MEV Strategies in 2026: The Binance Hack Feeding Frenzy

On the evening of October 29, 2026, someone emptied roughly $6.3 billion out of Binance’s hot wallets in under four hours. Blockchain forensics firm Elliptic called it the largest theft in the history of crypto, around four times the size of the roughly $1.5 billion Bybit hack, and, like TRM Labs, attributed it to North Korea’s TraderTraitor group.

The theft itself is not what this article is about. It is about what happened in the minutes after, while the stolen coins were still in motion. To cash out before issuers could freeze the funds, the attacker dumped hundreds of millions of dollars of stolen stablecoins into decentralized exchanges, and a swarm of automated bots pounced. They bought the discounted coins, arbitraged the price gaps the fire sale tore open across venues, sandwiched the attacker’s own swaps, and paid Ethereum’s block builders a fortune to go first. According to on-chain analysis by Bitquery, the attacker lost about $119 million to slippage and swap costs, and roughly $60 million of that landed with arbitrage traders.

That is Maximal Extractable Value, or MEV, running at full throttle, in public, on the worst possible night. What follows is a field guide to the strategies those bots run, the hidden supply chain that powers them, what they quietly cost ordinary traders, and how, if at all, you can defend yourself.

The night the dark forest woke up

The sequence was brutally fast. Between 20:04 and 23:56 UTC, Binance hot wallets on nine blockchains (Ethereum, Bitcoin, Tron, the XRP Ledger, Solana, Arbitrum, Polygon, Optimism and Base) were drained, according to Bitquery’s on-chain reconstruction. Around $729 million of the stolen USDT was swapped into ETH and TRX before Tether’s freezes could land, at roughly 16 percent slippage through intent-based routers like UniswapX and 1inch Fusion. A thief in that much of a hurry, selling that much into automated markets, is a gift to anyone watching the chain.

Plenty of people were watching. On the night of the hack, MEV bots paid Ethereum’s block builders 10,251 ETH in tips, against just 4.6 ETH over the same hours the night before, per Bitquery’s who-profited analysis. Builders and validators collected about $30 million in tips and fees from the frenzy; in one arbitrage window around a single token, bots paid $2.67 million in direct builder bribes across 900 transactions plus another $241,000 in tips. Gas on individual transactions ran $1,000 to $1,500, with at least one burst costing $4,000. The notorious sandwich bot jaredfromsubway even ran a sandwich on the attacker’s own sells and booked a tidy $157,000.

Paradigm researchers Dan Robinson and Georgios Konstantopoulos gave this environment its name back in August 2020, describing the public mempool as a dark forest: a place where predatory bots watch every unconfirmed transaction and devour anything profitable the instant it appears. For most of 2026 the forest had grown quiet, as private order flow pulled transactions out of public view. The Binance hack briefly set the whole thing ablaze.

It is worth stressing what the breach was and was not. It was not a cracked private key or a buggy smart contract in the ordinary sense; investigators traced it to a compromise of Binance’s hot-wallet signing infrastructure, part of a broader run of operational and wallet attacks across the month that we covered in our look at the Bitget breach. The on-chain response was swift: Tether froze 1.76 billion USDT, Circle froze 823 million USDC, the Arbitrum Security Council locked attacker funds, and a handful of other issuers froze the rest, roughly $2.7 billion in all. What the bots scooped up in the gaps, nobody is getting back.

MEV in one minute

Strip away the jargon and MEV is simple. As the Ethereum documentation puts it, Maximal Extractable Value is the most value that can be squeezed out of producing a block, over and above the standard block reward and gas fees, by adding, dropping, or reordering the transactions inside it. Whoever builds the block holds that power, and they sell it to the highest bidder.

The term was coined in a 2019 paper memorably titled Flash Boys 2.0, by Phil Daian and co-authors, who warned that the race to capture ordering profits could destabilize consensus itself. Back then the M stood for miner. After Ethereum’s 2022 Merge handed block production to proof-of-stake validators, the industry quietly relabeled it maximal, and the acronym stuck.

