Price impact vs slippage: the two numbers on a swap screen, what each one costs and how to size a trade to the pool

A swap screen shows two percentages that both describe losing money to the price. One is a fact about your trade, computed exactly from the pool before you send; the other is an estimate about everyone else's trades in the next few seconds. Confusing them leads to two common mistakes: raising the tolerance setting to 'fix' a high impact number, which does nothing, and buying a position that can never be sold at the chart price because nobody looked at impact on the way out. This guide separates the two with the pool arithmetic, shows what each one costs at different trade sizes, and gives a sizing rule that works for both buys and sells. What is slippage in crypto covers the tolerance setting and bots in depth; here the focus is the distinction and the sizing.

Two definitions

• Price impact: the gap between the pool's spot price before your trade and the average price you actually get, caused by your trade moving the reserves. It depends only on your trade size and the pool's size and is identical whether you trade now or in an hour, as long as the pool is the same.
• Slippage: the gap between the price you were quoted and the price at execution, caused by trades that landed in the pool after your quote and before your transaction. It depends on timing, volume and how long your transaction waits.

A quote already includes impact. If the screen says 'you receive 950,000 coins, price impact 4%', the 950,000 is the number after impact. Slippage is what might make it 920,000 by the time the transaction lands, and the tolerance setting is the floor below which you refuse.

The impact arithmetic

In a constant-product pool, if you add a fraction f of the native reserve (ETH, BNB or USDC), the coins you receive are the coin reserve times f divided by one plus f. Your average price is the old spot price times one plus f, and impact, measured the way DEX interfaces measure it, as the share of the at-spot output you do not receive, is f divided by one plus f. The new spot price is the old price times one plus f squared, so the chart moves by about twice your impact. What is a liquidity pool derives this; here is the table that matters:

• Trade is 0.5% of the native reserve: impact 0.5%, spot after 1.0% higher.
• Trade is 1%: impact 1.0%, spot after 2.0% higher.
• Trade is 5%: impact 4.8%, spot after 10.3% higher.
• Trade is 10%: impact 9.1%, spot after 21% higher.
• Trade is 50%: impact 33%, spot after 125% higher.
• Trade is 100% (you add as much ETH as the pool holds): impact 50%, spot after 300% higher; you receive exactly half the pool's coins.

Note that impact is always smaller than the spot move: you get the average of the curve you traded along, not its end point. Uniswap's documentation defines the figure its interface shows at docs.uniswap.org. A buyer who sees the chart jump 20% on their own trade paid about 9% above the previous price, not 20%. The reverse holds for sells: selling coins worth 10% of the pool's native reserve costs about 9% impact and leaves the spot about 17% lower.

A worked buy and sell

Pool: 10 ETH and 400 million coins, ETH at $3,000, so $30,000 of ETH and a spot price of $0.000075. You buy with 0.5 ETH ($1,500), 5% of the reserve. Ignoring the fee, you receive 400 million times 0.05 divided by 1.05, which is 19.05 million coins, at an average of $0.0000787: 4.8% impact. The spot is now $0.0000827, 10.3% higher. An hour later, with no other trades, you sell all 19.05 million coins. The pool returns to 10 ETH and 400 million coins and you receive your 0.5 ETH back. Impact on the way in and out canceled exactly, because nothing else moved the reserves. Add a 1% fee on each leg and the round trip costs 2%; impact itself cost nothing because you reversed your own move. Impact only becomes a real loss when someone else trades in between, which is where it meets slippage.

Why splitting a trade does not reduce impact

A common suggestion is to split a large buy into several smaller ones to reduce impact. In the same pool with no other trades in between, this changes nothing: two buys of 0.25 ETH end at exactly the same reserves as one buy of 0.5 ETH, and deliver exactly the same total coins, because each piece starts where the previous one ended. The constant product does not care how many transactions got it there. Splitting helps in three other situations: when other traders may sell between your pieces (you buy the dips they create), when you want to limit how much any single transaction exposes to a sandwich bot, and when the same coin trades on several pools so each piece can take the deepest one. Those are slippage and MEV reasons, not impact reasons.

Sizing a buy to the pool

A practical rule: keep each buy under 1% of the pool's native reserve for a trade that barely registers, and under 5% for one you consider deliberate. Above 10% you are the market, and the chart move you cause will be read by others as a signal, which cuts both ways. Coin pages report liquidity as the dollar value of both sides, so halve it to get the native reserve. A page showing $60,000 liquidity has roughly $30,000 of ETH; 1% of that is $300, 5% is $1,500. On a four-chain launch race, the four pools share a combined starting market cap of about $3,000, so early race buys are large relative to their pool by design; the race page shows each pool, and how big should your first buy be works through what that means for a creator's opening buy.

Sizing the exit, which is the one that matters

Buyers size entries and forget exits. A position bought at 2% of the pool can become 20% of the pool after a run, not because the buyer added anything but because the price rose and other holders are the ones who now have to absorb the sell. Suppose you bought $300 of a coin when the pool held $30,000 of ETH, and the coin then rose 25 times while the pool, after the run and some sells, holds $50,000 of ETH. Your position is worth $7,500 on the chart, 15% of the reserve. Selling it all at once costs about 13% impact and leaves the spot about 24% lower: you receive about $6,500, and everyone else's position is marked down. The chart said $7,500; the pool says $6,500 minus fees. Take profit and stop loss for meme coins covers staging exits; the impact table above is the reason staging matters at all.

1. Before buying, note the pool's native reserve and your position's share of it.
2. As the coin moves, recompute: position value divided by current native reserve. The number you care about is that ratio, not the gain percentage.
3. If the ratio passes 5%, your exit has real impact; if it passes 10%, you cannot exit at the chart price and should plan to sell in pieces as other buyers arrive, since each piece lets the reserve refill between sells.
4. Deeper pools fix this. A coin whose liquidity grows with its market cap keeps exits cheap; a coin whose market cap grows 50 times on a pool that grows 2 times has a chart that nobody can realize.

Reading both numbers on a coin page

A good coin page shows liquidity, the pool's native reserve or something equivalent, and on the buy screen an expected fill and an impact percentage. The chart tells you about the past; the impact figure tells you what your entry does to the present; the ratio of your position to the pool tells you about your exit. How to read a meme coin chart explains why a vertical candle on a tiny pool is less information than it looks, and this is the arithmetic behind that: a $2,000 buy into a pool holding $5,000 of ETH draws a nearly 100% candle and represents one person. On sasa the buy screen quotes in USDC, so the expected fill is already the dollar cost after impact; what remains uncertain is slippage, which the floor on the quote handles. Impact is yours to size; slippage is yours to bound; neither is a surprise once you know which is which.