MEV, Sandwiches, and Front-Running for Mortals
MEV stands for Maximal Extractable Value — the profit a block producer can make by reordering, including, or excluding transactions in their block. To you, that boils down to one thing: a bot can see your pending swap and trade in front of it. Here is when that matters and how to stop it.
The mempool is not private
When you sign a swap, your wallet broadcasts the signed transaction to a public mempool — a holding area where pending transactions wait to be picked into a block. On Ethereum, BSC, Base, and most EVM chains, the default mempool is fully public. Anyone running a node, including a bot watching for profitable patterns, can see every pending transaction including yours.
That visibility is the foundation of MEV. A searcher (the bot) reads your transaction, simulates it, and asks: "if I place a transaction before this one, can I profit?" If yes, they pay the validator a tip to order their transaction first. The validator gets paid, the searcher gets paid, you absorb the cost.
Anatomy of a sandwich
The classic MEV attack on retail is the sandwich. It works only when you're swapping in a pool where your trade size is large enough to move price meaningfully — usually 0.5%+ of pool depth. Three transactions, all in the same block:
The bot's profit is bounded by your slippage tolerance. If you set 0.5% on a major pair, the bot can extract at most 0.5% from your trade. If you set 5% on a memecoin entry, the bot can extract up to 5%. This is why slippage tolerance is not free — it caps your worst-case execution but it also caps the bot's best-case extraction. Setting it lower hurts the bot first.
When you are actually a target
Not every retail swap gets sandwiched. Bots have minimum-viable-profit thresholds; a $200 swap on a major pair is rarely worth the gas. Here is when you should expect to be sandwiched:
| Trade context | Sandwich risk |
|---|---|
| $50–$500 on Ethereum mainnet major pair | Low — profit doesn't beat gas |
| $500–$5,000 on a thin-liquidity altcoin | High — you are the ideal target |
| $1,000+ on Base / Arbitrum / Optimism | Medium — cheap gas raises the bot's hit rate |
| Any size on a brand-new pool (under 24h old) | High — sniper bots are camped on every new pool |
| Solana trades through Jupiter | Low–medium — Jito bundles + Jupiter routing limit exposure |
| BSC trades on PancakeSwap | High — the chain's mempool is heavily watched |
You can confirm after the fact. EigenPhi and libMEV let you paste a transaction hash and see whether it was sandwiched, by whom, and for how much. Look at any failed-to-deliver swap from the last week and check.
Mitigations that actually work
1. Use a private RPC / private mempool
The single biggest mitigation. Submit your transaction directly to a block builder bypassing the public mempool. Searchers literally cannot see it until it's already on-chain.
- Ethereum / Base / Polygon: Flashbots Protect RPC. Add as a custom RPC in MetaMask. Free. Blocks ~99% of sandwich attempts on Ethereum.
- Ethereum: MEV Blocker. Same idea, bundles your tx with searchers who can rebate you a share of MEV instead of extracting from you.
- BSC: BNB48 MEV-protect RPC. Limited but better than nothing.
- Solana: route through Jupiter with the "Jito Bundles" option enabled — uses Jito's auction to land your tx atomically. Most major Solana wallets enable this by default now.
# MetaMask custom network — Flashbots Protect
Network name: Ethereum (Flashbots)
RPC URL: https://rpc.flashbots.net
Chain ID: 1
Symbol: ETH
2. Set realistic slippage tolerance
Cap the bot's profit by capping your tolerance. On a major pair, anything above 0.5% is leaving money on the table. See the slippage math piece for the full ladder.
3. Split big trades across time and pools
A single $20k buy is a sandwich magnet. Four $5k buys spaced 2 minutes apart, routed through different pools, are individually below most bots' threshold and collectively land at a better average price. Aggregators do route-splitting automatically; time-splitting requires a TWAP / DCA mode (Jupiter has it built in; on EVM use 1inch Fusion+ or CoW Swap).
4. Use intent-based / batch-auction DEXs where available
CoW Swap, UniswapX, and 1inch Fusion don't broadcast a swap to the public mempool. They post your intent to a sealed auction, where solvers bid to fill it. The execution happens atomically with no front-runnable order. Pure-intent DEXs are functionally MEV-resistant by design.
5. Know which chains are exposed
L2s with centralized sequencers (Base, Arbitrum, Optimism, zkSync) currently have private mempools by default — sandwiches require collusion with the sequencer, which Coinbase / Offchain Labs / OP Labs publicly commit not to do. That changes when sequencers decentralize. Solana's leader-rotation model + Jito bundles produce a different threat surface; vanilla front-running is rare but bundle-stuffing exists.
Failed-transaction MEV — the silent tax
Even if you aren't sandwiched, you can pay MEV through transaction failure. A bot front-runs your buy with a buy of its own, eating the price impact you would have caused. Your transaction reverts because the price moved beyond your slippage tolerance. You pay gas for nothing; the bot exits with the price-impact profit minus a small back-run.
This pattern is most common on hot new launches. On a token that just deployed, half the failed transactions you see in the explorer are not "low slippage" — they're MEV-induced reverts. Mitigation is the same: private RPC + realistic tolerance.
0x1771, check the block. If a known MEV bot landed a buy + sell of the same token in the same block, you were not unlucky — you were a target. Switch to a private RPC and try again.
The honest summary
MEV isn't going away. The blockchain you are trading on has a finite block space, and that block space is contested by professional searchers with hundred-millisecond infrastructure. As retail you are not racing them — you are opting out of the race by routing through systems that don't expose your trade to them.
If you do nothing else, configure Flashbots Protect on EVM and trade through Jupiter with bundles on Solana. That single change captures most of the practical mitigation. The rest is sizing discipline.