The numbers are not small. Cumulative MEV extracted on Ethereum has passed $1.2 billion, and more than half of that total has come from sandwich attacks on ordinary traders, according to industry trackers. The core mechanic never changes: on a public blockchain, your pending transaction is visible to everyone before it confirms, and whoever orders the block decides who goes first. Visibility plus discretionary ordering is the raw material. The Binance hack simply delivered an unusually large, unusually fast pile of it in a single night.

A field guide to MEV strategies

MEV is not one thing. It is a family of strategies, some of which keep markets honest and some of which are a straight transfer from your wallet to a bot’s. The table below is the map; the sections after it walk the territory. Keep one distinction in mind throughout: a strategy is benign when the value it captures would otherwise be left on the table (a mispriced pool, an underwater loan), and toxic when it manufactures a loss for a specific user who did nothing wrong.

Arbitrage: the benign backbone that ate the hack

Arbitrage is the oldest trade in finance: buy low in one place, sell high in another. On-chain, it splits into two flavors. Atomic arbitrage happens entirely in one transaction, often across two or three DEX pools, and is effectively risk-free because the whole thing reverts if it is not profitable. Non-atomic arbitrage, which spans a centralized exchange and an on-chain pool, carries inventory risk because the two legs do not settle together. That second flavor, CEX-to-DEX arbitrage, is the largest and least visible form of MEV, run by a small number of sophisticated firms with low-latency connections to both worlds.

The Binance hack turned into the single biggest arbitrage event of the year precisely because the attacker behaved like a desperate, price-insensitive seller. Dumping $729 million of stolen USDT into DEX routers at speed pushed on-chain prices far out of line with the rest of the market; one widely cited reconstruction found WETH trading well above its true price against the stolen stablecoins, and smaller tokens like TRON’s TRX and the token OGN dislocating even harder. Every one of those gaps was an arbitrage invitation, and bots took it, buying the cheap side and selling the expensive side until the prices snapped back. Bitquery’s accounting puts the traders’ net haul at roughly $60 million after gas and tips.

Here is the uncomfortable nuance: most of that activity was, mechanically, the good kind of MEV. Arbitrage is what keeps a token worth the same on fifty venues at once; without it, DEX prices would drift and every quote you see would be stale. The bots did not cause the theft and did not worsen anyone’s position except the thief’s. They simply did, at enormous scale and in a few frantic hours, what they do quietly every single day.

Sandwich attacks: the toxic corner

If arbitrage is the acceptable face of MEV, the sandwich is the reason the word has a bad reputation. The mechanic is predatory and precise. A bot spots your pending swap in the public mempool, buys the same asset just ahead of you (pushing the price up), lets your trade execute at that worsened price, then sells immediately after for a profit. You are the filling. You still get your tokens, but you get fewer of them, and the gap is the bot’s margin. Unlike arbitrage, a sandwich creates no efficiency; it is a pure, engineered loss imposed on someone who simply wanted to trade.

One bot has dominated this corner for years: jaredfromsubway. By widely cited on-chain estimates it has accounted for a majority of all sandwich activity on Ethereum at its peak, pulling tens of millions of dollars a year from ordinary traders. Its targets are nobody’s idea of fair game: in April 2026 it even sandwiched a roughly $4 token swap by Ethereum co-founder Vitalik Buterin, wrapping it in over a million dollars of volume, as CoinDesk reported. On Binance-hack night, it turned the same playbook on the thief.

There is a coda worth savoring. In June 2026 the hunter became the hunted: an attacker fed jaredfromsubway a stream of fake, profitable-looking opportunities over several weeks, tricking the bot into granting token approvals to malicious contracts, then drained roughly $7.5 million from it in a single counter-exploit, as Chainalysis documented. The most feared sandwich bot on Ethereum had been sandwiched by its own greed. The lesson generalizes: the strategies in this family are an arms race, not a steady business, and the edge belongs to whoever is one trick ahead.

Liquidations, backrunning, JIT, and the long tail

Between the benign and the toxic sits a broad gray middle, where most day-to-day MEV actually lives.

