What is a MEV Bot?

How MEV bots work on Ethereum

Every MEV bot runs the same four-stage loop, regardless of strategy:

1. Mempool monitoring. The bot subscribes to pending transactions through a low-latency Ethereum node (Geth, Erigon, or Reth) or a specialized mempool feed like BloXroute or Eden. Every pending DEX swap is decoded in real time — token in, token out, amount, slippage tolerance, deadline.
2. Opportunity detection. The bot simulates each pending transaction against current pool state to determine whether sandwiching, arbitraging, or providing just-in-time liquidity around it would be profitable after gas + builder bribe.
3. Bundle construction. When an opportunity clears the profit threshold, the bot crafts an atomic transaction bundle — front-run + victim + back-run for sandwiches, flash-loan + multi-hop swap for arbitrage — signed with the operator's key.
4. Bundle submission. The bundle goes to block builders via Flashbots Relay, Beaverbuild, Titan, rsync-builder, BloXroute Max, and Builder0x69. Each bundle includes a coinbase transfer — a direct ETH tip to whichever builder includes it. The bundle either lands atomically in the next block or doesn't land at all (no partial fills, no exposed front-runs).

The full cycle from mempool detection to bundle submission happens in under 100 milliseconds for competitive bots. That latency budget is why MEV is an infrastructure game more than a code game — you need direct peer connections, co-located hardware, and relationships with builders.

The three MEV strategies: sandwich, arbitrage, JIT liquidity

Sandwich attacks

A sandwich attack wraps a pending DEX trade with a front-run (buy before the victim) and a back-run (sell after the victim). The victim's trade pushes the price up; the back-run sells into the elevated price for a profit. Sophisticated bots like JaredFromSubway run multi-layer sandwiches — 5-layer or 7-layer attacks that trap multiple victims and execute multiple price manipulations in a single bundle, dramatically improving capital efficiency.

Arbitrage

An arbitrage bot finds price differences between DEXs (Uniswap vs SushiSwap vs Curve), CEXs and DEXs, or across chains. The bot buys low on one venue, sells high on another — atomically when possible. The most advanced version is flash-loan arbitrage: borrow capital from Aave or Balancer, execute the entire arbitrage, repay the loan, and keep the spread — all in one transaction with zero upfront capital.

JIT liquidity

Just-in-time liquidity is a Uniswap V3-specific strategy where the MEV bot adds concentrated liquidity around a pending large swap, collects the LP fees on that swap, then removes the liquidity immediately after. JIT is strictly more profitable than sandwiching for large trades because the LP fee captured is bigger than the price-manipulation profit a sandwich could extract — and JIT doesn't harm the victim (their trade still executes at their expected rate).

Liquidation

Lending protocols like Aave V2/V3, Compound V2/V3, and MakerDAO let users borrow against collateral. When the value of that collateral drops and a position becomes undercollateralized (health factor below 1.0), anyone can trigger a liquidation — repay part of the debt and receive the collateral at a 5–15% discount. MEV bots monitor open positions across all major lending protocols, detect the moment a position crosses the threshold, fund the repayment with a flash loan, call the protocol's liquidation function, and capture the bonus. Like flash-loan arbitrage, liquidations are zero-capital strategies — the flash loan provides the repayment funds and is repaid atomically inside the same transaction. The JaredETH bot runs liquidation alongside sandwich and flash-loan arbitrage as a third independent strategy.

How MEV bots interact with the mempool

The Ethereum mempool is the global pool of pending transactions waiting to be included in a block. Every node maintains its own view of the mempool, and MEV bots subscribe to one (or many) for low-latency visibility. A typical mempool sees 30,000–80,000 pending transactions at any moment, with several thousand DEX swaps per minute during peak hours.

MEV bots decode each pending DEX swap by inspecting its calldata against known router ABIs — Uniswap V2 Router, Uniswap V3 SwapRouter02, SushiSwap RouterProcessor3, 1inch AggregationRouterV5. From the decoded swap, the bot extracts the token path, amount in, minimum amount out, deadline, and the victim's effective slippage tolerance. That slippage number is the most important variable: it determines how much the bot can move the price in its front-run without causing the victim's trade to revert.

Increasingly, sophisticated users protect themselves with private mempools — Flashbots Protect, MEV-Share, CowSwap, 1inch Fusion — where their transactions never hit the public mempool and can't be sandwiched. MEV bots watch for these protected transactions and skip them; trying to sandwich a private-flow transaction is wasted gas.

Flashbots, block builders, and the MEV supply chain

Since the Ethereum merge in September 2022, blocks are produced through a proposer-builder separation model. Validators (proposers) outsource block construction to specialized block builders — Beaverbuild, Titan, rsync-builder, BloXroute Max, Builder0x69, Flashbots. Builders compete to produce the most valuable block; the validator picks the highest-paying builder's block.

MEV bots submit their bundles directly to builders through their respective relays. The bundle includes a coinbase transfer — an ETH payment from the bot's contract to whichever builder includes the bundle. The size of this tip determines inclusion priority. Most MEV bots tip 90–97% of their expected profit; JaredFromSubway tips up to 99.9% — keeping only 0.1% per attack. The high bribe rate strongly favors inclusion against competing bots.

Builder relationships matter as much as code quality. The top three builders (Beaverbuild, Titan, Builder0x69) collectively produce over 80% of Ethereum blocks at most times. A MEV bot that submits to all six major relays with high tips gets included in roughly 60% of blocks at peak activity — that's the block presence stat you see in MEV dashboards.

