MEV: A Reference

Why MEV matters

MEV (Maximal Extractable Value is profit block producers and searchers extract by reordering, including, or excluding transactions in a block) still drains tens of millions from Ethereum users annually — but sandwich attacks fell ~75% in 2025 as private RPC adoption grew. An active retail trader on default settings is still losing 1–2% on every medium-size swap. The fixes are free and take a minute. Last verified: 2026-05-27.

Sandwich attacks drained ~$40M from Ethereum users in 2025, down from peaks near $10M/month in late 2024 to ~$2.5M/month by October 2025 per EigenPhi data, even as monthly DEX volumes climbed from $65B to over $100B. Meanwhile MEV Blocker (now under Consensys after a January 2026 acquisition) has paid out 6,177 ETH in cumulative rebates and protected over $60B of DEX volume as of May 2025 (MEV Blocker is a free private RPC that routes transactions through a private mempool, eliminating public sandwich risk). The infrastructure has matured — the remaining losses fall almost entirely on traders still using default RPCs with loose slippage. This guide closes that gap.

MEV in 30 seconds

Whoever orders transactions in a block can profit by inserting their own, reordering others, or excluding some. That profit is MEV. On Ethereum today the ordering happens via MEV-Boost — a sealed-bid auction among block builders — and the next upgrade, Glamsterdam, will enshrine this process into the protocol itself. Last verified: 2026-05-27.

Every blockchain produces blocks containing ordered transactions. Whoever controls ordering — historically miners, now block builders (block builders are specialized nodes that assemble Ethereum block contents and bid for inclusion via MEV-Boost relays) selected via MEV-Boost (MEV-Boost is the relay system used to build 90%+ of Ethereum blocks, outsourcing block construction to specialized builders) — can profit by:

1. Inserting their own transactions before or after yours.
2. Reordering transactions to capture arbitrage or sandwich profit.
3. Excluding transactions (rare; mostly censorship-related, e.g. OFAC sanction screening on some builders).

That profit is called MEV — Maximal Extractable Value. Originally coined as "Miner Extractable Value" in Phil Daian et al.'s 2019 paper Flash Boys 2.0; after Ethereum's transition to proof-of-stake and the rise of block builders, the M was redefined to Maximal because the extractor is no longer necessarily the miner. Today the value flows along a clear pipeline: searcher (finds the opportunity) → builder (assembles the optimal block) → proposer (validator who publishes it). Each layer takes a cut.

The five main types of MEV

Arbitrage (benign, keeps prices aligned), sandwich attacks (purely extractive), liquidations (necessary for lending solvency), NFT sniping, and long-tail strategies like JIT liquidity. Only one of these — sandwiches — is unambiguously extractive at user expense. Last verified: 2026-05-27.

1. Arbitrage MEV (mostly benign)

Two DEXs show different prices for ETH/USDC. A searcher atomically buys cheap on one, sells expensive on the other, pockets the difference. Result: prices converge across venues, and the market becomes more efficient. You benefit (better prices going forward) because spreads stay tight. Arbitrage is the bulk of "healthy" MEV and is the reason DEX prices track CEX prices within fractions of a percent.

2. Sandwich attacks (purely extractive)

You submit a large swap. A bot:

1. Front-runs you (buys the same token first, pushing price up).
2. Lets your transaction execute at the worst possible price within your slippage tolerance.
3. Back-runs you (sells immediately, capturing the spread).

You pay the price impact twice. Common on swaps with high slippage tolerance, especially the default 1–3% that wallets ship with. Monthly sandwich extraction on Ethereum collapsed from ~$10M in late 2024 to ~$2.5M by October 2025, driven by private RPC adoption.

3. Liquidation MEV

When an on-chain loan becomes undercollateralized, anyone can liquidate it for a fee (5–10% of position). Bots race to be first; the winner captures the fee. Mostly necessary — liquidations keep lending protocols solvent, and without competitive bot infrastructure, undercollateralized positions could persist long enough to leave the lender insolvent. The 2020 Black Thursday Maker incident, where some liquidations cleared at $0 due to gas-price spikes, is the canonical lesson in what happens when the liquidation pipeline fails.

