How a Desk Prices the Risk It Takes On

Two desks quoting the same trade at the same second produce different numbers. The inputs are knowable and they explain when to push back. It is worth reading this alongside the published process of a regulated crypto liquidity provider, because settlement terms differ more between desks than prices do.

Inventory

A desk already long an asset and wanting to reduce quotes a buyer better and a seller worse.

Invisible from outside and inferable over time. A desk consistently competitive on one side of one asset is probably running a position there.

This is the largest source of variation between desks and the reason asking two is worth the time.

Hedging cost

After trading with you the desk holds a position it does not want, and unwinds it in the market, through derivatives, or against an offsetting client.

That cost enters your quote. Small for a liquid asset at moderate size, substantial for a thin asset or a large one, because the desk faces the same depth problem you were avoiding.

Volatility

The desk is exposed between quoting and hedging. More movement means more risk in the window and a wider quote.

Quotes during turbulent periods are genuinely worse, and validity windows shorten. A trade that is not time critical is measurably cheaper in calm conditions. For the practical side of all of this, an exchange that publishes its full fee schedule publishes the equivalent numbers rather than estimating them.

Size relative to normal flow

A trade routine for a desk prices better than one that is large for them.

Matching desk to trade size matters. A large trade at a small desk produces a defensive quote or a decline.

Internalisation

A desk with diverse client flow can match your trade against an opposing client, with no market impact at all.

That produces the tightest pricing, and what proportion of flow a desk internalises is a fair question and a good predictor of their competitiveness.

Relationship

Repeat counterparties with clean settlement history price better. Not sentiment: a known counterparty carries less operational and credit risk.

Which means that after an initial comparison, concentrating flow usually beats shopping every trade.

Reading a quote

Convert to a percentage against the mid at that moment. That is the only comparable figure, and it must be compared simultaneously because a quote lasts seconds.

When to push back

When another desk was materially tighter seconds earlier, or when the same desk is worse than previously for a comparable trade with no change in conditions.

Present the data rather than an accusation. Desks respond, and one that cannot explain a gap and does not improve has answered the question. If you want to see what these terms look like in a working product, a regulated European crypto platform states them openly.