Slippage

An order book is a stack of finite offers, not a single price, so a market order bigger than the best offer climbs to worse ones. On October 21, 2021 a buggy client algorithm walked the Binance.US book from about 65,760 dollars down to 8,200 in under a minute. For a 200 dollar purchase the effect is usually pennies, and knowing when it is not is the whole point.

How it works

Slippage comes from the gap between what a screen shows and what a book contains.

An exchange displays one headline number, normally the last trade or the midpoint between the best bid and the best ask. Behind it sits the book: a list of individual offers at rising prices, each with a limited size. A market order instructs the engine to buy whatever is available until your amount is filled, so it consumes the cheapest offer, then the next, then the next. The average of every price you touched becomes your fill price, and the distance between that average and the headline number is slippage.

Three things make it larger. Size is the first, because an order that exceeds the depth sitting near the top of the book has to reach for worse prices. Thinness is the second: the same 5 bitcoin order that barely dents a major BTC/USD book can move a small local pair by several percent. Timing is the third, because during a fast move the book empties faster than market makers refill it, which is why the effect is worst in exactly the minutes when people feel most urgent about trading.

Slippage also runs in your favor. If the price ticks down between submission and execution, the market order fills cheaper than quoted. Exchanges never advertise this, but positive slippage is common on quiet books, and it is why slippage is a distribution of outcomes rather than a fee.

Why this matters when you buy bitcoin

Slippage decides how much of a purchase survives execution, and for most readers of this site it is smaller than the costs they are already ignoring.

At retail sizes, a market order on a deep BTC/USD or BTC/EUR pair typically fills within a couple of dollars of the quote. Kraken Pro's 0.40 percent taker rate on a 1,000 dollar buy is 4 dollars, while the slippage on that same order is usually a fraction of a dollar. Worrying about slippage while paying instant-buy pricing is the wrong optimization, because the widget's built-in markup costs several times more than either.

There are three situations where it stops being a rounding error. Large orders are the obvious one: past five figures per trade, splitting the order or resting a limit order keeps you off the expensive rungs. Small local markets are the second, because an exchange serving one country often quotes far less depth against the local currency than against dollars, so a mid-size order visibly moves it. Volatile minutes are the third, when quotes move between the instant you tap and the instant the matching engine sees the order.

The defenses cost nothing. A limit order names the worst price you will accept and simply does not fill beyond it. Routing through the deepest pair and converting afterward often beats trading an illiquid direct pair. And when a platform offers a slippage tolerance setting, read a generous tolerance as permission to charge you badly rather than as a convenience.

A 250,000 dollar order into a thin book

Arithmetic shows where slippage comes from, one book level at a time.

Picture a book where the best ask is 100,000 dollars for 0.5 bitcoin, the next is 100,050 for 0.8, and the third is 100,200 for 1.2. A market order for 2.5 bitcoin consumes all three. You pay 50,000 dollars for the first half coin, 80,040 for the next 0.8, and 120,240 for the last 1.2, a total of 250,280 dollars. Your average price is 100,112 against a quoted 100,000, so slippage cost 280 dollars, or 0.112 percent. The 0.40 percent taker fee on that trade is about 1,001 dollars, three and a half times larger, and only one of those two numbers appears in most people's mental arithmetic.

Now delete the middle two levels, which is what a thin book looks like. The same 2.5 coin order takes 0.5 at 100,000 and 2 at 101,000, an average of 100,800 dollars and 2,000 dollars of slippage. Nothing about your order changed. Only the depth did.

Slippage vs bid-ask spread

The bid-ask spread is what the book costs you before you trade; slippage is what your own order costs you while it trades. A spread exists whether or not you show up, since it is the standing gap between the best bid and the best ask, and crossing it once is the minimum price of immediacy. Slippage is the extra distance your order travels past that best ask because it is too large or the market moved underneath it. A venue can quote a tight spread over a shallow book and still fill you badly, which is why depth matters more than the headline spread on smaller platforms.

Slippage vs volatility

Volatility is the market moving; slippage is your order missing the price you saw. The two overlap in the seconds around a trade, and people routinely blame the wrong one. If bitcoin falls 4 percent during the hour after your order filled, that is volatility, and nothing was mispriced. If your order filled 0.6 percent above the quote at a moment when the market barely moved, that is slippage, and a limit order would have stopped it.