Liquidations. When a loan on Aave, Compound, or a perp venue falls below its collateral threshold, bots race to repay it and claim the protocol’s liquidation bonus. This is written into the rules and it keeps lending markets solvent, so the outcome is healthy; the competition to win it, down to the ordering of a single block, is textbook MEV.

Backrunning. Placing a transaction immediately after a known one, to capture the opportunity it creates without touching the original user, is the gentlest strategy of the lot. It harms no one directly, and it is the engine behind the order-flow auctions we will meet in the defenses section, which turn backrunning into a rebate for the person whose trade created the opportunity.

JIT liquidity. A just-in-time liquidity provider adds a large, tightly concentrated position right before a big swap and withdraws it immediately after, capturing most of that swap’s fee while bearing almost no risk. It tightens pricing for the one big trade but quietly siphons fees away from the passive liquidity providers who were there all along.

The long tail and OEV. Token launches, NFT mints, and thinly traded pairs all generate their own scramble for first position. A newer sibling, oracle extractable value, has bots backrunning price-feed updates to win the liquidations those updates trigger; a growing number of protocols now auction that right off and recycle the proceeds, turning a leak into revenue.

Cross-domain and multi-chain MEV

The Binance hack was a nine-chain event, and that is the clue to where the frontier of MEV has moved. Cross-domain MEV chases opportunities that span multiple chains, rollups, and centralized venues at once. It is harder and riskier than single-block arbitrage because there is no atomicity across domains; a bot can land the first leg on Ethereum and then watch the second leg on Solana or a CEX move against it. That risk is exactly why the professionals dominate it. During the hack, market makers were reportedly buying the stolen USDT within minutes and hedging the exposure on venues like Hyperliquid, running a cross-domain book that spot-only bots could not touch.

Layer-2 networks add another wrinkle. On most rollups a single sequencer decides transaction order, so MEV migrates from a public mempool to that sequencer. Arbitrum has gone as far as auctioning an express lane through its Timeboost mechanism, selling a brief head start to the highest bidder every minute. The strategies are the same ones in the table above; they have simply relocated to wherever ordering is decided, which is increasingly not the open Ethereum mempool at all.

The supply chain: searchers, builders, relays, proposers

Behind every extracted dollar is an assembly line, and understanding it is the difference between seeing MEV as random bot chaos and seeing it as the industry it has become. Ethereum runs on a design called proposer-builder separation, and value flows along it in four steps.

• Searchers scan the chain for opportunities and bundle transactions that capture them.
• Builders take those bundles, assemble a full, maximally profitable block, and bid for the right to have it proposed.
• Relays sit between builders and validators, holding the block so the validator can commit to it without seeing (and stealing) its contents.
• Proposers (the validators) pick the most valuable block offered and add it to the chain, pocketing the bid.

Most of this happens through software called MEV-Boost, which runs the sealed-bid auction that now assembles roughly 90 percent of Ethereum blocks. On a normal day the machinery is invisible. The Binance hack made it briefly, vividly legible: searchers found the dislocations, builders packed them into blocks, and the bids flowing up the chain spiked to that 10,251 ETH in tips, with $2.67 million in direct bribes passing through in a single token’s arbitrage window. The feeding frenzy was not a breakdown of the system; it was the system working exactly as designed, at full tilt.

The builder oligopoly

Here is the part that keeps researchers up at night. The block-building that captures MEV has consolidated into a handful of firms. Over a recent 24-hour window tracked by relayscan.io, a single builder, Titan, produced just over half of all Ethereum blocks, and the top three built roughly 86 percent between them. The relay layer is just as concentrated, with two relays handling about two thirds of delivered blocks.

Concentration this tight is a censorship risk, not just a competition problem. A small group of builders decides what goes into blocks, and after the 2022 Tornado Cash sanctions some relays began filtering sanctioned addresses, so the share of blocks passing through such compliance-minded relays is now tracked publicly. The most prominent response is BuilderNet, a collaboration between Flashbots, Beaverbuild, and Nethermind that runs block-building across multiple operators inside secure hardware enclaves and refunds MEV back to the users and apps that generate the order flow, rather than letting a single integrated firm keep it. Whether a decentralizing project can out-compete the incumbents on raw profitability is the open question of the next year.

