Automated market maker: how reserves set token prices
What is an automated market maker in crypto? It prices swaps from pool reserves: each trade changes the reserve ratio, moving the next quoted price.
By Crypto Docket Newsroom#0a27935 min read

An automated market maker (AMM) sets a token’s price from the ratio of assets in its liquidity pool, rather than from bids and asks in an order book. The curve determines how much a trader receives for an input, and the resulting reserve change shifts the price for the next trade.
A constant-product AMM such as Uniswap v2 follows the rule x · y = k: the product of the two reserves stays constant in the idealized curve calculation. A trade adds one token and removes the other, so the ratio—and therefore the marginal price—moves as the swap proceeds.
For a swapper, the networks supported by an aggregator affect which pools and routes can be considered for a trade. KyberSwap Aggregator searches liquidity sources across its supported networks; see kyberswap supported chains to check whether a network you use is in scope.
How does an AMM turn reserves into a quote?
A constant-product AMM takes an input token, applies the pool’s fee, and solves the curve for the amount of the other token it can release. With reserves x and y, the no-fee relation is x · y = k. For an input Δx, the output is y − k/(x + Δx).
The ratio of reserves also gives the pool’s marginal spot price. If a pool holds x units of token A and y units of token B, one unit of A is priced at about y/x units of B before fees and trade impact. That is the price for a very small trade; a larger swap moves along the curve and gets a different average price.
Fees matter to the quote. Uniswap v2 charges a 0.3% swap fee, so the curve calculation uses less than the full amount the trader sends in. The fee remains in the pool, so the actual reserve product rises slightly after a swap; x · y = k describes the pricing curve between trades, rather than an exact invariant across fee-paying trades.
Why does a larger trade move the price?
A constant-product curve releases progressively less output for each additional unit of input. As a trader buys one token, its reserve shrinks while the reserve of the payment token grows. The changing ratio makes later units cost more than earlier ones.
Take a pool with 100 ETH and 300,000 USDC. Its reserve product is 30,000,000, and its initial spot price is 3,000 USDC per ETH. To buy 10 ETH, the pool must retain 90 ETH. At that reserve level, it needs about 333,333.33 USDC, so the trader adds roughly 33,333.33 USDC to the pool before fees.
The average price is about 3,333 USDC per ETH, roughly 11% above the initial spot price. That difference is price impact: the trade itself moves through the curve. The next trader sees the new reserves and a higher marginal price for ETH. Fees add to the trader’s cost, but they are separate from this reserve-driven impact.
In the formula, orient x as the USDC reserve and y as the ETH reserve. The required USDC input for 10 ETH out is found by setting the remaining ETH reserve to 90 and solving for the new USDC reserve. This is the same curve calculation as y − k/(x + Δx), with the reserve labels chosen to match the direction of the swap.
Who brings the pool price back toward the market?
Arbitrageurs trade when an AMM’s marginal price differs from prices available elsewhere. If ETH becomes cheaper in the pool than on external markets, they can buy ETH from that pool; if it becomes more expensive, they can sell ETH into it. Those trades shift the reserves back toward a ratio closer to the external price.
The pool does not read an external price feed to make this adjustment. Arbitrage activity connects its reserve-based price to other markets, while the curve sets each pool’s available trade and the amount needed to move its price. The alignment can lag or vary with trading activity and available liquidity.
This is the main change from an order book: a trader does not match with a particular resting order. The pool’s formula supplies the quote, and arbitrageurs respond when that quote diverges from other venues. A deeper pool generally takes a larger trade to move by the same amount, while a shallow pool can shift more sharply.
How do different AMM curves change the trade-off?
Constant product gives liquidity across the full range of possible prices, which makes the formula simple but can leave much of a pool’s capital away from the current trading price. Curve’s StableSwap uses a different curve designed for assets expected to trade near parity, such as stablecoins; its pricing behavior differs from a constant-product pool.
Concentrated liquidity changes where capital is active. Uniswap v3 and KyberSwap Elastic let liquidity providers choose price ranges, concentrating liquidity within them. This can provide more depth around the prices where trading occurs, but liquidity outside the selected range is inactive for swaps at the current price.
For multi-pool swaps, an aggregator can compare available quotes instead of relying on one pool. KyberSwap Aggregator reads pool reserves to calculate outputs and routes, then selects a route for the trade. When evaluating a quote, watch the pool depth, the fee, the price impact, and whether the route crosses multiple pools: each affects the amount that reaches your wallet.