REBUILT 7 September 2026. This page carried two exchange rates. Revenue was translated at the R$6.9931 to the dollar the company itself states in its 2025 annual report, and every other dollar figure at about 6.783, so net income, the balance sheet, the cash flow statement and the dividend all ran 3.1% high and every ratio crossing the two was wrong by that factor. The revenue field was worse: the $2,226M published matched no filed period on either rate, while the page's own price to sales of 0.65 implied the filed $2,746M. Everything below is now translated at the single stated rate and checked by a numeric gate. Earnings per share are the weighted-average diluted figures the annual report gives, not net income divided by a period-end share count, which is why they fell from $7.24 to $6.29. The variable interest entities produced 80% of revenue, not the 95% previously stated. Qifu is, on the raw multiples, the cheapest profitable credit platform on the public market. It earned $856M on RMB 19.2B (roughly $2.75B) of FY2025 revenue at a 31.2% net margin, holds about $535M of net cash against a $1.80B market cap, pays a $1.54 per-ADS dividend for a 10.4% yield, and has retired roughly a quarter of its shares in three years, from 322.8M to 243.8M ordinary shares. At 2.1 times trailing earnings and 3.1 times forward, the tape prices a business in terminal decline. The honest problem is that the decline is now real, not imagined. China moved its 24% all-in consumer-loan rate cap from a judicial guideline to a hard legal ceiling in October 2025, regulators are pushing the all-in cost toward roughly 12% by 2027 which makes above-cap interest legally unenforceable, Qifu was summoned for review, and its outstanding loan balance fell from about RMB 126B at the end of 2025 to about RMB 114B by March 2026. Management then guided Q2 2026 net income down 47% to 51% year over year. So the deep-value screen is right that the stock is cheap and wrong that it is safe. The bull case is narrow and specific. Even a lender whose book contracts under rate caps can return most of its market value in dividends and buybacks over a few years, and Qifu is doing exactly that, paying out roughly what it earns while sitting on net cash. The bet is cash extraction outrunning regulatory attrition, and the reported 86% free-cash-flow yield overstates the real cushion because a big part of it is non-cash loan-loss provisions added back and loan-book funding that moves through the cash-flow statement like any lender's. Buy it for the payout and the buyback, not for growth, and size it for the regulatory risk that is actively repricing the earnings.