Back to Library

Dutch Auction

// Descending Price Auction

Auction where the price of an asset starts high and gradually decreases over time.

In a Dutch auction, the seller sets a high initial price, which decreases according to predetermined rules until enough bids are received to clear the available supply. The distinctive feature is the downward price movement, which creates a tension between waiting for lower prices and the risk of missing out on other bidders. This mechanism can be implemented with various clearing rules: buyers might pay their bid price (pay-as-bid), or all pay the same final clearing price (uniform price). The price decrease can be time-based (following a schedule) or demand-based (responding to purchasing activity).

The concept dates back to ancient times, with Herodotus describing a similar descending price auction in Babylon. In 17th-century Holland, this auction format was used for estate sales and paintings, where it got its name. The method appeared in England by the 17th century, where it was called "mineing”. In the blockchain context, Gnosis pioneered its implementation during its ICO in April 2017, followed by their launch of the DutchX platform in 2018. The DutchX smart contracts became a reference implementation for the broader DeFi ecosystem, though Gnosis later identified limitations regarding gas price sensitivity and last-minute trading dynamics. This mechanism has since been used in various contexts, including token trading and market making, NFT launches and token sales– most notably with Liquidity Bootstrapping Pools (LBP).

Variations

  • GDA (Gradual Dutch Auction): Breaks up a single sale into a sequence of Dutch auctions, allowing for efficient sales of assets without liquid markets. It is beneficial for NFT sales, where assets need to be sold in whole units, with each auction having a higher starting price than the last.
  • VRGDA (Variable Rate GDA): A modification that adjusts prices based on how sales track against a predetermined schedule, raising prices when sales are ahead and lowering them when behind. This creates a self-adjusting mechanism for maintaining target sale rates over time.

Advantages

  • Efficient Price Discovery: Effectively discovers the highest willingness to pay in markets with uncertain demand through automated price reduction.
  • Urgency Generation: Creates natural buying pressure through declining prices, helping clear markets quickly and preventing stagnation.
  • Fair Distribution: Enables equal access and prevents price manipulation by removing the advantage of timing or market power.

Limitations & Risks

  • Less Price Control for Sellers: Sellers lose the ability to influence the final sale price once the auction begins.
  • Winner's Curse: There's a risk of overbidding, potentially leading to price volatility after the auction.
  • Market Timing Risk: Participants must balance the trade-off between waiting for lower prices and risking losing the opportunity to buy.
  • Liquidity Fragmentation: When used for token trading, Dutch auctions fragment liquidity across different price points and venues, mainly since they are typically limited to a single token pair.

Design Considerations

  • Price Clearing: Define how final prices are determined. Options include pay-as-bid auctions, where participants pay their actual bid amount, maximizing revenue but encouraging last-minute bidding strategies, and uniform-price auctions, where all participants pay the final clearing price, improving fairness but requiring robust refund mechanisms.
  • Price Decay: Select how the price decreases over time. Consider time-based decay, where the price drops at fixed intervals to create predictability, and demand-based decay, where price adjustments respond to purchasing activity, accelerating if demand is low and slowing when demand is high.
  • Starting Price: Optimize the initial price to balance market engagement and efficiency. Look into historical benchmark pricing, where past market data informs the initial value, and tiered starting prices, where different asset classes or participant types (e.g., early supporters) start at different initial price levels.

Resources

Examples

Gnosis Token Sale

In 2017, Gnosis conducted one of the first Dutch auctions for a token sale, raising $12.5 million in just 10 minutes. The rapid completion time and high valuation demonstrated both the effectiveness and potential limitations of the Dutch auction format. While successful in price discovery and fundraising, the speed of completion suggested that the starting price or decay function might have been suboptimally calibrated, leading to fewer price points being explored than intended. This implementation became a reference point for subsequent token sales and influenced the development of more sophisticated Dutch auction variants.

Azuki NFT Launch

Utilized a Dutch auction format for their NFT mint in 2022, starting at a higher price point and decreasing over time. This approach helped manage the initial demand surge and reduced gas wars common in fixed-price NFT launches.

UniswapX

Employs Dutch auctions for trade settlement with solver competition, where multiple parties compete to find the most competitive prices for assets. This system protects users from MEV through third-party execution while maintaining AMM-like speed, though it's limited to single-token auctions.

Words3

Fully onchain PvP word game like Scrabble, where letters are placed on an infinite grid to spell words. Letters are priced via a VRGDA, which increases the price of letters when demand to use them is high and decreases when demand is low.

SudoSwap GDA

Uses discrete GDA for NFT liquidity pools. Their implementation treats each NFT sale as part of a continuous sequence of mini Dutch auctions, where the starting price of each subsequent auction is determined by the previous sale price and time elapsed. This approach helped solve the challenges of NFT liquidity by allowing prices to automatically decrease during periods of low demand while quickly adjusting upward after sales, preventing rapid depletion during high demand.