On 12 March 2026 a wallet swapped about $50.4 million of aEthUSDT, the token Aave issues for USDT deposits, for AAVE and received tokens worth roughly $36,000, CoinDesk reported. The trade was routed through pools far too thin for its size and slippage ran past 99%. According to Aave founder Stani Kulechov, the Aave app had warned about extraordinary slippage and required the user to tick a box before it would go ahead. Nothing was hacked. The swap executed exactly as the settings allowed.
Slippage is the gap between the price you are quoted and the price you actually get. Many guides stop at that line and move on to tips. That leaves out the parts that cost people money, which are the difference between slippage and price impact, what the tolerance setting actually controls, and how bots exploit a tolerance set too wide. This piece covers all three with worked numbers, then shows how to keep the cost small.
What slippage means
Investopedia notes that slippage can run in either direction. Negative slippage means you pay more or receive less than quoted, positive slippage means the market moved in your favour before the order filled. In practice traders complain about the negative kind.
Slippage shows up in two very different places. On a centralised exchange your market order is matched against other people's orders in an order book. On a decentralised exchange (DEX) your swap usually trades against a liquidity pool, a smart contract holding two tokens that sets the price by formula. The causes overlap, but the mechanics and the fixes differ, so it pays to take them one at a time.
Slippage vs price impact
These two terms get used interchangeably, and Uniswap's own help centre says plainly that they do not mean the same thing. Price impact is the change in price caused by your own trade. Price slippage is the difference between the price you expected to receive and what you actually received once the swap completed. In everyday use, and in most wallet warnings, slippage covers the whole gap between quote and fill, price impact included. This piece uses the narrow sense where the difference matters.
The distinction matters because the two have different owners. Price impact is known before you click. It depends on your trade size and the depth of the pool or book, and a good interface shows it to you. Slippage in the narrow sense is what happens between the quote and the execution, when other trades land first, the price drifts or a bot gets in the way. You control price impact by choosing your size and venue. You limit slippage with the tolerance setting. The $50 million swap above shows both senses at once. Its warning called the loss slippage, but it was mainly price impact, a trade far larger than the pools it was routed through could absorb, and the settings the user confirmed let it execute anyway.
How a DEX pool sets the price
The classic AMM (automated market maker) design is the constant product formula behind Uniswap v2. The pool holds two tokens, and the product of the two balances has to stay constant after every trade. As Uniswap's pricing docs put it, when a token is withdrawn a proportional amount must be deposited to maintain the constant, and the ratio of tokens in the pool determines the price a swap executes at.
Take an illustrative pool holding 100 ETH and 200,000 USDC. The numbers are round for easy arithmetic, not a market quote. The product is 20 million and the pool prices ETH at $2,000. Spend 10,000 USDC on ETH and the USDC side grows to 210,000, so the ETH side must shrink to 20 million divided by 210,000, which is about 95.24 ETH. You receive the difference, about 4.76 ETH, at an average price of $2,100.
| USDC spent | ETH received | Average price paid | Above the $2,000 starting price |
|---|---|---|---|
| 1,000 | 0.4975 | $2,010 | 0.5% |
| 10,000 | 4.7619 | $2,100 | 5% |
| 50,000 | 20.0 | $2,500 | 25% |
The pattern is clean in this simple design. Before fees, the premium you pay equals your trade size as a share of the pool's USDC side. Spend 0.5% of that side and you pay about 0.5% over the starting price, spend a quarter of it and you pay 25% more. The swap fee comes on top, and the next buyer faces a worse price still, since after the 10,000 USDC trade the pool prices the next ETH at about $2,205. Arbitrageurs then buy cheap ETH elsewhere and sell it into the pool until its price lines up with the wider market again.
Two refinements matter in practice. Uniswap v3 and v4 use concentrated liquidity, where liquidity providers, the people who deposit tokens into the pool, place capital in chosen price ranges, so traders get deeper liquidity around the current price than the flat formula above suggests. That depth sits only where providers choose to place it, so a pool can be deep at the current price and close to empty a little way from it. And depth is a property of the specific pool you trade against, not of the protocol's headline size, which is why a protocol's total value locked tells you little about how much your swap will move the price. Stablecoin pairs show the stakes most clearly. They normally trade with tiny impact, until one coin loses its peg and every holder tries to swap out through the same pool, as in the episodes our depeg explainer walks through.
Slippage on order book exchanges
On a centralised exchange, and on onchain order book venues such as the perpetual futures exchange in our Hyperliquid review, the price comes from orders already waiting in the book rather than a formula. A market order takes whatever is on offer, starting with the best price and working down the queue until it is filled. Traders call this walking the book.
Picture a thinly traded token where the best bid is $0.998 and the asks are 2,000 tokens at $1.000, 3,000 at $1.004 and 5,000 at $1.012. A market buy for 10,000 tokens fills all three levels and costs $10,072, an average of $1.0072, or 0.72% above the best ask you saw on screen. The spread, the 0.2% gap between the best bid and the best ask, is a separate cost. You would pay it even on a tiny order if you bought and sold straight back. The extra 0.72%, caused by your own order being bigger than the first level of the book, is price impact, the order book version of the pool example, though exchanges and traders usually just call it slippage.
