A trader who doubles an order size usually finds the extra cost more than doubles. The relationship is not linear, and the reason lies in how a book is consumed.
Each level is a separate price
A limit order book is a stack of offers at increasing prices, each with a finite quantity behind it. A market order fills against them from the best price outward.
Small orders finish inside the first level and pay the quoted price. Larger ones exhaust that level and continue into the next, and then the one after.
The average fill price is therefore a weighted blend across every level touched. Adding size does not just buy more units; it buys them at successively worse prices.
Depth thins as you move away from the touch
Books are rarely flat. Market makers concentrate size near the current price and post progressively less further out, because quoting far from the market ties up capital for a low probability of a fill.
That shape means each additional unit of order size walks further per unit consumed. The cost per coin rises as the order grows, on top of there simply being more coins to pay for.
The combination produces a cost curve that bends upward, which is why execution algorithms treat large orders as a fundamentally different problem rather than a scaled-up small one.
The book refills, but not instantly
Once a sweep clears several levels, market makers requote and depth returns, usually within seconds on a liquid pair.
Splitting a large order into pieces separated by time lets that replenishment do the work, so each slice meets a fuller book than a single sweep would have.
The trade-off is exposure to price drift while the order is being worked, which is the choice every execution schedule is balancing.
Automated pools price size differently
An automated market maker has no discrete levels. Its price comes from a curve over its reserves, so cost rises continuously with the fraction of the pool being traded against.
The effect is the same in character: a trade that touches a small share of reserves is cheap, and one that touches a large share moves the price sharply against itself.
Because the curve is public, the cost can be computed exactly before sending the trade, which is why on-chain interfaces quote expected slippage in advance while order-book venues can only estimate it.
Signalling raises the cost further
Visible large orders change other participants' behaviour. Market makers widen quotes and other traders position ahead of the expected flow.
This is why sizeable trades are often routed through venues that do not display resting size, or negotiated directly between two parties away from the public book.
The cost of moving a large position is therefore part mechanical and part informational, and the informational part grows with how predictable the order pattern looks to anyone watching.