MEVDAO

Custom MEV Bot Development: What to Know Before You Hire

By Merunas Grincalaitis · · 5 min read · 0 views

I get a lot of messages that start with "I want an MEV bot, how much?" And I understand why. You've read that bots make money, you've got some capital, and you'd rather pay someone than spend a year learning Solidity, mempool mechanics and block building.

The problem is that "an MEV bot" isn't a product, any more than "a website" is. A sniper for new token launches, a liquidation bot for one lending protocol and a long-tail arbitrage bot are completely different systems. So before you hire anyone, me or anyone else, here's what you should know to get something that actually works and to avoid getting scammed on the way.

First: do you even need a custom bot?

Honestly, often not.

  • If you want to snipe launches, ready-made tools already do the hard parts: watching for the launch, building the bundle and sending it to builders. That's what the MEVDAO Sniper Bot is, and it's in the €200/month subscription with the sniping group and the Token Rescue Bot. Try that first before you pay for custom development.
  • If you want to rescue tokens from a hacked wallet, the Token Rescue Bot covers the common case. I explained how it works in how to rescue tokens from a compromised wallet.

Custom makes sense when you have a specific edge that no off-the-shelf tool covers: a strategy for one protocol, unusual order flow, a launch with special rules, or a combination of strategies that needs its own contract.

What a real MEV bot is made of

When people imagine an MEV bot, they think of the smart contract. That's maybe a fifth of the work. A production bot has roughly these parts:

  1. Data in. A node you control, or a very fast provider, with access to the mempool and new blocks. For serious strategies that means running your own node, often reth or another client, close to the builders.
  2. State. A local copy of what matters: pool reserves, borrower positions, token rules. It's updated every block, sometimes every pending transaction.
  3. Strategy. The logic that finds an opportunity and calculates the exact amounts. This is where your edge lives.
  4. Simulation. Every candidate gets simulated before it's sent. Fast bots use an embedded EVM like revm instead of making RPC calls, because a round-trip is too slow.
  5. The on-chain contract. Executes the trade atomically, checks profit and reverts if it isn't there. It has to be gas-efficient and secure, because it holds or moves money.
  6. Submission. Bundles sent to multiple builders, with a bidding strategy for how much to pay them.
  7. Monitoring and risk. Logs, alerts, profit tracking, max position sizes, and a kill switch.
  8. Key management. The hot key that signs transactions, kept separate from where the profits go.

A developer who only talks about part 5 hasn't built a bot that competes.

What to bring to the first call

The more specific you are, the faster and cheaper it gets. Before you message any developer, write down:

  • The strategy, in one sentence. "Backrun large swaps on these three pools." "Liquidate positions on this specific lending protocol." "Snipe launches with this kind of anti-bot contract."
  • Why you think it's profitable. Ideally with examples: transaction hashes where someone else made money doing it. If you can't find any, that's important information too.
  • Chain and venues. Ethereum mainnet only? Which DEXs or protocols?
  • Capital. How much you're willing to put in the contract, and how much you're willing to lose while you learn.
  • Infrastructure. Do you have servers and a node, or does that need to be part of the job?
  • Ownership. Do you need the full source code? (You should.) Who runs it after delivery?

If your idea comes from a how MEV bots make money type of article, that's fine as a start, but you need to narrow it down to one concrete strategy.

How a good project usually goes

Every project is different, but a sane process looks like this:

  1. Research. Look at historical data for your strategy. How often does the opportunity show up? Who wins it now? How much do they pay builders? This step sometimes ends the project, and that's a good result, because it's the cheapest way to find out an idea doesn't work.
  2. Prototype. A minimal version that detects opportunities and simulates them, without sending anything. You learn what it would have made.
  3. Fork testing. Run it against a mainnet fork with Anvil on replayed historical blocks.
  4. Small live test. Real money, small size, with strict limits.
  5. Iterate. Gas optimization, better bidding, more venues. This never really ends, because competitors adapt.

What it costs

I'll be transparent about my side: my rate is $5,000 an hour. That's not cheap, and it's not meant to be. You're paying for knowing which ideas will fail before you spend weeks building them, and for a bot that handles the edge cases that eat profits.

Whoever you hire, budget for:

  • Development, which depends heavily on the strategy. A simple, well-specified bot is a different job from a multi-strategy system with its own node.
  • Infrastructure. Servers, nodes or RPC plans, every month.
  • Capital and losses. The first weeks live are a learning period.
  • Maintenance. Protocols upgrade, competitors change, and strategies decay. A bot nobody maintains stops making money.

Red flags

This industry is full of scams, and "MEV bot" is one of the favourite keywords. Leave the conversation if you see any of these:

  • Guaranteed returns. Nobody who actually runs MEV can guarantee profit. Competition alone makes that impossible.
  • "Deploy this contract and send ETH to it." This is the most common scam. YouTube tutorials share a "bot" contract for Remix, and when you fund it, the ETH goes to the scammer. SentinelOne documented a series of these scams that drained more than $900,000 from victims.
  • They want your private key or seed phrase. A developer never needs your main wallet's key. The bot gets its own fresh key, and you fund it with what you're willing to risk.
  • No code ownership. If you pay for a custom bot, you should get the source, or at least be able to audit the contract that holds your money.
  • They can't show you anything on-chain. Ask for transactions from bots they built, or at least a detailed explanation of how they'd approach your strategy. Real developers can talk about specifics.
  • "Profit sharing only, but first a setup fee." Usually a scam.

What I can build for you

Custom work I can help with includes snipers for launches with unusual rules, liquidation bots, long-tail arbitrage, rescues for complicated compromised wallets, and trading bots outside of MEV, like the Polymarket bots I wrote about. You can also read how a same-block sniper bot works to get an idea of the details involved.

If you have a strategy in mind, send me a message on Telegram with the one-sentence description and any example transactions. I'll tell you honestly whether I think it can work. And if you're not ready for custom yet, start with the MEVDAO subscription, 3 days free, and join the MEV Secrets email list right below.

← All articles