Quant Memo
Core

Depth At Touch And The Shape Of The Book

The best bid and ask are only the first line of a much deeper story. Plotting size against price away from the touch reveals how much a trade will actually move the market.

Prerequisites: Order Book Mechanics

"Depth at touch" is the size sitting at the single best bid and best ask — the two numbers everyone quotes. But a trade of any real size never fills at the touch alone; it walks the book, eating through price levels one after another. The depth profile is the full picture: size available at every price level, not just the first one.

Building a profile and walking an order through it

Suppose the offer side of a book looks like this:

PriceSize at levelCumulative size
50.02300300
50.03500800
50.042001,000
50.059001,900

A trader needs to buy 1,200 shares immediately (a market order). It sweeps: 300 @ 50.02, 500 @ 50.03, 200 @ 50.04, and then the remaining 1,2001,000=2001{,}200 - 1{,}000 = 200 shares come from the 900 sitting at 50.05. The order clears at 50.05, not at the 50.02 touch — the average fill price is

pˉ=300(50.02)+500(50.03)+200(50.04)+200(50.05)1,200=50.0308\bar{p} = \frac{300(50.02) + 500(50.03) + 200(50.04) + 200(50.05)}{1{,}200} = 50.0308

versus a touch price of 50.02 — 1.1 cents of slippage purely from walking the book, before any information effect at all. Notice the jump from 50.04 to 50.05: only 200 shares deep at 50.04 means any order over 1,000 shares necessarily prints above 50.04, a cliff a depth-only-at-touch view would never reveal.

50.02 50.03 50.04 50.05 300 500 200 900 200 used 1,200-share buy sweeps 3.7 levels deep
The amber slice is the last 200 shares of a 1,200-share order, taken from the 50.05 level after the first three levels (1,000 shares) were exhausted.

Depth at touch tells you the price of the next share. The full depth profile tells you the price of the next thousand shares — and it's the profile, not the touch, that determines market impact.

Where it's used. Pre-trade cost estimates for execution algorithms are built directly from the depth profile: given an order size, walk the book to get an expected average fill price before sending a single share. Depth profiles also feed Order Book Features For Machine Learning — the shape of the near-touch levels (steep versus flat) is one of the more predictive short-horizon signals in limit order book data. And market makers watch the profile, not just the touch, to judge how much size they can absorb before needing to move their own quotes.

A displayed depth profile only shows displayed liquidity. Iceberg orders and reserve size mean the real depth at a level can be several times what's shown — a book that looks thin at 50.04 might not be. See Detecting Hidden Liquidity for how to infer the difference.

Related concepts

Practice in interviews

Further reading

  • Bouchaud, Bonart, Donier & Gould, Trades, Quotes and Prices (ch. 1-2)
ShareTwitterLinkedIn