MEV and sandwich attacks: how much they actually cost you

A sandwich attack is one form of MEV — value captured by whoever decides what order transactions execute in. A bot buys immediately before your buy, lets your order push the price up, then sells immediately after into the price you just created. The amount it can take is bounded by the slippage tolerance you set.

What MEV actually is

MEV stands for maximal extractable value: the profit available to whoever chooses which transactions enter a block and in what sequence. It is not a flaw in one protocol but a consequence of three ordinary facts — trades settle in discrete blocks, somebody decides each block's contents, and a trade's price depends on what executed immediately before it.

Most MEV research is written for searchers and validators. You cannot reduce the supply of MEV; you can only reduce how much of your own order is exposed to it. Of the common shapes, only one reaches you directly — arbitrage between pools costs you nothing, liquidations are irrelevant unless you have borrowed. Sandwiching comes out of your fill, and it scales with a number you choose.

How a sandwich works, step by step

Your swap is not private between clicking and confirming. In that window a third party can read it — from a public mempool on EVM chains, or from transaction flow reaching a block producer on Solana — including your input size and, critically, your minimum acceptable output.

From there it is mechanical. The bot places a buy in the same block, ordered before yours, moving the pool's reserve ratio so your price is worse than quoted. Your trade executes at that worse price and pushes it further the same way. The bot sells immediately after you, into the price your order created. Its profit is the spread, net of fees and whatever it paid for that ordering.

None of it requires an opinion about the token. The bot holds the position for a fraction of a second across your trade, which is why sandwiching works on tokens nobody would otherwise touch. It is profitable on memecoins specifically because of pool depth. In a constant-product pool, price impact is a function of your order relative to the reserve — the mechanism is set out in how a constant-product reserve responds to an order. A thin launch-hour pool converts a modest order into a large price move, and that move is the raw material a sandwich needs.

Why your slippage tolerance sets the size of the loss

Your transaction carries a minimum-output figure derived from the tolerance you set, and the pool reverts your trade if the fill drops below it. That protects you from an unlimited loss — and tells the bot exactly how far it may push the price before your trade breaks. A rational sandwicher sizes its front-run to leave you fractionally above your own floor.

So the tolerance is not only a safety limit — it is an upper bound on the extraction. A 1% setting caps the take near 1%; a 25% setting authorises something much larger. Raising slippage until trades stop failing does work, and the cost appears nowhere on your receipt: it looks like a bad fill and gets filed as volatility. The method for setting the number is in how to size a slippage tolerance to your own price impact.

The second lever is time. Every block spent waiting is a block in which your order can be seen and ordered against, so bidding for faster inclusion is a different and better fix than widening tolerance. On Solana that bid is a compute-unit price, covered in what a priority fee actually buys you.

Where sandwiching happens on Solana, and why it looks different

Solana has no global public mempool. Transactions are forwarded straight to the current and upcoming block leaders rather than gossiped to everyone, so the EVM pattern of watching an open queue does not apply here.

That has not removed sandwiching; it has moved it. The parties who see your transaction before it lands are those close to block production — leaders, the block-engine infrastructure most validators run, and anyone with a private order-flow arrangement. Jito withdrew its public mempool product in 2024 after it became a sandwiching venue; the tip-and-bundle mechanism that remains lets a searcher submit an ordered transaction set that either all executes or none does — a better tool for this than a public queue ever was.

So "no mempool" is not protection, and any interface marketing it that way deserves scepticism. What helps is routing that does not expose your order before execution: an MEV-protected RPC endpoint, a venue submitting directly to a leader, or a route settling as a bundle. Read what the setting claims — "private" can mean a direct submission path or a preference flag that degrades under load.

Robinhood Chain: one sequencer is a structurally different risk

Robinhood Chain is an Arbitrum Orbit L2 with roughly 250ms blocks, on mainnet since 1 July 2026. L2BEAT's assessment as of 31 August 2026 records that it does not reach Stage 0, listing among its findings a centralised sequencer with MEV capability, fraud proofs resting on two whitelisted validators, critical contracts upgradeable by an externally owned account, and an authorised filterer able to force any transaction hash to fail.

