MEV searchers optimize for the highest risk-adjusted profit they can extract from a pending transaction before another searcher or the market removes the opportunity.
That means the target is not simply a profitable trade. It is a trade that remains profitable when the transaction executes, can reach a block builder in time, survives competition, and still pays less than the value it creates.
What a searcher needs before acting
A searcher starts with a pending transaction, usually from a public mempool or a private order-flow channel, then reconstructs the state in which it may execute. For a DEX swap, that means reading the calldata, pool reserves, token balances, fee tiers, and expected price impact. For a liquidation, it means checking collateral values, oracle updates, debt balances, and whether another transaction could make the account healthy again.
The basic calculation is expected profit: the value created if the transaction lands in the desired position, minus gas, the payment to the builder, capital costs, and the chance that the attempt fails. A trade with a larger theoretical payoff can therefore be worse than a smaller trade with a higher inclusion probability.
- Detect the state change. The pending transaction must reveal something actionable, such as a large swap, a liquidation threshold, or a price difference between venues.
- Simulate the exact sequence. The searcher tests its transaction before, after, or alongside the user transaction against a current state snapshot.
- Price competition. It estimates how much of the surplus can be paid as priority fee or builder bid while remaining profitable.
- Submit with timing and ordering constraints. The transaction may go directly to builders or inside a bundle whose order matters.
The result is often an atomic arbitrage bundle: the user's swap moves one pool's price, the searcher trades against that imbalance elsewhere, and the bundle pays the block-building supply chain from the surplus. If the sequence cannot execute as simulated, the opportunity is usually worthless.
Why inclusion probability matters more than headline profit
Pending transactions do not come with a guaranteed place in the next block. They compete with other transactions, builder strategies, nonce rules, gas limits, and changing state. A searcher therefore optimizes latency and reliability alongside price.
For a public-mempool swap, a bot may try to trade before the user, after the user, or both. A sandwich requires the searcher's first trade to land before the user's swap and its second trade immediately after it. A backrun needs the user transaction to execute first, but still needs enough block space and a bid that beats competing backrunners.
Under proposer-builder separation, searchers generally sell their opportunities to builders, while builders optimize the ordering of many transactions and bundles for the highest block value. This changes the path to inclusion, not the underlying arithmetic. The searcher still wants the largest net surplus, but it must express that surplus as a credible payment without giving away all of it.
This is why gas efficiency remains a competitive edge. Lower gas usage leaves room for a higher bid, and a smaller bundle is easier to fit within block limits. The practical costs are substantial: low-latency nodes or RPC access, continuous simulation, builder connectivity, engineering time, and capital that may sit idle while an opportunity disappears.
When the model stops applying
Not every pending transaction is an MEV signal. A trade may have too little price impact, a slippage limit may make the transaction fail, or the apparent spread may vanish before inclusion. A transaction sent through private order flow may be invisible to a searcher without access to that channel. A strategy that depends on two chains executing together also cannot be priced like a single-block Ethereum arbitrage.
That distinction matters for cross-chain activity. Across Protocol relayers price gas, capital opportunity cost, settlement timing, and fill deadlines. LayerZero Protocol adds destination verification and execution conditions through its DVNs and Executors. Axelar Network uses validator-mediated cross-chain delivery. In each case, the pending source transaction is only one part of the opportunity, so speed alone does not determine the best action.
As of September 2026, Ethereum's MEV design is moving toward more constrained and less transparent order flow. FOCIL is locked as the consensus-layer headliner for the planned Hegotá upgrade, giving eligible public transactions a stronger inclusion path. LUCID is a proposal for encrypted transaction commitments whose contents are revealed only after scheduling decisions are fixed. Neither turns today's searcher market into an encrypted mempool yet. Glamsterdam's planned gas repricings also make old simulation and gas-golfing assumptions worth revisiting.
The practical rule is simple: optimize the whole execution path, not the visible trade. A pending transaction is valuable only when its state change, ordering, access path, deadline, and builder payment all fit together.
The wider choice is between a route-specific execution model and a single cross-chain interface that absorbs these differences.
That comparison is the domain of Universal Bridge.