JaredFromSubway — the most profitable MEV bot on Ethereum

JaredFromSubway (contract 0x1f2F…F387) is the most profitable MEV bot on Ethereum mainnet. Since March 2023 it has extracted 5,924 ETH in net profit across more than 40,000 sandwich bundles, all publicly verifiable on Etherscan.

JaredFromSubway leads on two fronts: it pioneered 5-layer and 7-layer sandwich techniques that competitors haven't replicated, and its 99.9% builder bribe rate outbids every alternative on inclusion priority. The full breakdown is on the JaredFromSubway page.

MEV bot GitHub repos — why open source doesn't win

Search GitHub for "mev bot" and you'll find hundreds of repositories — rusty-sando, mev-template-rs, subway-rs, sandwich-bot-py. Most of them are educational. None of them generate sustained profit. The reasons are structural:

• Latency. Public MEV repos use generic RPC providers. Profitable MEV requires sub-20ms latency to builders, which means co-located hardware and dedicated peer connections.
• Builder relationships. Top builders throttle untrusted senders. Established bots have direct lines to builder ops teams; GitHub forks don't.
• Bribe ceilings. Open-source bots typically cap bribes at 80–90% of profit. Production bots go 99%+. You can't out-tip the market with a 90% strategy.
• Strategy depth. Public repos do basic 3-layer sandwich. Production bots do swap-JIT, 5/7-layer sandwiches, dual-purpose transactions, and adversarial bundles that the public has never seen documented.

We cover the build-vs-buy decision in detail on the build a MEV bot guide and document why public repos fail at length on MEV bot GitHub explained.

Is running a MEV bot profitable in 2026?

Yes, for operators with infrastructure and builder relationships. The reference deployment behind JaredETH has extracted 5,900+ ETH in net profit since March 2023.

For new entrants building from scratch, the bar is high. The competitive bots have a three-year head start on optimization and an entrenched relationship with every major block builder. New entrants are better off either licensing an existing bot (1 Year License or Full Source Code) or specializing in a niche that the incumbents haven't saturated — Layer 2 MEV on Base or Arbitrum, NFT MEV, or governance MEV on specific protocols.

MEV bots vs traditional HFT and algorithmic trading bots

A MEV bot is the on-chain equivalent of a high-frequency trading (HFT) bot. Both extract value from price discovery mechanics by acting faster than human or generic counterparties. A traditional HFT firm co-locates servers next to NASDAQ's matching engine to shave microseconds off order routing; an Ethereum MEV bot co-locates against block builders to land bundles before a competitor's alternative bundle. Same fundamental edge — latency, infrastructure, and proprietary strategy — just a different venue.

The differences matter for anyone evaluating MEV vs other algorithmic trading bot categories:

• Venue. HFT runs on regulated exchanges (NYSE, NASDAQ, CME, Eurex). MEV bots run on permissionless DEXs (Uniswap, Curve, Balancer) and the public mempool — no broker, no exchange membership required.
• Capital requirements. Wall Street HFT shops run $50M–$500M of working capital. MEV bots with flash-loan integration run on near-zero balance sheet — Aave or Balancer fronts the capital for each attack and gets repaid atomically.
• Settlement risk. Traditional HFT carries T+1 settlement risk and clearing-house margin. MEV bundles settle atomically in the same block they execute — either the entire arbitrage succeeds or nothing happens (and you only pay gas if it landed).
• Strategy overlap. Many MEV strategies are direct translations of traditional algorithmic trading: cross-venue arbitrage, statistical arbitrage, market making (which is what JIT liquidity essentially is), and order-anticipation (front-running). The math is the same; the rails are different.
• Barrier to entry. Setting up an HFT operation requires exchange membership, regulatory licenses, and 7–8 figures of infrastructure. Running a MEV bot requires a server, a wallet, and the right software — and JaredETH ships the software so you can skip the multi-year build.

For anyone coming from a traditional finance background, MEV is the most accessible on-ramp into algorithmic trading on crypto rails. The strategies you already know (arbitrage, market making, latency-sensitive execution) all have direct DeFi analogs, and the regulatory overhead is dramatically lower.

How to verify a MEV bot on Etherscan

Every MEV bot's entire profit history is public. To verify a MEV bot on Etherscan:

1. Open the bot's contract address on Etherscan.
2. Check the Transactions tab for recent activity — competitive bots show new transactions every block or two.
3. Use the Analytics tab to see the ETH balance over time; a profitable bot trends up.
4. For sandwich confirmation, click into any transaction and look at the Logs tab — you'll see Transfer events for the front-run buy, victim swap, and back-run sell in sequence.
5. Cross-reference with Flashbots MEV-Inspect for an independent attribution of the bot's historical profit.

See the live JaredFromSubway profit dashboard on the Etherscan verification page.

Further reading and references

• Flash Boys 2.0 (Daian et al., 2019) — The original MEV paper that named the phenomenon and catalogued the on-chain attack surface.
• Flashbots documentation — Reference implementation for MEV-Boost, MEV-Share, and the relay infrastructure that production bots integrate with.
• Flashbots MEV-Inspect — Open-source MEV attribution pipeline that classifies sandwich, arbitrage, JIT, and liquidation transactions across all Ethereum operators.