over-year. Fine numbers. But the most consequential thing said on the call had nothing to do with insurance underwriting — it was an admission that management's own capital-return guidance was simply wrong, in the shareholders' favor. Last quarter, CFO Dan Kobell told analysts to expect full-year buybacks “more towards the upper end of that $200 million to $250 million range” — Dan Kobell, CFO · 2026-08-05. Radian has instead repurchased $379 million year-to-date — 8% of beginning shares outstanding — after $135 million in Q3 and another $117 million in October. Weinbach's explanation is blunt: “we didn't anticipate the opportunity to buy back shares at near book value and then even more below book value” — Michael Weinbach, CEO · 2026-11-05. Over $200 million of that was executed in the last three months at an average price below the $35.78 book value per share. That is the quarter. Why it matters: this is a company that just spent $1.7 billion on Inigo, repaid its revolver, and still had the balance sheet to retire 8% of its equity in the same year. Holding company leverage is now under 19%, group liquidity is $376 million, and Radian Guaranty will pay a $145 million Q4 dividend, lifting the 2026 total to $660 million against initial guidance of $600 million. The allocate capital story is doing more work here than the underwriting story.

What went quiet

Two themes vanished, and both are informative. First, the Middle East. In the second quarter it was the center of the Specialty debate; Kobell walked analysts through roughly $30 million of reserving — “the combined ratio was 98% in the second quarter. So that included the reserving related to the Middle East” — Dan Kobell, CFO · 2026-08-05. In the Q3 transcript the phrase never appears. What replaced it was a combined ratio of 80%, helped by $43 million of favorable prior-period development that Russell Merrett described as spread across short-tail and liability classes, with catastrophe development "a little bit less than half of the total." A 98%-to-80% swing that is half cat, half catch-up, and zero conflict-related — that is the whole Specialty narrative in one quarter, and it is not repeatable. Second, the mortgage business itself. Radian's own keyword history shows the Mortgage Insurance theme fading hard this quarter even as the segment generated $206 million of the $254 million in adjusted pretax operating income. Specialty gets 57% of net premiums earned and 34% of operating income, but Mortgage still supplies the overwhelming majority of the dollars. The narrative has become a specialty-insurance story years before the P&L is one.

Two risks nobody asked about a year ago

Bose George's question on FICO and VantageScore now sitting on the same grid produced the sharpest new disclosure on the call — and Weinbach did not soften it.

The same score under one model might mean a different level of credit risk than it does under another model. So pricing them the same could potentially be concerning because obviously, other things being equal, if lenders are best exiting the execution and we price the same, we could find ourselves inadvertently taking more risk for the same pricing or the same risk for lesser pricing.
Michael Weinbach, CEO · 2026-11-05

That is a mortgage insurer saying, politely, that score harmonization could compress its pricing power on a book with $286 billion of insurance in force. It lands in a quarter when the global keyword set is dominated by FICO's own Preferred Program discussion — the two companies are describing the same plumbing from opposite ends. Radian's riposte is operational readiness plus underwriting selectivity, but nobody on the call attached a number to the risk. The second new item is Washington. An executive order piloting simpler cancellation of mortgage insurance where home price appreciation supports it, using automated valuations, is a direct attack on the duration of an MI policy. Radian's answer was candidly unquantified: “It's way too early to have any specifics on what that looks like” — Michael Weinbach, CEO · 2026-11-05. Kobell was more useful, explaining that persistency at 83% is far more rate-driven than policy-driven — the automatic and borrower-requested cancellations have been stable quarter to quarter. True today; not necessarily true if AVMs shorten the tail.

Reading the default rate correctly

The company's single highest-momentum keyword this quarter is default rate, and the temptation is to read it as deterioration. It rose to 2.68%, as guided, and Kobell expects another increase in Q4 — for a structural reason rather than a credit reason: “as the book has seasoned and extended as we've seen fairly low level of cancellation activity, some of the older vintages, which typically tend to carry higher default rates just naturally are now making up a larger and larger percentage of the book... more so that dynamic than any deterioration that we see in terms of credit performance” — Dan Kobell, Executive · 2026-11-05 The offsetting evidence is cure behavior: cumulative cures are trending toward 96%–97% against a reserving assumption of 92.5%. That gap is why the loss provision is not the problem. It is also why the stock has not moved. Radian peaked at $39.71 on July 29 and sits 6.1% below that, essentially flat over 90 days — a strong quarter, a large buyback, and no re-rating, because none of it was a surprise to the tape.

The balance sheet after the transformation

The fundamentals confirm the shape of the pivot. Trailing total revenue was $575 million in Q2, up 81% year-over-year, while operating income was $151 million, down 22%. The acquisition was funded from the investment portfolio, which is why effective net cash fell 45% year-over-year to $2.5 billion — a deliberate swap of a 4%–5% yielding asset for a business expected to earn mid-teens through the cycle. That is the bet. On the call's own capital synergies question, Kobell pointed to the letter-of-credit facility as the first tangible proof, upsized and repriced for the 2027 underwriting year. That is real but small; the bigger capital release, as Geoff Dunn pressed, comes only when Inigo generates more than Lloyd's requires — and in a softening market, that is a 2027 story, not a 2026 one. So what changed? A CEO who is willing to say his prior guidance was too conservative, a specialty business whose 80% combined ratio flatters a benign cat quarter, a mortgage book still earning the money while the keywords migrate away from it, and two genuinely new policy-and-scoring risks that the market has so far treated as noise. Buybacks below book are a fine answer to a flat stock. They are not an answer to a credit-ordering change.