Timing Advantages in Onchain Auctions

Onchain auctions would be a huge unlock for many decentralized applications—NFT launches, liquidation engines, DEX routers, even wallets. If they were possible, then applications or users could use them to capture most of the maximal extractable value (MEV) they create.

But in a decentralized protocol, such auctions are very difficult to implement. As previous research has discussed, the ability of block proposers to reorder, censor, or peek at transactions gives them unique power to capture MEV from onchain auctions. While there are many proposed technical solutions to reduce the block proposer’s ability to extract this value, there is one advantage that is particularly difficult to eliminate: latency. Because timing delays can arise naturally in a decentralized setting, one party may be able to submit or cancel transactions later than anyone else.

Previous work has shown that if a block proposer can censor or peek at transactions, they can capture up to 100% of the value of a single-block auction. Is the same true of a block proposer who only has a timing advantage?

In our new paper, we find that while such a block proposer can extract some value from the auction, these auctions do not degenerate in the same way—they can still be competitive enough to capture some value for the seller. We quantify the profit that a proposer can extract from their timing advantage. We also derive the “timing pressure” on this auction (akin to the incentive for a block proposer to delay their block using timing games).

This paper preserves some hope that despite the inevitability of timing advantages, competitive decentralized auctions are still possible and can preserve some value for users. It also motivates work (like multiple concurrent proposers, FOCIL, and threshold encryption) that could add censorship-resistance and privacy to decentralized auctions, as well as work (like leaderless auctions) that tries to reduce timing advantages as much as possible.

Details

To see how our results work in a little more detail, consider two bidders, Alice and Bob, bidding to fill a UniswapX order. The price of the tokens (and therefore the value of filling the order) evolves in real time. One bidder (Alice) submits a bid early. The other bidder (Bob) moves later and has more up-to-date information (but cannot see Alice’s bid).

We show that Bob’s advantage translates into a strictly positive expected profit, even though he doesn’t know Alice’s bid. Alice, despite bidding strategically, earns zero profit in expectation. Importantly, the auction doesn’t completely collapse—value still accrues to the seller—but Bob is able to extract a significant surplus due to his timing edge.

How much? Well, let's consider two cases. If the UniswapX order has no reserve price, there’s a straightforward option-theoretic interpretation. Bob’s profit, in expectation, looks exactly like the payoff from a financial derivative: specifically, an exchange option that lets you trade one asset for another. If instead the order does have a reserve, the answer is slightly richer—Bob’s payoff becomes a portfolio of exotic options (which include range options and binary calls). This lens allows us to apply tools from financial mathematics—particularly the Black-Scholes model—to quantify how Bob’s edge depends on the volatility of the underlying asset and his latency advantage.