What is slippage in crypto? Why what lands isn't always what you were shown, and how to keep the gap small.

01Slippage in one picture

You were shown one amount. A different one landed. That gap has three parts, and once you can tell them apart, you can shrink each one.

Strictly, slippage is the middle part: the price moving between your order and its execution. In everyday use people call the whole gap slippage, and it can go either way. Researchers note a trade "can receive more asset Y than expected", too.

The sections below take the three parts one at a time. For every cost in a swap, the main guide on how to swap crypto has the full receipt.

02Price impact: the one formula behind the number

Most decentralized exchanges price trades from a pool of two coins. The rule is almost embarrassingly simple: the two reserves multiplied together stay constant.

The research on these markets writes it as "constant product k = RαRβ", and a study found this model held "over 75% market liquidity" in 2020.

Here's what that means for you. When you buy, you add one coin and take the other, and the price moves against you as you go. The bigger your trade compared with the pool, the further it moves:

a trade worth 1% of the pool moves the price about 1%. A trade worth 10% moves it about 9%. As the research puts it, "the larger the amount of reserves, the less any one trade will cost".

03Price impact vs slippage: two different gaps

The two get mixed up constantly. The research separates them cleanly:

• Expected slippage, which is price impact: the change "based on the volume to be traded and the available liquidity".
• Unexpected slippage: any further change "from the submission of a trade commitment to its execution".

Why does the second one happen at all? Because your quote is built on "a past blockchain state, which may change between the transaction creation and its execution". Other trades land first. Prices move.

Bottom line: price impact is yours and visible. Slippage is the world moving while you wait.

04Slippage tolerance: what to set

Slippage tolerance is the worst price you'll accept, written as a percentage. ethereum.org puts the rule simply: at 1%, "the minimum you can expect to receive = expected output-1%".

Every guide wants to give you a magic number. There isn't one. There's a method:

1. Cover only what can move while you wait. Price impact is already in the quote, so don't pad for it.
2. Too low, and the trade fails instead of filling at a worse price. You can try again.
3. Too high, and you hand over room. Every extra percent is value the price, or a bot, can take.

And most people never touch it: ethereum.org notes "Most users will leave default slippage anyway". Try your own numbers:

05Sandwich attacks: how bots use your tolerance

Here's why a loose setting costs real money.

ethereum.org describes it step by step: a bot "will watch the mempool for large DEX trades", buys just before yours, and sells just after. The result: "Users who are sandwiched face increased slippage and worse execution on their trades."

The detail that matters: the attacker has to stay inside your limit, or your trade fails and the attack with it. The research says the bot sizes its first trade so that your "slippage protection does not trigger a transaction revert". Your tolerance is the bot's budget.

1. 2018 to 2021. 750,529 sandwich attacks worth $174.34 million to the attackers, a third of them sent privately to miners.
2. Late 2024. 2,932 private sandwiches cost their victims $409,236.97 in two months.
3. 2026. A three-year study found 38,567 sandwiches on Tron and 30,607 on Ethereum against trades that were supposed to be protected.

Private routing helps less than people expect. A tight tolerance is the defence you control.

06Why a quote can move on a swap service

A swap service isn't a pool you hit in one block. You send coins, they confirm, and only then does the trade happen. The wait is the sending chain's.

On Bitcoin, "New blocks are added to the blockchain approximately every 10 minutes on average", and bitcoin.org itself says the price "is still very volatile". On Ethereum, a sent transaction waits in "a transaction pool" until a validator includes it.

There's a second effect people miss. Network fees "are unrelated to the amount transferred", so a fixed cost takes a bigger share of a small swap. Our desk's readings below show it plainly. How confirmations work is in the swap guide's confirmations chapter.

07Fixed vs floating rates

Swap services quote in one of two ways, and the difference is simply when the amount is set.

With a fixed rate, the service carries the price risk while your coins confirm, so its quote has to cover that. A floating rate passes the market straight through, up or down.

08What our desk does, with our own readings

We quoted the same pair at five amounts on 3 November 2026 at 13:44 UTC: Fast, USDT on Tron to Bitcoin. Here's the share of the market value that landed:

60 USDT landed 0.00069610 BTC, 90.08% of its market value. 900 USDT landed 0.01143050 BTC, 98.61%. On a small swap, the gap is mostly fixed costs, not slippage.

And here's a floating amount moving: 300 USDT quoted 0.00375920 BTC at 13:35 UTC and 0.00376480 BTC at 13:44, 0.15% more, nine minutes later.

09The playbook: keep the gap small

Six steps, in order of how much they save:

1. Read price impact first. If it's high, trade smaller or find a deeper pool.
2. Set tolerance to what can move while you wait, and not a percent more.
3. Big trade plus wide tolerance is the sandwich target. Tighten one or the other.
4. On a swap service, pick the rate type on purpose: fixed when an exact amount must land, floating when a small move is fine.
5. Quote your own amount. Fixed costs weigh more on small swaps.
6. Send promptly after you quote, so the price has less time to move.

Holding USDT on Tron and want Bitcoin? Quote USDT on Tron to Bitcoin and see the live amount before you send, on our swap with no ID.