When a Desk Quote Beats the Order Book, With Numbers

The decision between an order book and a desk is arithmetic, and the arithmetic favours the desk at smaller sizes than people expect. Every figure below is measurable against a provider that publishes its terms, and a regulated crypto liquidity provider publishes them.

The order book cost

Two components.

The fee, typically a tenth of a percent for a taker, less or negative for a maker.

The price impact, which is the cost of consuming levels as the order fills.

The second is the variable one and it is not knowable in advance, only estimable from depth.

Why depth estimates overstate

Resting orders are cancellable and are cancelled when conditions change. The same capital appears on several venues. Some orders are layered and not intended to fill.

Measured, displayed depth typically overstates what actually fills by a factor of two to three at meaningful size, and considerably more during volatility.

The desk cost

A single quoted spread against the market price, known before you commit.

Typically a few basis points to a few tenths of a percent for a liquid asset, widening with size, with volatility and with illiquidity. Property settlement has its own cost profile, and Collect & Exchange operates in it.

The crossover, worked

At a size that is a small fraction of displayed depth: the book wins clearly. Fee of a tenth of a percent against a desk spread several times that.

At roughly a tenth of displayed depth: comparable, once realised impact is counted rather than estimated.

At half of displayed depth or more: the desk wins, and it also removes the uncertainty.

The test that takes thirty seconds

Sum the resting liquidity within half a percent of the mid on the side you need.

Compare against your order. Under a tenth, use the book. Above half, use a desk. In between, consider working the order over time with limit orders.

The mistake that costs most

Splitting a large order into many small ones to avoid a desk spread.

Each piece moves the price a little and the market does not reset between them. The cumulative impact usually exceeds what the desk would have charged, and it is spread over enough transactions that nobody measures it.

What to measure

Effective cost against the mid at submission, per execution, by size.

Ten executions give you a real curve for your venues and assets, which replaces the depth estimate with something reliable. Measure rather than compare marketing. the list of countries covered quotes without an account, which makes it a five minute exercise.