Suppose a pension fund's treasury desk needs to move €40 million of bitcoin into custody by Friday. Put that size through a public order book and the fund pays for its own footprint: the book thins as the order eats through resting quotes, and the fund's own trade becomes the price move it was trying to avoid. The alternative is to trade off the book, over the counter, against a counterparty who takes the other side of the whole order at one price. What that involves, who bears which risk, and what the institution should have in writing before it happens, is the subject of this post.
The three ways a large order gets worked
Request for quote against a principal desk
In a request for quote, or RFQ, the institution asks one or more desks to price the whole order, and the desk responds as principal, taking the position onto its own book and warehousing the risk until it can lay it off. A large share of institutional-size crypto trading happens this way, bilaterally against a market maker's balance sheet rather than on an exchange order book. The desk is not acting as the client's agent; it is the client's counterparty, and its quote already contains the cost of hedging and unwinding the position it is about to hold.
A single dealer quote gives the client nothing to compare it against, so size is usually shown to two or three desks at once and the best price taken. Nothing about the process is displayed to anyone else.
Agency execution and algorithmic working
The alternative is to keep the order and work it, either through a human trader or an execution algorithm that slices it into smaller pieces across venues and time to reduce market impact, in the way a VWAP or implementation-shortfall algorithm would in equities. The broker never takes the position; it earns a commission or spread for execution skill rather than for warehousing risk. This suits a fund that can tolerate execution over hours rather than needing certainty at a single price, and it keeps the counterparty risk of the trade itself low, because the broker is never the other side.
Dark and conditional venues, where they exist
Some venues let institutions post non-displayed or conditional orders that interact only with other qualifying institutional flow, reducing the chance that a block order signals its own presence before it fills. In digital assets these venues are fewer and less standardised than their equity or foreign-exchange equivalents, and liquidity on any single one can be shallow enough that the client is still relying on RFQ or agency working to complete the size. Where a dark venue exists, it is worth using for the portion of an order it can absorb; it is rarely, on its own, the answer for a whole block.
Our earlier analysis of crypto market fragility is relevant here: displayed liquidity is often conditional and can thin precisely when a large order needs it, which is the structural reason block trades are routed away from the public book in the first place.
Pre-trade: credit, collateral and how a desk prices the risk
Before a desk will quote, it needs a credit line or pre-funded collateral in place. A principal quote commits the desk's own balance sheet to the other side of the trade, and it needs assurance the client will settle. Institutional counterparties typically operate under a bilateral master agreement setting credit limits, margin terms and default remedies, and post collateral, cash or crypto, against the line before trading rather than after.
The quote itself is not the mid-market price. A principal desk prices in the cost of hedging the position it is about to take on, the expected slippage of unwinding that hedge, the capital it must hold against the exposure, and a margin for the risk that the market moves against it before the hedge is complete. Size, volatility at the time of the request, and how liquid the specific asset is all widen or narrow that spread; a quote for a thinly traded token will carry materially more hedging cost than the same notional in bitcoin or ether. None of it is disclosed as a breakdown. It arrives as a single all-in price, which is why comparing quotes across desks, rather than accepting the first one, is the client's main protection.
The same balance-sheet logic, priced risk rather than a fixed fee, sits behind quoting on exchange as well; see our piece on how crypto market makers work.
Information leakage and last look
Asking for a quote is itself information. A desk that sees a client's request, even one it does not win, learns that size is coming to market, and if it trades on that knowledge the client's subsequent execution gets worse. Foreign-exchange markets have lived with this problem for longer, and the reference framework institutions use is Principle 17 of the FX Global Code, which covers last look: the practice of giving a quoting party a final window to accept or reject a trade request after seeing it. The Global Foreign Exchange Committee's Execution Principles Working Group report of August 2021 restates that last look, if used, should be a risk control mechanism to verify validity and price; that a market participant should not conduct trading activity that uses the information from the client's trade request during the window, whether pricing or hedging, unless a documented and disclosed cover and deal arrangement applies; and that a participant should disclose, at a minimum, whether and how price changes in either direction affect the decision to accept or reject, the expected period for making that decision, and the purpose for which last look is used. The Code is voluntary, not a statute, but it is the clearest articulation available of what fair use of a quote window looks like, and crypto desks quoting institutional size are measured against it by counterparties who trade both markets.
An institution should therefore ask any principal desk, before the first trade, whether it applies last look, what the window is, and its policy on trading during that window. A desk that will not answer has told the client something too.
Settlement: windows, payment versus payment, and the exposure in between
Agreeing a price is not the same as being paid. Between execution and final, irrevocable settlement there is a window in which one party can have delivered value without having received it, the risk foreign-exchange markets call Herstatt risk, after the failure of Bankhaus Herstatt on 26 June 1974: the bank's licence was withdrawn after it had taken in deutschmark payments from counterparties and before the corresponding dollar leg was paid out in New York. The industry's answer in foreign exchange was CLS, which has settled on a payment-versus-payment basis since 9 September 2002, releasing both legs of a transaction only if they can be made simultaneously, so that neither party is exposed to the other defaulting mid-settlement.
