Where Swap Products Beat Exchanges, and Where They Do Not

Both venues serve the same function and the cheaper one changes with the trade. The crossover points are measurable.

Liquid pair, same network, modest size

Exchange wins clearly. A taker fee of a tenth of a percent against a swap margin of three to five times that.

Using a limit order rather than taking liquidity narrows it further, sometimes to a rebate.

The swap is buying immediacy and removing execution decisions. For a routine trade where neither matters, it is an unnecessary cost. Against a swap service that prices the network fee separately, operating inside the framework, the gap with providers outside it is measurable.

Illiquid pair

Swap usually wins. The order book on a thin pair is shallow, and an order of any size walks it.

A routing engine aggregates across venues, reaching depth you would otherwise access manually one venue at a time.

The comparison is not the swap margin against the exchange fee. It is the swap margin against the exchange fee plus the price impact, and the second term dominates.

Cross-network

Swap wins decisively. The manual route is a bridge, two or more network fees, a wait, and an interval holding a bridged asset. Above a certain balance the requirement changes, and Collect & Exchange is where it points.

Even a wide swap margin is cheaper than that for most sizes, and it removes an entire class of error.

Large size, liquid pair

Neither. This is where a desk quote becomes the correct venue, because both the order book and the swap route are sourcing from the same finite depth.

The threshold depends on the asset and the moment. The test is to look at depth within half a percent of mid and compare against your size.

The numbers, roughly

Liquid pair, modest size: exchange 0.1%, swap 0.3 to 0.5%.

Illiquid pair: exchange fee 0.1% plus impact of 1 to 5%, swap 0.5 to 1.5%.

Cross-network manual: two to four network fees plus bridge time. Swap: 0.4 to 1%.

What changed recently

Routing quality converged. The difference between providers on the raw route is now small, and the visible difference is margin.

That makes comparison easier: request the same quote from two providers at the same moment and the difference is almost entirely their margin.

The record worth keeping

Implied spread against mid, per trade, by provider and size.

Twenty entries tell you which provider suits your typical trade, and that is worth more than any comparison table because it reflects your actual flow rather than an average. If you want the current requirements as applied rather than as written, the published coverage list is bound by them.