The invisible tax on ordinary traders

Spectacular hack-night numbers obscure the real story, which is mundane and constant. For a normal person swapping tokens, MEV is a small, silent toll paid on a huge share of trades. Researchers have taken to calling it crypto’s version of payment for order flow, the controversial practice that lets stock brokers monetize retail trades. One academic estimate found that the visible swap fee is only around 28 to 36 percent of a DEX trade’s true all-in cost; the rest is implicit, and MEV is a big part of it. You never see a line item, because the loss is baked into the price you receive.

That is why execution quality, not headline fees, is where retail traders actually win or lose, a point we made at length in our breakdown of what a perp DEX trade really costs. The same logic governs a plain spot swap. None of this is helped by a jittery market, either: in the days around the hack, ETH was changing hands near $2,240 and SOL around $96, both down roughly 6 percent on the week, per CoinGecko, the kind of volatility that widens spreads and fattens every sandwich.

How to defend your trades

You cannot stop arbitrageurs from correcting prices, and you should not want to. But you can make yourself a very hard target for the toxic strategies, and the tooling in 2026 is genuinely good. Defense comes in three layers, summarized below and then explained.

Keep your trade out of the public mempool. This is the single highest-impact move, because a transaction a bot cannot see is a transaction it cannot sandwich. A private RPC endpoint routes your swap directly to block builders. MEV Blocker is free, splits any backrun value 90 percent to the user and 10 percent to the validator, and reports having protected well over $200 billion in volume. Flashbots Protect similarly hides your transaction and, through MEV-Share, returns about 90 percent of any eligible MEV to you. Both are a one-time change of RPC URL in your wallet.

Trade through intents. Instead of broadcasting a specific transaction, you sign what you want (sell X, get at least Y) and let competing solvers find the fill. CoW Swap, which settled more than $23 billion in 2026, clears trades in batch auctions at a single uniform price, a structure that makes sandwiching essentially impossible because there is no in-between price for a bot to exploit. UniswapX runs a Dutch auction to similar effect. The tradeoff is that you are trusting a solver network to get you the best outcome, but the sandwich surface simply disappears.

Tune your settings, and mind your custody. Set slippage tolerance as tight as the market allows and use short deadlines, so there is little room for a bot to operate and stale transactions expire. And remember the night’s broader lesson: funds sitting on an exchange carry a different risk entirely, one that no slippage setting can fix, which is the whole argument for self-custody that we weighed in our hardware-wallet comparison and in our coverage of October’s wallet-security wave.

Solana’s separate forest

Solana, one of the chains drained in the hack, plays a different game. It has no global public mempool, so the classic Ethereum sandwich, which depends on watching pending transactions, has a smaller surface. Instead, most Solana MEV flows through Jito, whose client runs on the large majority of the network’s stake and hosts an off-chain auction where searchers submit bundles and tip for priority, as Chainstack explains. The JitoSOL liquid-staking token then recycles those tips back to stakers, turning MEV into yield.

Solana has also been more willing to police the toxic end directly. In 2024 the Solana Foundation removed more than 30 validators from its delegation program for running private sandwich mempools that preyed on retail users. Tim Garcia, who leads validator relations at the foundation, described the purge as final, a line in the sand that Ethereum, with its credibly neutral and permissionless block market, has never been able to draw. It is a genuine philosophical split: discretionary enforcement by a foundation, versus protocol rules that treat every bidder the same.

Is any of this legal?

The honest answer, in the United States, is that nobody quite knows yet. The landmark test is the case of brothers Anton and James Peraire-Bueno, charged in the Southern District of New York with wire fraud and money-laundering conspiracy over a roughly $25 million exploit of a MEV-Boost relay in April 2023. It was billed as the first criminal MEV case, and it did not go cleanly: the first trial ended in a mistrial in late 2025 after the jury deadlocked, prosecutors moved for a retrial, and as of late October 2026 no verdict had been reached. The defense leaned on a prior precedent: Avraham Eisenberg’s $110 million Mango Markets manipulation convictions were vacated in May 2025, a judge finding insufficient evidence of misrepresentation given the protocol’s permissionless design.

