There Is No Third Auction. Your Foreclosure Math Is Wrong Twice.

TL;DR: Most explainers describe China's court foreclosure sales as a three-auction staircase: first auction, second auction, third auction, then a sale by listing. That staircase belongs to the old traditional-auction rules. Under the current online judicial auction rules, there are exactly two auctions. The first must start at no less than 70 percent of the appraised value. If nobody bids, a second auction follows within 30 days, with a cut of no more than 20 percent. If that fails too, the property moves to a listing sale. There is no third auction — and anyone pricing a deal off the wrong staircase has the wrong timeline by 30 to 60 days and the wrong price floor by 8 points.

Start with the floor, because it is where the bad math compounds. Multiply the two legal extremes — 70 percent, then another 20 percent off — and you get 56 percent of appraised value. That number circulates as if it were the price foreclosures clear at. It is not a price. It is the extreme that occurs only if a court pushes both steps to the statutory minimum. In the first ten months of 2026, residential foreclosures in China actually sold at 79.0 percent of appraised value, across the whole market. In a worked Shenzhen example from the same data, the second auction opened at 64.0 percent of appraisal and was bid back up to a sale at 73.0 percent. Courts often do not push the first step to the floor at all. Anchor a negotiation on 56 percent and you will not sound disciplined. You will sound uninformed, in every negotiation, all year.

The second error is treating the appraisal as a market price with a discount attached. If homes clear at 79 percent of appraisal as a national average, the appraisal itself is systematically high — by roughly a fifth to a quarter. The discount rate is not a national constant either. It is a function of city tier and property type, and the spread is not subtle.

What the clearance data actually says

Clearance rate — sold units divided by listings whose sale window has closed, not by everything ever listed — tells you which assets the endgame market will absorb at all. For January to October 2026: residential cleared at 38.4 percent, industrial at 24.5 percent, land at 22.5 percent, commercial at 16.3 percent, and parking spaces at 9.2 percent. The residential unit and the shop on its ground floor live in different liquidity worlds: a 2.4-times difference in clearance and an 11-point difference in discount depth, in the same building. Underwrite a commercial unit with residential exit assumptions and the model is fiction before the first payment.

Volume is the third misreading. Listings rose 31.6 percent and completed sales rose 34.6 percent by count — but only 19.0 percent by value. Do the division and the average sold unit fell about 11.6 percent. That is not a recovery. That is a market clearing inventory by cutting price: more units, less money per unit. Reading rising foreclosure volume as a warming market, and raising your offers into it, is the most expensive interpretation error available this year. One honest caveat: these are one research institute's database figures, reported in full by two financial outlets — not a court-published census — and city-level clearance and discount rates for the two cities I work in most were not in the public release. Until I compute them from my own sample, I label every use of the national 79 percent as a national figure, not a local measurement.

What I changed in my own underwriting

I do not bid at these auctions. My real estate work is pre-foreclosure: negotiating a direct takeover with the owner before the staircase runs. Which is exactly why the staircase matters to me — not as a buying channel, but as a public clock and a public ruler.

A failed first auction is a free referendum. The market has just announced, in public, that 70 percent of appraisal was too much. The court says to the owner what I could never say credibly across a table. So my outreach is now sequenced by the ladder, not by debt size: first contact within seven days of a failed first auction, a second touch when the second auction is announced, and the listing-sale phase treated as the best-evidenced window of all. My data layer now records, for every property, which regime it is under and which step it sits on — read from the court's own notice, never assumed — because the two regimes differ by a month of timeline and 8 points of floor. And commercial and parking assets no longer pass through my intake on residential exit assumptions; at 16.3 and 9.2 percent clearance, they need their own exit story, in writing, or they do not pass.

The staircase is a pricing machine built out of deadlines. Most commentary reads it as a discount chart and gets both the steps and the floor wrong. Read it as a clock instead: value does not slide down it smoothly — it falls in steps, on published dates, in public. The negotiation is won by whoever shows up right after a step lands.

— Jiahui Miao, founder of vertciti, building a personal AI operating system for one.