Reading liquidity through the order book

Most slippage models start with the mid-price and a spread, then bolt on an impact term that scales with size. It is a clean abstraction, and it is wrong in the way that matters: it treats the book as a price when the book is really a queue. What you pay is not a function of where the mid sits, but of how much resting interest stands between your order and the other side — and how fast that interest is replenished while you wait.

Once you measure fills against the state of the book at the moment of execution rather than against a smoothed reference price, the comfortable picture comes apart. Two trades at the same nominal spread can realize very different costs, and the difference is almost entirely explained by depth and queue position.

Depth is a distribution, not a number

Top-of-book size tells you almost nothing on its own. What governs the cost of a meaningful order is the shape of resting liquidity over the first several price levels — how quickly size accumulates as you walk the book, and how stable that profile is across the seconds it takes to work an order.

A book that looks two ticks deep can be a tenth as resilient as one that looks identical at the top but thickens sharply behind it. We model depth as a distribution and track its moments through the day; the tails, not the mean, are where execution cost lives.

The mid-price tells you what the market thinks something is worth. The book tells you what it will cost to act on the thought.

Queue position is an option you are short

When you post passively, you join a queue. Your place in it is an option the market holds against you: if the move you feared arrives, the orders ahead of you fill and you are left holding the worst of the adverse selection; if nothing happens, you wait. Honest cost accounting has to price that option, not assume the passive fill was free.

In practice this means three things for how we read a fill:

• Time-to-fill matters as much as the fill price — a passive order that rests through an adverse move was never cheap.
• Cancellation pressure ahead of you reveals how real the displayed depth is; queues that thin out under stress were always thinner than they looked.
• Replenishment rate separates a genuinely deep book from one that merely refreshes quickly between trades.

What honest slippage looks like

When we re-price historical fills against book state instead of the mid, two things happen. Realized costs rise — sometimes uncomfortably — because the mid had been quietly subsidizing the numbers. And the variance of those costs collapses, because most of what looked like noise was depth and queue effects we simply had not been measuring.

That second result is the one that changes how a desk behaves. A cost you can predict is a cost you can size around. The goal was never a lower number on a slide; it was a number we can defend, and act on, in live markets.