Slippage

Slippage is the difference between the price a trader was quoted when they submitted a swap and the price the swap actually executed at. On a decentralised exchange that gap has two contributors: the effect of the trade's own size against pool depth, which is price impact, and whatever else touched the pool between the quote and the fill. Slippage is the realised outcome. Price impact is the part of it you could have calculated in advance.

Slippage and price impact are used interchangeably and they are not the same measurement. One is what your size does to the price because of depth. The other is what you actually got versus what you were shown. Confusing them means a trading interface tells you a trade is safe when the risk is sitting in the part it did not measure.

Scroll to see the full diagram

Step one already prices your own size, so the residual gap comes from step four. That is the honest boundary between price impact and slippage, and it is also the window an attacker needs.

The boundary, stated so it stays put

Price impact is a function of the pool. Your trade consumes a fraction of the reserves, the curve reprices as it does so, and the result is computable from the contract's current state before you sign anything. It does not depend on timing, on other traders, or on luck.

Slippage is a function of time. It is measured against the quote you were shown, and it exists because the world moved between that quote and your fill. If nothing touches the pool in between, and your quote already priced your own size, realised slippage is zero and you still paid full price impact. That single sentence is the cleanest way to hold the two apart.

Paradigm's cost breakdown for a decentralised exchange trade treats them exactly this way, listing price impact, broker or trading fees, slippage, and the blockchain's own transaction fee as four separate components.² Four causes, four mitigations. Deeper reserves fix the first. A different fee tier changes the second. Faster inclusion or private order flow addresses the third. Nothing you do to the pool addresses the fourth.

Why the vocabulary is genuinely contested

The formal literature does not use the word the way a trading interface does, and pretending otherwise causes more confusion than it resolves. A peer-reviewed treatment of costs in AMM networks defines linear slippage as how a trade's size affects the trader's return and angular slippage as how a trade's size affects the subsequent market price, alongside divergence loss for the provider's opportunity cost and load for how those costs are shared.¹ In that framing, both of the numbers a trading interface splits into price impact and slippage are types of slippage.

The same paper shows these costs obey conservation laws, meaning they can be shifted between traders and providers but never fully removed by a choice of AMM design.¹ That is the part worth carrying into design work. A curve that gives traders better execution has moved the cost onto providers, and a curve that protects providers has moved it onto traders. There is no configuration where it goes away.

So use the operational definitions when you are talking to a trader, use the formal ones when you are reading a paper, and never let a document use both without saying which.

Tolerance is a setting, and it is also an attack surface

Slippage tolerance is the worst execution price you have pre-authorised. Set it too low on a shallow pool and your transactions revert, which costs gas and looks like the venue is broken. Set it high and you have told the world exactly how much value can be taken from you without the transaction failing.

That second case is not theoretical. The foundational academic work on decentralised exchange front-running documented sandwich attacks, where an attacker observes a pending trade and inserts one transaction immediately before it and another immediately after, capturing the price movement inside the victim's tolerated band.³ The tolerance is the attacker's budget. You set it.

For a project, this shows up as a support problem and a reputation problem rather than a treasury one. Holders with a default tolerance on a thin pool get poor fills, and the pool is why. The fix is depth, not a help article about changing a setting.

What this means at launch

Model the impact curve first, because it is deterministic. Find the trade sizes at which your pool produces 1%, 5% and 10% impact, then compare those against the position sizes your actual holders carry. If 5% arrives at a $3,000 trade in a community whose median holder is well above that, the pool is not sized for the people you sold to.

Then treat residual slippage as an execution problem separately. It is driven by how contested block space is on your chain, how much automated flow watches your pair, and how long a transaction sits pending. Those are chain and venue decisions, and no amount of extra liquidity fully removes them.

Both numbers belong in the launch model with different names. A single line item called slippage is where the two get silently averaged and the actual exposure disappears.

Where the confusion costs real money

The expensive version of this mistake happens in treasury execution. A team plans to sell protocol-owned tokens to fund operations, checks the quoted slippage on a single large order, sees a number it can tolerate, and executes. The interface priced the impact of that order against current reserves and reported it under whichever label it uses. What it did not price is the second, third and fourth order, or the fact that the pool does not refill on its own between them.

Splitting a large sale into tranches reduces impact per trade and increases exposure to everything that happens between trades. That is the tradeoff, and it is a real one with no free side. Deciding it deliberately, with the two costs named separately, is the difference between a funded quarter and an unplanned drawdown.

Academic work on AMM slippage continues to formalise these effects, and the practitioner literature is still catching up on the vocabulary.⁴ Until it settles, define the term in your own documents before you use it.

Common questions

What is the difference between slippage and price impact?

Price impact is what your own trade does to the price because of pool depth, and it is computable from the pool's current reserves before you sign. Slippage is the gap between the price you were quoted and the price you actually received, which exists because other transactions landed in between. If your quote already priced your size and nothing else touched the pool, slippage is zero and price impact is not.

What is a good slippage tolerance?

Low enough that a sandwich attack is not profitable, high enough that ordinary block-to-block movement does not revert your transaction. There is no universal number, because it depends on pool depth, pair volatility and how contested block space is on that chain. Treat a high tolerance as what it is: a pre-authorisation for how much value someone else may extract from your trade.

Why did my swap execute at a worse price than quoted?

Something changed the pool between your quote and your fill. That could be ordinary trading, an arbitrage transaction correcting the pool against another venue, or a deliberate sandwich in which an attacker traded immediately before and after you to capture the movement inside your tolerance. The quote was accurate for the reserves at that moment; the reserves were different by the time your transaction executed.

Can slippage be eliminated by better AMM design?

No. Formal analysis of costs in AMM networks shows they obey conservation laws, so they can be shifted between traders and liquidity providers but not removed. A curve that gives traders tighter execution has moved the cost onto providers, and the reverse also holds. Design decides who pays, and depth decides how much. Neither makes the cost disappear.

See Token Launch Strategy for how this applies in practice.

Sources

Loss and Slippage in Networks of Automated Market Makers (OASIcs Vol. 97, Tokenomics 2021)
Daniel Engel, Maurice Herlihy, Brown University (Schloss Dagstuhl, DOI 10.4230/OASIcs.Tokenomics.2021.13), 2021
Peer-reviewed definitions of linear slippage, angular slippage, divergence loss and load, with the proof that these costs obey conservation laws and cannot be fully eliminated.
Understanding Automated Market-Makers, Part 1: Price Impact
Paradigm, 2021
Enumerates price impact, trading fees, slippage and blockchain transaction fees as four separate components of the cost of a decentralised exchange trade.
Flash Boys 2.0: Frontrunning, Transaction Reordering, and Consensus Instability in Decentralized Exchanges (arXiv:1904.05234)
Daian, Goldfeder, Kell, Li, Zhao, Bentov, Breidenbach, Juels, 2019
Original formalisation of decentralised exchange front-running, including sandwich attacks that exploit a victim's pre-authorised slippage tolerance.
Slippage in AMM Markets (SSRN 4133897)
Irene Aldridge, SSRN, 2022
Cited for provenance of continuing formal work on AMM slippage. The abstract page blocks automated requests, so nothing is quoted from it here.

Last reviewed 2026-08

Related terms

Know the terms but not sure how they apply to your project? That is what an engagement is for. We design, document, and stress-test the whole token economy inside the Tokenomics Data Room.

Book a discovery call

100+ projects advised. Complete tokenomics in 4 to 6 weeks.