MEV: transaction ordering as an execution cost

Execution includes ordering

Maximal extractable value describes value associated with the inclusion, exclusion or ordering of transactions beyond ordinary protocol rewards. Ethereum’s MEV documentation discusses arbitrage, liquidations and sandwiching among the relevant activities. The economic consequences differ: some activity helps align prices or maintain lending systems, while other activity worsens execution for a particular user.

For a trader, the key issue is that submitting a transaction is not the same as executing it at the displayed quote. The market state can change before inclusion. A transaction’s permitted bounds determine whether it reverts or accepts an updated outcome. Those bounds are part of the order, not a cosmetic preference in the interface.

Distinguish price impact from slippage

Price impact is the effect of the trade itself on available liquidity. Slippage, in common interface usage, is the difference between the expected result and the eventual result as conditions change. Definitions vary, so read the platform’s labels. A large swap can have predictable impact even in a quiet market; a small swap can face unexpected slippage during rapid updates.

In a practice comparison, retain the initial quote, input amount, minimum acceptable output, actual output and transaction fees. Without all five, it is easy to blame every shortfall on ordering when some was already present in the quote. Separate ordinary market movement, route changes, impact and evidence of adverse ordering rather than treating a low-quality fill as proof of one cause.

Why wide tolerances create exposure

A permissive minimum-output setting allows the transaction to complete through larger adverse changes. That may reduce reverts but also permits worse execution. A very tight setting can fail repeatedly and incur network costs, depending on the chain and transaction outcome. Neither extreme is a universal optimum.

Consider a quote for 1,000 output tokens. A 1% tolerance implies a minimum near 990 under this simplified convention; 5% implies 950. The second order authorizes a materially worse outcome. It does not mean the market is expected to move 5%, and it does not turn the 5% into a fee paid automatically. It defines a boundary on acceptable execution under the transaction’s actual rules.

Evaluate protected routes as services

Private submission, batch auctions and other mechanisms can change who sees an order and how it is processed. Their protection depends on implementation, counterparties and fallback behavior. A private route may reduce one form of exposure while introducing trust in relays, builders or an auction mechanism. Verify whether failed private submission falls back to a public route.

For an educational comparison, request or reconstruct quotes for the same size and time, then account for fees, rebates, inclusion delay and success rates. A better headline quote that rarely executes may not produce better realized outcomes. A service claiming protection should explain its scope: hiding an order from some observers is different from guaranteeing a best available price.

Build an execution-quality record

Use a consistent benchmark such as the quote at submission, while acknowledging that it is not always executable later. Track effective cost by trade size, route, congestion and asset liquidity. Include failed transaction expenses instead of evaluating only successful swaps. A small sample of favorable fills is insufficient to establish a durable advantage.

The purpose of studying MEV is to understand market structure and reduce unexamined execution assumptions. Avoid assuming that an apparent arbitrage is easy money: competition, latency, failed transactions and adverse price changes can consume the spread. Historical opportunities visible after the block was built do not imply that an ordinary participant could have captured them beforehand.

When recording fills, compare like with like. A route that returns a different wrapped asset or settles on another network has changed the deliverable. Include the cost and delay of converting it into the desired asset before claiming an improvement. Apparent execution savings can disappear when the final inventory is made comparable, especially for small orders where fixed conversion charges dominate.

Your minimum output is a limit, not a target

Suppose a fictional swap quote shows 1,000 output units and a 1% tolerance sets a minimum of 990 under the quoted interface’s convention. Receiving 995 satisfies that limit, but it is still five units worse than the original quote. Receiving 990 is not evidence that the route found the best available execution; it only establishes that the transaction did not breach that particular bound. Check the actual minimum-output value rather than assuming every interface calculates a displayed percentage identically.

Evaluate the transaction against a timestamped benchmark and record the amount actually received, network cost, any service charge and whether earlier attempts failed. A route that saves three units but incurs five units of extra cost has not improved the all-in result. Conversely, a rejected transaction may have protected the minimum output while still consuming a fee, depending on how it was submitted and processed.

Do not diagnose every disappointing fill as an attack. Ordinary price movement, your own price impact and changes in available liquidity can also explain a difference. Transaction ordering is an additional mechanism: Ethereum’s MEV documentation describes how sandwiching can worsen a user’s execution. Attribution requires evidence about the sequence, not just a bad outcome.

Compare the whole order, not its best fragment

Splitting an order can reduce the size of each individual trade, but it also creates more fee-paying actions and more time between the first and last fill. In a fictional comparison, route A executes the entire order once. Route B uses five smaller swaps. Add all five fills and all associated costs before deciding whether B was better; selecting its best fill would answer a different question.

Use the same starting time and intended quantity for both paper routes, then record any amount left unexecuted. If one route completes only half the order, its attractive average price should not be presented as the cost of completing the full instruction. Include the remaining exposure to price changes and the conditions that would make you stop waiting.

This kind of record is also useful when comparing private submission or protected execution services. Their value is not established by the word “protected.” The meaningful questions are what threat the service addresses, what information it receives, what happens if execution fails and whether the full result improves after costs. One route can reduce ordering exposure while adding a different operational dependency.