Volatility adds slippage in the narrow sense. In a fast market, orders get cancelled and replaced between the moment you see the book and the moment your order arrives, so even a small market order can fill at a worse level than the one you saw. Thin books and fast markets arrive together during news and liquidation cascades, and small coins run thin books in quiet weekend hours too. Those are the moments when slippage is at its worst.
What slippage tolerance actually does
Slippage tolerance is the worst result you will accept, expressed as a percentage below the quoted amount. The swap carries that minimum with it, and if the pool cannot deliver at least that much when the transaction executes, the swap is cancelled rather than filled at a worse rate. Uniswap's v2 docs describe the original rule. The interface computed a minimum from the rate it observed, and the router contract refused any fill worse than that, with 0.5% as the default tolerance. As of November 2026 the Uniswap app offers an Auto setting that, per its support page, lands between 0.5% and 5% depending on network cost and swap size, and you can override it.
Run the example pool again. You are quoted 4.7619 ETH for 10,000 USDC and set 0.5% tolerance, so the minimum you will accept is about 4.738 ETH. Before your transaction is included, another trader spends 5,000 USDC on ETH in the same pool. Your swap now returns about 4.538 ETH, roughly 4.7% short of the quote, so it fails. On a chain like Ethereum that failure still costs money, since, as Uniswap explains, the network charges a gas fee for the attempt even when the transaction fails.
Set the tolerance at 5% instead and the same swap goes through at 4.538 ETH. That is the trade-off in one example. A tight tolerance protects your price and risks failed transactions. A wide one gets the trade done and leaves you exposed to anything that moves the price before you, up to the whole margin you allowed.
Sandwich attacks and MEV
That margin is what MEV bots hunt. Maximal extractable value is, in the words of ethereum.org, the maximum value that can be extracted from block production in excess of the standard block reward and gas fees by including, excluding and changing the order of transactions in a block. In plain terms, control over the order of transactions lets someone profit by placing trades around yours. Swaps waiting in the public mempool, the waiting room where transactions sit until a block includes them, are visible before they execute, and that visibility is the opening.
A sandwich attack works like this. A bot sees your pending swap and its minimum output. It buys the same token just before you, pushing the price up as far as your tolerance allows, lets your swap execute at that worse price, then sells straight after you into the price your trade pushed even higher. Ethereum.org describes the result for the victim as increased slippage and worse execution. Your tolerance setting is the ceiling on what a bot can extract from you.
The losses can be enormous even on major stablecoins, assets that should not move. In March 2025, Cointelegraph reported that on 12 March a trader swapped $220,764 of USDC for just $5,271 of USDT in a Uniswap v3 pool. According to the onchain analysis it cited, the bot traded just ahead of the swap by pulling all the USDC liquidity out of the pool, then put it back after the trade executed. It kept about $8,000 and tipped $200,000 to the block builder, the party that assembles a block and sets its order, to secure that position. Some analysts suggested the trade could have been a deliberate way to move funds, but the mechanics hold either way. The swap's settings let it accept almost any output, and a stablecoin swap with that much room has no price protection at all.
Risk is not spread evenly. A study of Uniswap trades by Austin Adams and co-authors found that gas dominates costs for small swaps while price impact and slippage account for most of the cost on large ones, and that the probability of adversarial slippage, the extra loss from bots trading around a swap, was about 80% higher when trading the memecoin PEPE than when trading a mature asset like USDC.
How to reduce slippage
Start by reading the price impact line before you touch the tolerance. If a swap only works with a wide tolerance, the honest message is usually that the pool is too thin for your size, and raising the tolerance just hands the difference to whoever trades ahead of you.
On a centralised exchange, use limit orders. A limit order fills at your price or better, and Investopedia notes that limit orders and avoiding volatile periods can reduce negative slippage. The cost is that a limit order may never fill if the price runs away from it.
For a large trade, size matters more than anything else. Because price impact rises with your share of the pool or the book, splitting it into several smaller trades spaced over time lets arbitrageurs pull the pool's price back between them. Each extra swap pays its own gas and fee, and the price can move while you wait, so splitting only pays when the impact it saves is bigger than those costs. DEX aggregators do a version of this automatically, routing one order across several pools to reduce impact on any single one. Sticking to liquid pairs and deep pools does the same job from the other side, and so does trading when the market is calm rather than during news or a sell-off.
Protection against MEV comes from keeping your transaction out of the public mempool, or from not sending a raw swap at all. Flashbots Protect sends transactions to a private mempool where they are hidden from bots looking to trade around them, and users do not pay fees for failed transactions. UniswapX lets you sign an order that specifies what you want to receive and has third-party fillers compete to fill it, and its docs list protection against MEV and no cost for failed transactions among its features. None of these tools fixes a trade that is simply too big for the market. They stop other people from widening the loss.