The deeper problem is jurisdictional. The SEC’s anti-fraud Rule 10b-5 and the CFTC’s parallel rules can reach a registered intermediary that front-runs its own clients, but an anonymous bot on a permissionless DEX is nobody’s client and sits outside that authorized perimeter, a gap we traced in detail in our piece on SEC crypto enforcement after the comment window. The same ambiguity cuts both ways. The very technique that sandwiches a trader can also rescue one: in late October an MEV bot front-ran a $7.45 million exploit of an Avalanche Safe wallet, seizing the assets before the thief could and then entering bounty talks to return them. Predator or whitehat depends entirely on intent, which is precisely what the law has not figured out how to read on-chain.

The fix on the horizon: ePBS, FOCIL, and encrypted mempools

Ethereum’s longer answer is to change the plumbing so that extraction is less concentrated and less damaging. The headline move is enshrined proposer-builder separation, or ePBS, specified in EIP-7732 and set as the marquee feature of the Glamsterdam upgrade. According to the Ethereum roadmap, ePBS folds the proposer-builder relationship directly into the protocol, removing the trusted third-party relays that MEV-Boost depends on today, establishing trustless in-protocol builder payments, and widening the block propagation window from about two seconds to roughly nine. As of late October 2026 it was still being tested on testnets, with a Hoodi fork scheduled and a mainnet date expected in the fourth quarter of 2026 but not yet confirmed.

Even its champions warn it is not enough on its own. In March 2026, Vitalik Buterin laid out a plan arguing that ePBS must be paired with two further ideas: FOCIL, a scheme of committee-chosen inclusion lists that forces blocks to carry transactions builders might otherwise censor, and encrypted mempools, which hide transaction contents until after ordering is fixed. That last one is the structural kill-shot for sandwiching: a builder who cannot read your trade cannot possibly wrap it. None of it touches the hardest cases, though. Cross-exchange arbitrage and cross-domain MEV live partly off-chain, and no single network’s upgrade reaches them. The dark forest will get safer for the average swap; it will not disappear.

Frequently Asked Questions

What is MEV in crypto?

MEV, or Maximal Extractable Value, is the profit a block producer or a bot can capture by choosing which transactions go into a block and in what order. Common strategies include arbitrage, liquidations, and sandwich attacks. It is often called crypto’s invisible tax because the cost is usually hidden inside a trade’s execution price.

Did MEV bots really profit from the Binance hack?

Yes. As the attacker dumped stolen stablecoins into decentralized exchanges on October 29, 2026, arbitrage bots bought the discounted coins and traded the resulting price gaps. On-chain analysis by Bitquery estimates the attacker lost about $119 million to slippage and swap costs, with roughly $60 million captured by arbitrage traders and about $30 million going to Ethereum builders and validators in tips and fees.

What is a sandwich attack, and how do I avoid it?

A sandwich attack is when a bot spots your pending swap in the public mempool, buys just before you to push the price up, lets your trade fill at the worse price, then sells right after for a profit. The most effective defense is to keep your transaction out of the public mempool using a private RPC such as Flashbots Protect or MEV Blocker, trade through an intent-based venue like CoW Swap, and set tight slippage limits.

Is MEV illegal?

In the United States it is largely unsettled. The first criminal MEV case, against the Peraire-Bueno brothers, ended in a mistrial in late 2025, and an earlier manipulation conviction in the Mango Markets case was later vacated. The SEC’s Rule 10b-5 and the CFTC’s rules can reach a registered firm that front-runs its clients, but anonymous bots on permissionless exchanges sit outside that perimeter.

Will Ethereum upgrades like ePBS stop MEV?

Not entirely. Enshrined proposer-builder separation (EIP-7732), the headliner of the Glamsterdam upgrade, removes the trusted relay middleware and is expected on mainnet in the fourth quarter of 2026, but Vitalik Buterin has argued it must be paired with inclusion lists and encrypted mempools to curb censorship and sandwiching. Even then, cross-exchange arbitrage would survive.

Yuki Tanaka covers DeFi and on-chain market structure for HOGE Wire.