4. NFT sniping

A mispriced NFT listing appears in the mempool. A bot front-runs the buyer to grab it first, then re-lists at market price. This was rampant on OpenSea pre-Seaport; bot-resistant protocols like Blur and the newer NFT marketplaces have driven it down meaningfully.

5. Long-tail strategies

• Time-bandit attacks — reorganizing past blocks to capture historical MEV. Rare on Ethereum post-Merge because reorgs are penalized; was a theoretical concern in 2020–2022.
• JIT (just-in-time) liquidity — add concentrated liquidity to a Uniswap v3 pool right before a known large swap, capture the swap fees, remove liquidity. Looks like LPing; is really MEV against passive LPs.
• Atomic searcher strategies — multi-protocol combos chaining arb, liquidation, and flash loans into a single bundled tx.

How sandwich attacks work (deeper)

A sandwich attack (a sandwich attack is an MEV strategy where a bot front-runs a victim's DEX swap, lets it execute at a worse price, and back-runs to capture the spread) front-runs your swap to push price up, lets your trade execute at the worst price within your slippage tolerance, then back-runs to capture the spread — typically 1.5% of trade size on a default-slippage swap. Last verified: 2026-05-27.

1. You submit:    swap 100 ETH → USDC, slippage 2%
   → enters public mempool
2. MEV bot sees:  the same pool, same direction, big size
3. Bot submits:   buy ETH first, pushing price up 1.5%
4. Your swap:     executes at +1.5% worse price
5. Bot sells:     pockets ~1.5% of your trade value

Your loss: ~1.5% of trade size. Bot's profit: roughly the same, minus a gas premium it bid to outpace competing sandwich bots and to compensate the block builder for ordering preference. Sandwich profits flow: searcher → builder (via bribe) → proposer (validator). Despite the 2025 decline in sandwich volumes, over 4,400 attacks per day still occur on Ethereum — average profit per attack has simply dropped to ~$3 because most high-value swaps now use private RPCs.

How to avoid getting sandwiched

Switch your wallet RPC to MEV Blocker (free, one minute, drop-in). For trades above $10k, route through CoW Swap or 1inch Fusion. On Solana, Jupiter routes through Jito BAM bundles automatically. Set slippage to 0.5% on majors and 1% on alts. Last verified: 2026-05-27.

For most users: switch your wallet RPC to MEV Blocker. It's free, takes one minute, and stops virtually all sandwich attacks at zero usability cost.

How to set up MEV Blocker

1. Go to mevblocker.io.
2. Click "Add to MetaMask" (or copy the RPC URL into Rabby).
3. Set as default network RPC for Ethereum mainnet.
4. All subsequent transactions route through the private mempool — invisible to public-mempool searchers.

Flashbots vs MEV-Share vs CowSwap: the comparison most explainers skip

Flashbots Protect is a private mempool (pure prevention). MEV-Share is opt-in MEV redistribution (you share tx hints, you get a kickback — 90% of extracted value by default). CowSwap is a batch-auction DEX where solvers internalize MEV and rebate it to traders. They are not interchangeable. Last verified: 2026-05-27.

The trend across all three: extracted MEV is increasingly being rebated to the user submitting the transaction, rather than blocked entirely. This is "MEV redistribution," and it's how the industry has converged on solving sandwiches in 2024–2026.

Best MEV protection by use case

Everyday EVM users: MEV Blocker RPC. Large trades: CowSwap or 1inch Fusion. Solana: Jupiter. NFT mints: Flashbots Protect. L2s: built-in protection from centralized sequencer ordering. Last verified: 2026-05-27.

