Total Execution Cost: Desk, Book and Swap Compared
One framework covering all three venues, with the thresholds that determine which is correct.
Three venues, one comparison framework. The correct choice changes per trade and the decision takes thirty seconds.
The framework
The measurement is straightforward against a crypto OTC trading platform and difficult against providers that quote a single number.
For every route, compute the total cost as a percentage against the market mid at the moment of decision.
That single figure makes all three comparable. Fees, spreads and impact all collapse into it.
Order book
Cost: the fee, plus price impact.
The fee is known. The impact is not, and it depends on your size relative to depth.
Cheapest when: the pair is liquid, your size is a small fraction of displayed depth, and you can use a limit order.
Typical: 0.1 percent or less at small size, rising sharply as size approaches depth.
Swap
Cost: the provider’s implied spread, known before you accept.
Cheapest when: the pair is thin, or the assets are on different networks, or you want certainty rather than the last basis point.
Typical: 0.3 to 1 percent same-network, 0.4 to 1 percent cross-network where the alternative is far worse.
Desk
At company scale the cost structure changes, and a platform built for institutional allocations prices it accordingly.
Cost: the quoted spread, known before you accept.
Cheapest when: size is large relative to depth, or the trade needs to be private, or certainty on the whole amount matters.
Typical: a few basis points to a few tenths, widening with size, volatility and illiquidity.
The threshold test
Sum the resting liquidity within half a percent of mid on the side you need.
Under a tenth of it: book.
A tenth to a half: work it over time, or get a desk quote and compare.
Above a half: desk.
The adjustment for reality
Displayed depth overstates what fills, typically by a factor of two to three and more during volatility.
So the thresholds above are conservative, and the desk becomes correct at smaller sizes than the raw numbers suggest.
The mistake that costs most
Splitting a large order into small pieces to avoid a desk spread.
The market does not reset between pieces and the cumulative impact usually exceeds the spread avoided, spread across enough transactions that nobody measures it.
What to record
Effective cost against mid, per execution, by venue and size. Ten entries produce a curve and the curve replaces the estimate. To check any figure above against a real provider, a crypto exchange with published fees publishes the equivalent terms.
Fees move with network demand. Any figure here reflects the date shown above and should be re-checked before you act on it. Where a measurement was taken by hand, the article says so.