How much slippage tolerance to set
There is no single right number, but there is a sensible order of thinking. On deep pairs of major assets, a tolerance around the old 0.5% default is usually enough, and stablecoin-to-stablecoin swaps on deep pools should need less. Let the interface's automatic setting propose a figure, check that the minimum received is a number you would genuinely accept, and treat any tolerance above a few percent as a sign to look again at size, pool or timing. For thinly traded tokens, or tokens that charge a fee on every transfer, a higher setting may be unavoidable, and that wider margin is exactly what a sandwich bot looks for.
Where slippage meets yield
Most slippage costs have someone on the other side of them. The $50 million from the opening did not vanish. Within the same block, arbitrageurs extracted more than $43 million in profit and $32.6 million of that went to the block builder, figures BlockSec gave to CoinDesk. On an order book, the orders your market order fills often belong to market makers, firms that quote both sides and earn the spread. When a price gap opens between two venues, arbitrageurs close it, and the gap a retail trader sees on screen is usually smaller than the fees and slippage it would take to capture it. Market makers and arbitrageurs live or die on execution cost.
EarnPark's strategies sit in some of those seats, and their own descriptions say so. Maker Core, one of them, earns partly from market making, placing buy and sell orders on exchange order books and earning the spread when both sides fill, per its strategy description. The strategies EarnPark calls DeFi carry the Cross-Venue Risk Label (EarnPark labels each strategy by the kind of risk it takes) and trade a different kind of gap between venues. They hold matched long and short positions to earn differences in interest and funding rates rather than spot prices, and they count trading fees, transfer costs and execution expenses against what they earn. Our funding rate guide explains how that cross-venue trade works and where it breaks.
Being the market maker does not make slippage go away. The Maker Core description itself warns that in fast markets quotes may fill at prices that are already stale, and that thin depth or wide spreads may prevent positions being closed at the expected price. Maker Core's rate is a target, "an objective, not a guarantee". EarnPark's strategies use leverage, run on large centralised exchanges and onchain venues, carry no deposit insurance and can post negative results. Our risk statement covers the platform-level risks.
There is a plainer point too. Stablecoin holders who hop between coins and pools to chase a better rate pay a spread, some price impact and sometimes an MEV tax on every move. Earning on the coin you already hold means one entry and one exit instead of a string of swaps. As of November 2026 EarnPark publishes up to 10% APY on USDC from two strategies. USDC DeFi withdraws instantly and charges 0.3% plus a flat $3.20 when you take funds out, while USDC Maker Core pays out after a 30 day bonding period, a fixed wait between a withdrawal request and payment, for a flat $2. Those are the same published rates any account gets holding zero PARK, EarnPark's own token, as our page on token gating explains, and each strategy shows its current terms in the app.
Slippage is not bad luck. Most of what you lose to it can be seen coming, set by your trade size, the depth of the market and how much room you leave others to trade ahead of you. If earning on what you hold suits you better than trading in and out, compare the Risk Labels and terms in the app and Start earning when one fits.
FAQ
How much should my slippage tolerance be?
For deep pairs of major assets, around 0.5% is usually enough, the default the Uniswap interface used in the v2 era, and deep stablecoin pairs should need less. As of November 2026, Uniswap's Auto setting ranges from 0.5% to 5% depending on network cost and swap size. If a trade only goes through with several percent of tolerance, the pool is probably too thin for your size, and a wide setting gives sandwich bots room to take the difference.
What is the difference between slippage and spread?
The spread is the gap between the best bid and the best ask in an order book at a given moment, and you pay it on any round trip however small. Slippage is the difference between the price you expected and the price you got, whether from your order consuming several price levels (price impact) or from the market moving before your order filled. A wide spread often goes with high slippage, since both are symptoms of thin liquidity.
What happens if slippage tolerance is too high?
A high tolerance lets your swap execute at a much worse price than quoted. Anyone who trades ahead of you, including a sandwich bot watching the public mempool, can push the price against you up to the limit you set and profit from the difference. In March 2026 a $50.4 million swap went through after the user confirmed a slippage warning, and returned about $36,000 worth of tokens.
What happens if slippage tolerance is too low?
If the price moves past your tolerance before your transaction executes, the swap fails instead of filling. You keep your tokens, but on networks like Ethereum you still pay the gas fee for the failed attempt, and in a fast market you may need several retries.
Is price impact the same as slippage?
No. Price impact is the change in price your own trade causes, and it is visible before you submit because it depends on your size and the pool's depth. Slippage in the narrow sense is the gap between the quoted and executed price caused by everything that happens between submitting and execution, though wallet warnings often lump both together as slippage. You manage price impact with trade size and venue, and slippage with your tolerance setting and MEV protection.
Can slippage be positive?
Yes. If the price moves in your favour between the quote and the fill, you receive more than quoted or pay less. On a standard DEX swap the tolerance only sets a floor on what you receive, so a better price usually goes through as a better fill.