• Best MEV protection for everyday EVM users — MEV Blocker RPC (free, one-minute setup, drop-in MetaMask/Rabby).
• Best MEV protection for large trades ($10k+) — CowSwap (batch auctions) or 1inch Fusion (RFQ).
• Best MEV protection on Solana — Jupiter aggregator (uses Jito BAM bundles automatically).
• Best MEV protection for NFT mints — Flashbots Protect RPC + bundled tx.
• Best MEV protection on Arbitrum/Base/Optimism — Built-in (centralized sequencer ordering = no sandwich MEV today).
• Best MEV protection for sensitive transactions — Flashbots Protect RPC (routes via BuilderNet private mempool).
• Best MEV-aware DEX router — Uniswap app's MEV-aware routing or 1inch with Fusion enabled.
• Best MEV-redistribution setup — Opt into MEV-Share via Flashbots Protect for kickbacks (90% of extracted value by default).
• Best MEV protection for traders making 20+ swaps/day — Use a CEX for size; MEV is zero on internal CEX order books.
• Worst MEV exposure — Default Ethereum RPC + 5% slippage + large trade on a low-liquidity pair = guaranteed sandwich victim.

MEV on different chains

Ethereum L1 has the largest MEV market and the most mature mitigations. Arbitrum, Base, and Optimism have effectively zero sandwich MEV today (centralized sequencer). Solana now runs Jito BAM — an open block-building marketplace launched September 2025 — as its primary MEV infrastructure. Last verified: 2026-05-27.

L2s have effectively zero sandwich MEV today because the sequencer orders txs in first-come-first-served fashion and there is no public mempool for searchers to front-run from. Arbitrum reached Stage 1 decentralization (permissionless fraud proofs via BoLD) in 2025, but sequencer decentralization — the step that would recreate the L1 MEV market on the L2 — remains on the roadmap without a firm date. Optimism's Superchain approach targets sequencer decentralization in a future upgrade; no hard date as of May 2026.

Common misconceptions

"L2s have no MEV" — wrong, they have no sandwich MEV today because of centralized sequencing; arbitrage MEV still exists. "Private mempool = no MEV" — wrong, it just changes who extracts. "Flashbots is shady" — wrong, it's the most transparent layer in the stack. Last verified: 2026-05-27.

• "L2s have no MEV." L2s have no sandwich MEV today because the sequencer is centralized and there's no public mempool. Arbitrage MEV between L2 DEXs and across chains still exists and is captured by sequencer-affiliated bots.
• "Private mempool = no MEV." A private mempool means the public can't see your transaction. The bundler operating the private mempool still can. The trust assumption shifts from "the entire searcher market" to "this one bundler" — better, but not zero.
• "Flashbots is shady." Flashbots is the most transparent layer in the MEV stack. Builder bids, relay logs, and bundle inclusion are all publicly observable. Since December 2024, Flashbots no longer runs centralized block builders — all traffic routes through BuilderNet, a jointly-operated network including Nethermind and others.
• "MEV is theft." Sandwich MEV is extraction within the rules of a permissionless ordering market; you can lose nothing if your tx is never broadcast publicly. Arbitrage MEV is a competitive market-making function that benefits price-takers.
• "Higher slippage tolerance helps a swap go through." It also raises your maximum extractable loss. Bots size their sandwiches to exactly your slippage tolerance — set 3%, you can lose up to ~3%; set 0.5%, you can only lose up to ~0.5%.

MEV is becoming "redistribution," not prevention

Order-flow auctions (CowSwap, UniswapX), MEV-Share kickbacks (Flashbots, 90% default), and Jito BAM tips (Solana) increasingly route extracted value back to the user submitting the transaction. The endgame is a market where users are paid for the MEV their order flow creates. Last verified: 2026-05-27.

The 2024–2026 MEV trend is redistributing MEV back to users rather than blocking it:

• Order-flow auctions (CowSwap, 1inch Fusion, UniswapX) — solvers bid to fill orders, paying users for the right to capture small MEV. The user gets the best fill; the solver pockets the spread.
• MEV-Share (Flashbots) — user opts to share their tx data hints with searchers; if MEV is extracted from the hint, the user gets 90% of extracted value by default. Both MEV Blocker and Flashbots Protect return 100% of refunds to users.
• Jito BAM tips (Solana) — users tip block leaders for priority inclusion; BAM's TEE-based scheduler nodes distribute ordering fairly. Jito validator client covers ~95% of Solana stake.
• BuilderNet (Flashbots, December 2024) — a jointly-operated decentralized block-building network replacing Flashbots' centralized builders, with gas refunds routed to order-flow originators.
• SUAVE (SUAVE is Flashbots' proposed decentralized MEV marketplace and block-building network) — Flashbots' proposed decentralized MEV marketplace. BuilderNet is the practical production step; SUAVE remains a research-stage vision without a mainnet deployment timeline.

