How AMM Price Inefficiencies Create Arbitrage

AMM price inefficiencies are the engine behind most on-chain arbitrage. An automated market maker does not watch the market, read news, or react to sentiment — it applies a formula to its token balances every time someone trades. Between trades, the formula's output is frozen while the outside world moves on. That gap between a pool's formulaic price and the live market price is an inefficiency, and correcting it is literally the job arbitrageurs are paid to do through the spread they capture.

Understanding where these inefficiencies come from turns random gap-spotting into a repeatable edge. Some are mechanical — the unavoidable lag between a market move and the next trade against the pool. Some are structural — thin pools that overreact to every trade. And some are behavioral — large traders who accept terrible prices out of urgency or poor routing, leaving the pool badly mispriced in their wake. Each type behaves differently, lasts a different length of time, and demands a different response.

The Constant-Product Formula, Intuitively

The classic AMM rule is x * y = k: the product of the two token reserves stays constant. Imagine a pool holding 100 ETH and 300,000 USDC, implying roughly $3,000 per ETH. Buying 10 ETH removes ETH and adds USDC, forcing the price upward as the reserve shrinks — you might pay $3,050 on average instead of $3,000. Price impact grows steeply with trade size relative to the pool, so every large trade leaves the pool's price displaced from where it started.

After a large buy pushes a pool's price above the market, the pool is now offering to sell the token too cheaply relative to everywhere else — or rather, it is quoting a price nobody would pay voluntarily, which is exactly the arbitrageur's invitation. The arbitrageur sells into the pool (or buys from it, in the reverse case), collecting the difference between the pool's displaced price and the true market price, and the pool lands back near fair value. The original large trader paid for this whole cycle through their price impact; the arbitrageur simply harvests the distortion they left behind.

Stale Pools: When Nobody Is Watching

The most common inefficiency is staleness. A pool that trades infrequently — a mid-cap token's secondary pool, a new pool that has not attracted volume yet — can sit at an outdated price for minutes or even hours while centralized exchanges move on. When the market rallies 3% and the pool has not been touched, it is effectively offering a 3% discount to the first trader who notices. These are the gaps retail arbitrageurs can realistically compete for, because professional bots concentrate on the deepest, most profitable pools and may monitor thin ones less aggressively.

Why Inefficiencies Persist Instead of Vanishing

If arbitrageurs correct mispricings, why do any survive? Because correction is only worth doing when the profit exceeds the cost of the correcting trade. A 0.4% mispricing on a network where the round trip costs 0.5% in gas and fees will sit there uncorrected indefinitely — it is an inefficiency in theory but not an opportunity in practice. This is the single most important filter in AMM arbitrage: the question is never how big the gap is, but how big it is after every cost. Beginners who skip this filter end up executing textbook-perfect corrections that lose money.

Trading the Inefficiency Without Becoming the Victim

The irony of AMM arbitrage is that the arbitrageur's own trade creates price impact too — you are correcting the pool by trading against it, which moves the price as you go. Sizing is therefore everything: too small and gas eats the profit, too large and your own impact erases the edge you came to capture. The practical approach is to compute the optimal size where marginal profit turns negative, then trade well below it to leave a safety margin. Simulating the trade before sending — checking the expected output at current pool state — separates professionals from gamblers.

Risk management matters as much as math. Volatile tokens can move against you between detection and confirmation; MEV bots can sandwich your correcting trade; and thin pools sometimes belong to tokens with transfer taxes or trading restrictions that turn a clean-looking gap into a trap. Verify the token contract, set slippage limits, and never commit a size where a failed or frontrun trade would hurt. For the broader venue context of these trades, see our CEX vs DEX arbitrage guide, and for single-venue loops, triangular arbitrage.