Dutch Auction Liquidation

What is a Dutch auction liquidation, and how does it differ from the liquidation bonus discount mechanism covered in an earlier article?

The liquidation bonus discount covered in an earlier article is a 'fixed discount' mechanism — regardless of the current market condition, once a position triggers liquidation, a liquidator can buy the collateral at a pre-set fixed percentage discount (say, 95% of market price) — this discount ratio is static, hardcoded in advance. A Dutch auction liquidation, by contrast, is a dynamic pricing mechanism — collateral's asking price isn't a fixed discount, but starts from an initial point close to or even above market price, continuously decreasing as time passes, possibly lowering by a percentage at set intervals, or smoothly declining according to a continuous mathematical formula (commonly an exponentially decaying curve), until price drops to a level a liquidator becomes willing to take on, ending the auction.

The key difference from fixed-discount liquidation lies in 'whether the discount magnitude reflects genuine market demand in real time': fixed-discount liquidation's discount ratio stays unchanged regardless of how strong or weak the market's actual demand for this batch of collateral currently is — if market demand happens to be strong, with liquidators competing for it, a fixed discount could cause a borrower to suffer more loss than necessary; Dutch auction liquidation, by contrast, lets the discount magnitude dynamically adjust over time — if market demand is strong, a liquidator might become willing to take it on before the price has decreased much, with a relatively small discount magnitude; if market demand is weak (say, an overall sharp market decline with everyone dumping, the liquidation cascade scenario covered in an earlier article), the discount magnitude keeps expanding until attracting a willing buyer — this dynamic adjustment, theoretically, more closely matches the current genuine market supply-demand condition.

Why does a Dutch auction liquidation exist, and what problems with fixed-discount liquidation is it trying to solve?

A fixed-discount liquidation mechanism is simple to design and executes quickly, but carries a structural flaw — the discount ratio is a static figure set in advance, requiring the designer to guess in advance 'how large a discount, under various future market conditions, would be sufficient to attract a liquidator to act.' If the discount is set too conservative (too small), during a period of tightened market liquidity when liquidators bear higher risk, no liquidator might be willing to act, causing the liquidation process to delay and bad debt risk to rise; if the discount is set too aggressive (too large), then under normal market conditions where a liquidator actually wouldn't need such a high incentive to act, the borrower gets forced to bear more unnecessary loss than needed — this extra discount is, to some extent, an efficiency loss.

What a Dutch auction liquidation aims to solve is exactly this problem of 'a single fixed discount can't adapt to every market condition': by letting price dynamically adjust over time, the auction mechanism itself automatically finds the discount level the current market genuinely needs — during good market conditions, the discount might quickly converge to a very small figure; during poor market conditions, the discount naturally expands to a level sufficient to attract a buyer to enter. This, to some extent, converts the difficult prediction problem of 'guessing the correct discount ratio in advance' into a dynamic discovery mechanism of 'letting the market naturally find the answer through time,' without needing the protocol's governance to continuously manually adjust discount parameters in response to different market conditions.

How does a Dutch auction liquidation actually work, and what does one complete auction flow look like?

Taking a more mature implementation design as an example, a typical Dutch auction liquidation flow involves several steps:

1. Triggering the auction: once a position falls below the liquidation threshold, the protocol formally initiates an auction targeting this collateral, setting an initial price (usually referencing the current market price, plus a certain premium percentage as the starting quote), plus at what pace this price will decrease
2. Price automatically decreasing over time: once the auction starts, price decreases continuously and automatically according to a pre-set curve (a common design is an exponentially decaying curve, where the pace of decrease itself also changes over time), needing no human intervention — the contract calculates in real time 'what the price should be at this current moment'
3. A liquidator settling in real time: at any point, once a liquidator (the independent third party driven by economic incentive covered in an earlier article) judges the current auction price already profitable, they can immediately buy this batch of collateral at the current price — the entire transaction usually settles instantly, not needing to wait for a fixed bidding deadline to close like a traditional auction would
4. Settling the debt, returning remaining assets: the payment a liquidator makes goes toward repaying the original borrowed debt — if the auction proceeds exceed the debt amount, the remainder usually gets returned to the original borrower; if the auction runs for quite a long stretch, with price already dropping to an extremely low level and still nobody willing to take it on, it could trigger the protocol-level bad debt handling mechanism covered in an earlier article

Worth noting: some newer Dutch auction designs further lower a liquidator's participation capital barrier — a liquidator doesn't need to prepare sufficient capital in advance to buy the collateral, but can complete an entire flow of 'buying collateral, immediately reselling it for cash, repaying the originally fronted capital' within the same transaction by instantly tapping into another decentralized exchange's liquidity — to some extent letting any participant holding an extremely small amount of capital (only needing to pay the network transaction fee) genuinely participate in the liquidation auction, expanding the potential counterparty pool willing to participate in bidding.

What's the practical impact of a Dutch auction liquidation on everyday users, and how can you assess whether a protocol you use adopts this mechanism?

For a borrower, a Dutch auction liquidation theoretically prices closer to genuine market demand than fixed-discount liquidation — if the market's actual demand for this batch of collateral happens to be high, the liquidation loss a borrower bears could be smaller than under a fixed-discount mechanism. But if the Dutch auction's price starts declining from a relatively high premium starting point, and a liquidator doesn't act promptly enough, or overall market liquidity is insufficient, the auction might need a period of time before price drops to a level sufficient to attract a buyer — during this period, the borrower's position remains in an uncertain state of ongoing liquidation.

A few concrete verification directions when assessing whether a protocol you use adopts this mechanism: verify whether the protocol documentation explicitly describes the auction's starting price setting logic and the price-decline speed curve — setting the starting premium too high or the decline speed too slow could cause the auction to need a longer time to close, increasing uncertainty both the borrower and the entire protocol bear; verify whether this protocol also designs the mechanism covered in an earlier article letting a liquidator participate in the auction by instantly tapping into another exchange's liquidity — this kind of design effectively expands the counterparty pool willing to participate in bidding, lowering the probability of the auction failing to close (nobody willing to take it on); and, if the auction genuinely fails to close, running for quite a long stretch of time without any transaction, verify what kind of backup handling mechanism the protocol subsequently activates (such as the bad debt socialization or insurance fund covered in an earlier article), ensuring even if the most extreme scenario occurs, the protocol still has a clear response plan, rather than falling into an indefinite deadlock.