Direction: in 2027+, expect most retail MEV to be auctioned back to the user submitting the transaction. The hidden tax becomes a visible rebate.

Where MEV applies (and where it doesn't)

• DEX swaps — sandwich and arbitrage MEV.
• NFT mints and sales — sniping MEV.
• DeFi liquidations — liquidation racing.
• Cross-chain bridges — rare; bridge MEV exists but is small.
• Pure transfers — almost no MEV (no profit opportunity).
• On-chain governance votes — no economic MEV (voting on a token result has no atomic profit hook unless tied to a parameter change).
• Most L2 user transactions — sequencer-ordered, no public mempool to attack.

Looking ahead

A few specific signals worth tracking through 2027:

• Glamsterdam and ePBS (EIP-7732) — the next Ethereum hard fork. First generalized devnet launched late April 2026; tentative mainnet target is mid-2026 (community docs cite June) but may slip to Q3/Q4. EIP-7732 enshrines PBS into the protocol, replacing MEV-Boost with an in-consensus block-building auction. Watch ethereum.org/roadmap/glamsterdam for testnet launches.
• L2 sequencer decentralization — Arbitrum achieved Stage 1 (permissionless fraud proofs via BoLD) but the sequencer itself remains centralized. When full decentralization ships it will recreate the L1 MEV market on each L2. Watch the Arbitrum governance forum and Optimism Superchain roadmap for sequencer decentralization specs.
• Commit-Boost adoption — a community-built, non-commercial validator sidecar fully compatible with MEV-Boost. Reached ~20% of Ethereum validators by August 2025, growing to ~35% by early 2026. Its adoption signals validator preference for a vendor-neutral sidecar standard.
• Jito BAM expansion — BAM launched mainnet September 25 2025 with a target of 30%+ of Solana stake. As more external builders join the marketplace, Solana's MEV extraction dynamics will approach Ethereum's PBS model in transparency.
• Encrypted mempools — threshold-encrypted mempools (Shutter Network, in a partnership with Primev for Ethereum L1) attempt to hide tx contents until inclusion. Shutter is live on Gnosis Chain; the Ethereum L1 out-of-protocol version is in development. Protocol-level encrypted mempool is on Ethereum's Arhat fork roadmap (estimated 2027). Adds latency and trust assumptions; large-scale deployment remains unproven.

Verdict

MEV is permanent on public blockchains, but a default MEV Blocker RPC plus Jupiter on Solana neutralizes the 1–2% sandwich tax for everyday traders. Large trades route through CowSwap or 1inch Fusion. Active traders should track MEV-Boost share and opt into MEV-Share for kickbacks. The infrastructure story in 2026: BuilderNet replacing Flashbots' centralized builders, Glamsterdam bringing ePBS on-chain, and Jito BAM maturing Solana's block-building market. Last verified: 2026-05-27.

For everyday users:

• Use MEV Blocker as your default Ethereum RPC.
• On Solana, use Jupiter (already MEV-aware via Jito BAM).
• For large trades (>$10k), use CowSwap or 1inch Fusion.
• Don't set high slippage on volatile pairs.

For active traders:

• Track MEV-Boost data (mevboost.pics) to understand block competition.
• Use Flashbots Protect for sensitive transactions.
• Opt into MEV-Share for kickbacks (90% of extracted value returned by default).
• Consider a CEX for very large trades — no MEV at all on internal order books.

MEV is a permanent feature of public blockchains. With five minutes of setup, you can keep it from costing you 1–2% per trade.

---

Related: Best DEXs 2026 · Best Crypto Wallets 2026