CEX vs DEX Arbitrage: Where the Bigger Gaps Hide and Why

CEX-DEX arbitrage trades the same asset between a centralised exchange and an on-chain pool when their prices differ. It can show larger gaps than pure on-chain arbitrage, but needs funds on both sides, fast data and acceptance of custody and counterparty risk. Fees and timing decide whether the gap pays.

Why the two venues diverge

Centralised exchanges match orders on an order book, and their prices react to every bid and ask within milliseconds. On-chain pools update only when someone trades against them, and blocks arrive roughly every twelve seconds. When news or large orders move a centralised price, the on-chain pool keeps quoting the old price until an arbitrageur trades it into line. That lag is the classic CEX-DEX opportunity.

The ethereum.org MEV overview discusses how traders compete to capture value from such price differences across venues and how that competition is organised around block production. What matters for you is that the side of the trade on-chain is subject to the same competition and gas costs described elsewhere on this site, while the exchange side introduces new constraints.

Comparing the three arbitrage settings

Feature	DEX-DEX, atomic	CEX-DEX	CEX-CEX
Settlement	One transaction, all-or-nothing	Two separate systems, no atomicity	Two exchange accounts
Capital needed	Little with flash loans	Pre-funded on both sides	Pre-funded on both sides
Main costs	Gas, pool fees, slippage	Exchange fees, gas, pool fees, rebalancing	Exchange fees, withdrawals
Key risk	Failed or outbid transaction	Price moves between legs, custody	Withdrawal delays, counterparty
Speed bottleneck	Block inclusion	Block inclusion and API latency	API latency

Key takeaway: CEX-DEX gaps are wider partly because they are harder to close, and the difficulty is what you are paid for.

The practical hurdles

No atomic settlement. On-chain, both legs of a DEX-DEX trade can happen in one transaction. With an exchange and a pool there is no such guarantee. You might sell on the exchange and then fail to buy on-chain, leaving an unhedged position. Bots manage this with pre-funded balances and by hedging the leg that executes first.

Inventory and rebalancing. If you keep buying on the exchange and selling on-chain, your inventory drifts. Eventually you must move assets back, which costs withdrawal fees and time, and during the transfer you are exposed to price changes. A sensible plan includes rebalancing rules and a budget for those costs.

Custody and counterparty risk. Funds held on an exchange are subject to the exchange’s solvency and policies. Funds in a hot wallet used by a bot are subject to key security and contract risk. Neither is zero.

API and data latency. Exchange feeds, rate limits and outages can leave a bot trading on stale prices. On the chain side, RPC provider quality matters too.

Rules and compliance. Exchange accounts come with identity verification, terms of service and, depending on jurisdiction, reporting obligations. See whether crypto arbitrage is legal for a general overview.

What the on-chain leg looks like

The on-chain half is ordinary DEX trading and shares the costs described in how DEX arbitrage works: pool fees, price impact and gas. Because the other half is off-chain, flash loans do not help much with the exchange leg, since the exchange will not lend inside your transaction. A flash loan can still fund the on-chain leg in some designs, as explained in flash loan arbitrage explained, but the exchange side must be pre-funded.

Private routing remains relevant for the on-chain leg because a pending swap can be copied or sandwiched. The approach is described in Flashbots Protect and arbitrage.

Where bots specialising in Ethereum fit

Some commercial bots target on-chain venues only. CryptoMEV Bot’s Ethereum arbitrage page describes scanning decentralised exchanges, simulating, and sending through Flashbots, and it does not state support for centralised exchange accounts on that page. A separate page covers Binance arbitrage for a different setting. If you want to run CEX-DEX strategies, confirm directly with any vendor what is supported, because the integration is more complex than reading pool data.

A worked example of the inventory problem

Suppose a bot repeatedly finds that a token is cheaper on-chain than on an exchange. Each cycle it buys on-chain and sells on the exchange. After several cycles the wallet has run out of the quote asset it uses to buy, while the exchange account has accumulated the quote asset from the sales and has run out of tokens. To continue, funds must move: tokens from the wallet side are gone, so the operator has to withdraw quote asset from the exchange back to the chain, pay a withdrawal fee, wait for confirmations and accept price risk in the meantime.

This is why serious operators model inventory as a first-class variable. They set target balances on each side, widen the required gap as a side gets depleted so that the bot favours trades that restore balance, and budget for periodic rebalancing. A strategy whose gross profit per trade looks healthy can have poor net results once rebalancing costs are spread across trades.

The tooling question follows from this. A bot designed for on-chain venues, such as the product described on CryptoMEV Bot’s Ethereum arbitrage page, handles the on-chain side. Handling exchange balances, hedging and withdrawals needs additional integration and risk logic, and you should not assume it is included. Ask directly. The vendor’s separate Binance arbitrage page is the place to start for exchange-related claims, and the same rule applies there: read the features and the disclaimer, then test with small funds.

Should you attempt it?

CEX-DEX arbitrage tends to suit teams with engineering resources and capital already positioned on both sides. For a newcomer, the combination of custody risk, rebalancing and latency often makes pure on-chain strategies easier to reason about, even though they are more crowded. Whichever you choose, start with small size and track net results after all costs, as laid out in whether Ethereum arbitrage is still profitable.

Checklist for a CEX-DEX setup

• Pre-fund exchange and wallet with amounts you can afford to lose.
• Use exchange API keys with trade-only permissions and no withdrawal rights.
• Set limits on position drift and a rule for hedging a half-filled trade.
• Measure feed latency on both sides and discard stale quotes.
• Budget for withdrawal and rebalancing costs.
• Record every leg so you can reconcile profit and loss afterwards.

Summary

CEX-DEX arbitrage exploits the lag between fast order books and slower pools. Gaps can be bigger, but atomicity is lost and you take on inventory, custody and latency risks. Compare the real cost stack before committing, and check current rules in your jurisdiction. This is not financial advice.