Crypto has no equivalent industry-wide payment-versus-payment utility for OTC settlement. Some settlement happens on-chain against a stablecoin or a fiat wire, which leaves a gap between the two legs unless both move atomically. Some happens inside a shared custodial network, where both counterparties hold assets with the same custodian and the transfer is a book entry, which closes the timing gap but concentrates risk in the custodian instead, and makes the identity of the entity that is legally custodian of record the thing that decides what happens to collateral if a counterparty fails; our analysis of crypto prime brokerage covers that structure. An institution agreeing a block trade should ask, specifically, how the two legs settle, whether there is any window in which it has delivered without receiving, and what happens to its collateral if the counterparty fails inside that window.
What a trade confirmation and best-execution record should contain
Regulated securities and derivatives markets require a written confirmation of every trade, with defined content, precisely because a bilateral deal leaves no public tape. Crypto OTC has no equivalent statutory requirement in most jurisdictions, so the content of the confirmation is whatever the counterparties agree, which means the institution has to specify it. At minimum that record should show: the instrument and quantity; the all-in price and, where the desk will disclose it, any fee or spread component; the time the quote was given, the time it was accepted, and, if last look applied, the outcome of that check; the settlement instructions for both legs and the time each leg actually settled; and the identity of the legal entity that was the counterparty, not merely the trading brand the client dealt with. A best-execution record should also show which desks were asked to quote, what each quoted, and why the winning quote was chosen if it was not simply the best price. Without that record, an institution cannot later show its own board, auditor or regulator that it sought a competitive price rather than accepting whatever a single relationship desk offered.
What is actually regulated, and where crypto OTC sits outside it
The regulatory position is uneven and moving, and the stages should not be conflated.
In the European Union, the Markets in Crypto-Assets Regulation has applied in full since 30 December 2024, and the transitional period for firms already providing crypto-asset services under national law before that date ran only to 1 July 2026, so authorised crypto-asset service providers are now inside it. In a Q&A answer dated 14 October 2025, the European Securities and Markets Authority addressed the boundary between execution of orders and dealing on own account, pointing to Recital 87 of MiCA: a provider acting as counterparty should still obtain the best possible result for its client. Whether a given arrangement is execution of orders is a fact-specific assessment of how the order is actually fulfilled and how a reasonable person would understand the provider's role, not of the label in the contract, and where a retail client is involved and the position is unclear ESMA points to the execution reading. Institutional desks operating under MiCA authorisation cannot assume that quoting as principal puts them outside best-execution scrutiny.
In the United Kingdom, the Financial Conduct Authority published its final rules for the cryptoasset regime on 30 June 2026, across five policy statements covering trading platforms, admissions and market abuse, stablecoin issuance, prudential requirements and the wider Handbook. The regime does not take effect until 25 October 2027, and the authorisation gateway opens on 30 September 2026. As this is written, the rules are adopted but not yet applying, and OTC desks quoting UK institutions today are not operating under them.
In the United States, there is still no comprehensive federal market-structure law for crypto spot and OTC trading. The Securities and Exchange Commission and the Commodity Futures Trading Commission issued a joint interpretive release on 17 March 2026 setting out a taxonomy that treats certain crypto assets as digital commodities rather than securities, but that release addresses classification, not who supervises spot or OTC trading in them, and CFTC oversight of crypto remains centred on derivatives rather than the cash market. The Digital Asset Market Clarity Act, which would have given the CFTC explicit authority over crypto spot markets, failed a Senate cloture vote on 15 September 2026 by 49 votes to 50, short of the 60 needed to proceed; the bill is technically alive but is not expected to pass in 2026. Until Congress legislates, a US-based OTC desk trading spot digital commodities is, for market-structure purposes, largely governing itself, subject to state money-transmitter and anti-money-laundering law rather than the best-execution and confirmation regimes that cover securities and, by the Code, foreign exchange. Anti-money-laundering obligations apply regardless; our piece on travel rule compliance in practice covers what a desk must collect and transmit on a counterparty.
The last-look principles and the payment-versus-payment logic described above are not law anywhere for crypto. They are imported by analogy, because crypto OTC has not built its own equivalent, and an institution relying on them is asking a counterparty to meet them voluntarily rather than exercising a right it can enforce.
What an execution policy should answer
Before the first block trade, an institution's execution policy should be able to answer, in writing:
- Which method, RFQ against a principal desk, agency working, or a conditional venue, will be used for a given size and asset, and who decides.
- How many desks are shown a request, and what happens to that information if a desk declines to quote.
- What credit or collateral must be in place before a desk will quote, and who holds it.
- Whether counterparty desks apply last look, what the window is, and what their policy on trading during that window is.
- How each leg of settlement actually happens, what gap exists between them, and what the exposure is if the counterparty fails inside that gap.
- What a trade confirmation must contain, and who reviews it against the quote that was accepted.
- Which jurisdiction's rules, if any, apply to the counterparty desk, and whether that status is in force, adopted but not yet applying, or absent altogether.
None of this is tax, legal, accounting or investment advice; it is a description of market structure and a list of questions an institution's own counsel and risk function should answer before size moves.