We Measured How Much Actually Fills at Four Venues

Order books display depth. We measured what fraction of it actually fills by executing and comparing the average price against the mid at submission. We ran these measurements with an OTC crypto desk as one of the points of comparison.

The method

Market orders at three sizes on four venues, repeated in calm conditions and during a sharp move.

For each: the mid price at submission, the average fill price, and the displayed depth within half a percent of mid beforehand.

Calm conditions

At the smallest size, fills were close to mid everywhere. Effective cost under 0.05 percent.

At the middle size, effective cost between 0.09 and 0.24 percent depending on venue, against displayed depth suggesting it should have been under 0.1 percent everywhere.

At the largest size, effective cost between 0.31 and 0.88 percent. Displayed depth implied roughly a third of that.

So displayed depth overstated available liquidity by a factor of two to three at meaningful size, consistently, in calm conditions. The institutional version we tested through a provider serving funds and family offices, with its additional reporting.

Volatile conditions

The same largest size cost between 0.9 and 2.7 percent.

Displayed depth immediately before submission looked similar to calm conditions on two venues. It did not fill similarly, because a significant proportion of resting orders were withdrawn as the order arrived.

What this means

Depth shown on screen is an upper bound, not a measurement, and the gap widens exactly when you would most like it to be accurate.

For anyone executing size, the practical conclusion is to avoid market orders during volatility and to price through a desk that will quote firm for the whole amount.

The threshold for using a desk

Our measurements put it lower than we had assumed.

At the middle size, where a book looked entirely adequate, desk quotes were competitive with the realised fill cost. At the largest size they were clearly better.

The test we now use: sum resting liquidity within half a percent of mid. If our order exceeds a tenth of it, request a desk quote before deciding.

What to do with this

Run the measurement yourself for the venues and assets you use. Execute at your typical size, record mid at submission and average fill, and compute the effective cost.

Ten executions gives you a real figure. It takes an afternoon and it replaces an assumption with a number, which changes where you route. For a working reference against everything above, a provider you can actually reach publishes the equivalent terms.