Sandwich Attack Presale Risk: How MEV Bots Drain Buyers

If you bought a token on its listing day and watched your fill come in well above the chart price, only to see the price snap back seconds later, you probably weren’t unlucky. You were sandwiched. The sandwich attack presale problem is one of the most consistent ways retail buyers lose money on token launches, and it has nothing to do with whether the project itself is legitimate.

This guide walks through how the attack works, where in a presale lifecycle it actually applies, and what you can realistically do about it as a buyer who is not running their own validator.

What a sandwich attack actually is

A sandwich attack is a form of MEV — Maximal Extractable Value — where a bot exploits the public mempool. When you submit a swap on a public chain like Ethereum or BNB Chain, your transaction sits in a queue visible to everyone before it is mined. Bots scan that queue thousands of times per second.

Here is the sequence:

1. You submit a buy of token X for, say, 2 ETH with 5% slippage tolerance.
2. A bot sees your pending transaction in the mempool.
3. The bot pays a higher gas tip and submits its own buy of token X before yours. This pushes the AMM price up.
4. Your transaction executes next, at the now-elevated price, eating most of your slippage allowance.
5. The bot immediately submits a sell, dumping into the price impact your buy created. The bot exits with a profit equal to roughly the gap between the original price and your fill price, minus gas.

You paid the spread. You got fewer tokens than the chart suggested. The bot paid nothing of its own — your slippage tolerance was the budget.

The Paradigm essay “Ethereum is a Dark Forest” (2020) and Flashbots’ ongoing research are the canonical primers if you want the full technical picture.

Where presales fit in

Most fixed-price presales — where the token sells at a set USD value through a smart contract whitelist — are not vulnerable to sandwiching during the presale phase itself. There is no AMM, no slippage, no price curve to manipulate. You either get filled at the published price or you don’t.

The risk concentrates in three specific moments:

• Bonding-curve presales. Some launches use a continuous bonding curve where each buy moves the price up. These are mathematically identical to an AMM swap and are sandwichable in principle, though the bot economics depend on liquidity depth.
• The listing transaction. The single highest-risk moment is when liquidity is added to a DEX and trading opens. The first hundred blocks after listing are a feeding frenzy. Sniper bots and sandwich bots both compete here.
• Claim-and-flip flows. If the presale lets you claim and immediately swap, your claim-then-swap pattern is predictable and targetable.

If you are evaluating a project, the presale scam checklist covers the project-side red flags. This page is about the execution-side risk, which applies even to projects that are entirely legitimate.

How big are the losses, really

Chainalysis reported in 2023 that MEV extraction on Ethereum DEXs ran into hundreds of millions of dollars annually, with sandwich attacks making up a substantial share. On a per-trade basis, a buyer with default 5% slippage on a thinly-liquid token launch can routinely lose 2–4% of their position to a single sandwich. Across a portfolio of ten launches, that compounds.

The numbers get worse on chains with cheaper gas. Solana and BNB Chain have their own variants — Solana’s are structured differently due to its execution model, but the buyer-loss outcome is similar.

What you can actually do

There is no perfect defense without controlling block production, but the following tactics meaningfully reduce exposure.

Use a private transaction relay. On Ethereum, services like Flashbots Protect and MEV Blocker route your transaction directly to block builders without exposing it to the public mempool. This is the single most effective retail-accessible mitigation. Most major wallets now support adding a custom RPC. We discuss wallet-level support in the self-custody wallet shortlist.

Set slippage tighter than feels comfortable. A 1% slippage tolerance forces the trade to revert if the price has moved too much, which often means a bot can’t profitably sandwich you. The cost is failed transactions and gas burn during volatile listing windows. This is a tradeoff, not a free fix.

Avoid being first. The first ten minutes of a listing are the worst. Bots have pre-positioned and gas wars are at their peak. Waiting an hour, even at the cost of a slightly higher entry, often results in better net fills.

Split your buy. Multiple smaller transactions are individually less attractive to sandwich than one large one, because the bot’s profit must exceed gas costs.

Watch out for honeypot contracts that mimic sandwich losses. Sometimes what looks like a sandwich is actually a contract that taxes your buy at 30%. The token contract red flags guide covers how to tell the difference using a simulator like Tenderly.

What we cannot verify for you

We cannot tell you, in advance, which specific presale’s listing day will be heavily targeted. MEV bot behavior depends on liquidity depth, expected volume, and how predictable the listing time is. A surprise listing with shallow liquidity may be largely ignored. A heavily-promoted launch with a precise unlock time will be a war zone.

We also cannot verify that any given “anti-MEV” feature advertised by a project actually works. Some launchpads claim sandwich protection via batch auctions or sealed-bid mechanisms; some of those work, some are marketing. Test on small size before committing.

For broader context on how retail gets extracted from at every stage of a launch, see our news coverage of recent launch failures and the scoring methodology we use on individual projects.

Honest summary

Sandwich attacks are a structural feature of public-mempool blockchains, not a bug specific to bad projects. Even a perfectly legitimate token launch will have buyers losing 2–4% to MEV on day one if they use default settings on a stock RPC. The fix is not exotic: use a private relay, tighten slippage, don’t be first, and split orders. None of these eliminate the risk, but together they shift the math against the bots enough that you keep most of what the chart implies you should.