Take the MEV finding on its own terms. There is no public mempool for outside searchers, so the retail sandwich vector familiar from Ethereum is largely closed — genuinely better. But it is closed because ordering authority sits with a single operator rather than being contested. On Ethereum a competitive builder market means extraction is bid down by rivals and users can route around any one of them. With one sequencer there is nothing to route around and no competitor to bid against; what stands between you and ordering-based extraction is the operator's policy and reputation, not competition and not cryptography.

Be precise about the epistemics: L2BEAT describes a capability, not an observed abuse. We have seen no evidence the sequencer has extracted MEV from users. The point is that the risk is concentrated and policy-dependent rather than diffuse and market-priced, and a chain without fraud-proof decentralisation gives you no way to check independently. The wider posture is unpacked in what Stage 0 means and why this chain has not reached it. For daily chain-level coverage, Locksley — an editorial and data site built by TrustSwap — publishes briefings and a screener for the chain.

How to tell whether it happened to you

Open the transaction in a block explorer and look at what else executed in the same block or slot against the same pool. A sandwich leaves a shape: a buy immediately before yours and a sell immediately after, often from one address or two funded by a common source, sized to your order rather than to anything about the token. If no such pair exists, a worse-than-quoted fill was most likely your own price impact plus ordinary drift — not MEV, and not fixed by private routing.

What this doesn't tell you

We are not publishing a figure for what MEV costs the average memecoin trade, because we do not have one we would stand behind. Published estimates vary by an order of magnitude depending on chain, window and whether arbitrage is counted as a user cost.

Attribution at the individual level is weaker than it looks, too. Two transactions around yours can be a sandwich, an unrelated bot pair, or coincidence in a busy block, and on-chain data alone often cannot separate them. Treat a single suspicious block as suggestive, not proven.

None of it addresses the dominant cost. Execution quality is worth a few percent; whether the token should have been bought at all is worth everything else, and a perfectly protected trade into a pool that gets drained is still a total loss — a dynamic set out in why speed at the launch moment is not a contest you win.

Frequently asked questions

What is a sandwich attack on Solana?

A bot places a buy immediately before your swap and a sell immediately after, in the same slot, so you buy at the price its front-run created and it exits into the price yours created. Solana has no public mempool, so it depends on proximity to block production — leader access, block-engine bundles or private order flow — rather than on watching an open queue.

Does high slippage cause sandwich attacks?

It does not cause them, but it determines how large one can be. Your minimum-output figure tells the attacker how far price can be pushed before your trade reverts, and a rational one stops just short. A wide tolerance is an authorisation, not a convenience.

Is MEV a problem on Robinhood Chain?

Differently shaped. There is no public mempool for outside searchers, which closes the common retail vector, but L2BEAT records a centralised sequencer with MEV capability as of 31 August 2026 — ordering power concentrated in one operator rather than competed for. That is a capability finding, not evidence of abuse, and there is no independent way for users to verify sequencer behaviour.

---

Lock the liquidity before execution quality matters

A clean fill is worth little if the pool can be emptied afterwards. Team Finance — built by TrustSwap, which also builds Meme Central — locks LP tokens for a fixed term on Ethereum, Robinhood Chain, Polygon, Base and BNB, and the lock shows as a verified badge on the token's page in the cross-chain launch feed. A lock does nothing about transaction ordering and does not stop a creator selling their allocation. It guarantees only that the pool is still there when you exit.

---

Nothing here is financial, legal or tax advice. Memecoins are extremely high-risk: most lose most of their value, and the majority of tokens launched never reach a decentralised exchange at all. Never spend money you cannot afford to lose entirely. Meme Central does not recommend any specific token. Data described as Meme Central's own reflects tokens indexed by Meme Central and is not whole-market data.