Slippage

On an AMM, slippage has two components. The first is price impact, the move the trade itself causes by shifting the reserves along the x times y equals k curve, larger the bigger the order is relative to pool depth. The second is the variation between the user signing and the transaction being mined, because other trades alter the pool first. The user protects against this by setting a slippage tolerance (for example 0.5% or 1%) that defines the worst acceptable price. In the contract this maps to parameters such as amountOutMin (minimum tokens to receive) or amountInMax, plus a deadline. If execution falls outside the tolerance, the transaction reverts.

It is the real execution cost, beyond the fee and gas, and at the same time the safety parameter that prevents executing at an unacceptable price. The deeper the pool, the lower the impact for the same order size, and Curve's StableSwap curves keep slippage very low near the peg. Managing it well is central to designing routers, aggregators and trading bots.

Tolerance connects with MEV (Maximal Extractable Value). A high tolerance opens the door to sandwich attacks, where a searcher front-runs and back-runs the order to capture value, so the setting is a balance between execution probability and protection against worse prices. No EIP defines slippage; it is an emergent property of AMMs and is controlled at the contract layer with the Uniswap router parameters, present since v2 in May 2020. The MEV concept was popularized by the Flash Boys 2.0 paper (Daian et al., 2019). Real tooling includes the tolerance and deadline controls in the Uniswap interface, the 1inch and CoW Protocol aggregators, the latter using batch auctions to reduce slippage, and Curve's design for pegged assets.