What is a coincidence of wants?
Two orders that mirror each other can simply trade with each other. The mechanism that lets batches settle peer-to-peer before touching a pool, and the origin of CoW Protocol's name.
Last updated:
The short version: when one order in a batch sells what another order buys, the matched portion can settle directly between the two at the batch's uniform price: no pool, no liquidity-provider fee, no price impact. Whatever does not match routes to on-chain liquidity as usual. It takes a solver built to combine orders: live today on the CoW-hosted chains, not yet in the solver lineup on the Ophis-operated ones.
What is a coincidence of wants?
A coincidence of wants (CoW) is when two parties each hold what the other wants: one order sells the token another order is buying, at compatible prices. In that situation the two can trade directly with each other, and no market, pool, or intermediary is needed for the exchange itself.
How do batch auctions find coincidences of wants?
By collecting orders into discrete batches instead of executing them one by one. A solver that considers multiple orders together can spot overlap: a sell of WETH for USDC and a buy of WETH with USDC can settle against each other at the batch's uniform clearing price, with only the unmatched remainder routing to on-chain liquidity. Whether that actually happens depends on the solver lineup, not on the contracts: it takes a solver built to combine orders.
Why is a peer match better than routing through an AMM?
Because the matched portion skips the costs that pools impose: no liquidity-provider fee, no price impact from walking a curve, and no exposure to the pool's state at execution time. Two traders simply cross at one price. That saving shows up as better execution, and better execution is how a solver wins the batch, so solvers have a direct incentive to find every match that exists.
Do coincidences of wants happen on Ophis today?
On the CoW-hosted chains, orders settle through CoW Protocol's solver competition, which includes solvers that combine orders, so matches happen there when opposite flow overlaps. On the Ophis-operated chains (Optimism, Unichain, and Robinhood Chain), the current solver lineup builds each settlement against on-chain liquidity per order, so peer matches do not occur there today; the settlement layer supports them, and they arrive with the solver lineup, not with a contract change. Where matching runs, the rate depends entirely on opposite flow in the same window, so any fixed number would be made up.
Is this where the name CoW Protocol comes from?
Yes. CoW Protocol, the settlement layer Ophis is forked from, is named after the coincidence of wants. Ophis inherits the batch-auction mechanism and runs its own settlement deployments on its operated chains; whether peer matching is active on a given chain follows the solver lineup there.
Go deeper
The full batch-auction mechanism around this: what is an intent-based DEX aggregator. Who does the matching and why they bother: what is a solver. How the same batches shut down sandwich attacks: what is a sandwich attack.