CNFinance at 8% of Book: The 1H Report Says the Credit Unwind Isn't Done

CNFinance Holdings (NYSE: CNF) has every outward marker of a cheap stock: down about 64% so far in 2026, trading near $2.20 against a roughly $39 million market capitalization, priced around 8 cents on the dollar of book equity, with management having bought back roughly US$19 million of ADSs — about half the current market value. Then the Chinese home-equity lender reported first-half 2026 results on August 28, and the report answers the one question a cheap-stock checklist never does: is the business getting better underneath, or worse? On these numbers, it is still getting worse.

The headline was a net loss of RMB 399.5 million (US$58.9 million) for the six months ended June 30, about ten times the RMB 40.4 million loss in the same period of 2025. Total interest and fees income fell to RMB 89.5 million from RMB 415.7 million, a decline of roughly 78% year over year. And the line that cuts to the economics flipped negative: net interest and fees income was minus RMB 88.4 million, meaning the company now pays more to fund its loan book (RMB 177.9 million of interest and fee expense) than the book earns. A falling stock can be a mood problem. A negative net interest margin is a business problem.

The reason goes straight to credit, and to the model that built this company. CNFinance made its name lending to small business owners with their homes as collateral; its current business is facilitating home equity loans and managing loans for borrowers and trust companies. Regulation and the property downturn pushed the operation toward that facilitation model — banks and trust companies now make the loans — while the legacy loans still on its own books aged into arrears. The share of its originated loans that were past due reached 63.4% at June 30, up from 50.3% at year-end 2025 and 46.0% a year earlier. The non-performing loan ratio eased, to 32.3% from 35.6%, but from a level that would finish most lenders. Management booked credit-loss provisions of RMB 340.7 million for the half, versus RMB 31.2 million a year earlier, on conditions it attributes to economic uncertainty. That provision was about 3.8 times the company's entire revenue line and accounts for roughly 85% of the net loss: this was a credit charge, not an overhead problem. Operating expenses, by contrast, fell about 58% as the company cut staff and streamlined.

That is the distinction worth internalizing. Tape pain — a falling price on an intact story — is the pitch of every beaten-down-stock presentation. CNF is not that. The loan book is the business, and the balance sheet still carries it: RMB 1.94 billion of on-book loans, plus RMB 3.09 billion of loans held for sale, against roughly RMB 3.0 billion of borrowings. When the market prices this at 8% of equity, it isn't being temperamental about a good company; it is pricing a stressed collection book and expecting more of it to go bad.

None of that proves the current price is right, and the honest tension is real. Equity still stood at RMB 3.17 billion (US$468 million) at June 30 against a market cap of about US$39 million. Even a harsh scenario that wrote off the entire on-book loan portfolio — about US$287 million of gross loans — would badly cut, but not erase, that cushion. The gap between a 468-million-dollar equity line and a 39-million-dollar market cap is the market saying it expects materially worse than today's books show: more provisions, more shrinkage, and a replacement business nowhere near big enough yet. The commercial-bank partnership model that is CNF's best hope of fee income contributed just US$0.6 million of net revenue in the half.

This is where the cheap-stock props fail one by one. The US$19 million of buybacks are a real commitment, but a company repurchasing its own shares does not change the delinquency trend. The roughly US$11.6 million of trailing free cash flow on a market-data screen is a lender's mirage: principal repayments from a shrinking book flow through the cash-flow statement even when the income statement bleeds, so the positive number describes a wind-down, not earnings power. And the aggregate signal on AInvest's service labels the stock Buy — the kind of opaque shorthand this report tests, because the reported financials are the evidence and they are deteriorating.

What would change the case is measurable, and worth naming. Credit provisions have to peak and losses have to narrow, which requires the delinquency ratio to stop climbing. Net interest margin has to turn positive as funding costs fall faster than income, or as a smaller, better book replaces the runoff. And the fee businesses — partnerships, loan management for trust companies, supply-chain finance — have to grow the top line back above a cost base management has already cut by about 58%. The early signs exist, and they are only early: the NPL ratio ticked down, and cash recoveries from overdue loans ran about RMB 600 million at a 103% recovery rate, real money pulled out of a book it supposedly can't collect. A 103% recovery rate on part of a 63%-delinquent portfolio is not yet an inflection. The break condition this report trips is plain: if each half brings another RMB 300–400 million of provisions and arrears keep climbing, the cushion gets consumed at a pace the market is already discounting.

Buying here is a bet that the credit cycle turns before the book spends more of an equity cushion that the latest half's loss alone reduced by about 12.6%. That could happen. But the discipline this trade demands is the discipline the report makes you practice: name the evidence that would make you reach for an 8%-of-book stock, and wait until it shows up. I can be wrong again — which is exactly why the tripwire matters more than the price.