Digital Asset Investment Policy Statement for Family Offices

A digital asset investment policy statement settles, before any position exists, which accounts it covers, what may be held, at which custodian, who may instruct a transfer and above what size, and how the document itself gets amended. In my experience most drafts spend their length on allocation and give the authority and amendment clauses a sentence each, which is the wrong way round.

Part of our guide: Family Office.

The short version

• Write the authority section before the allocation section. Who may instruct, above what size, and who countersigns gets tested first.
• Compliance is measured against the facts as they stood when the decision was made (W.S. 4-10-908), so an overwritten file destroys the evidence.
• Define permitted assets by test rather than by name. A written eligibility test keeps deciding after the drafter has moved on.
• Fix one price source and one observation time inside the document. The IRS asks for a method applied “in a reasonable manner that is consistently applied” (Notice 2014-21, A-5).
• The amendment clause is the clause that gets used. With no route to a recorded exception, people work around the policy.

What the document has to decide

Scope, which doubles as a regulatory boundary. Name every legal owner the policy binds: the individuals, each holding company, each trust, any pooled vehicle. That roster matters twice, because the family office exclusion from the Advisers Act carries conditions:

“A family office is a company … that: (1) Has no clients other than family clients … (2) Is wholly owned by family clients and is exclusively controlled … by one or more family members and/or family entities; and (3) Does not hold itself out to the public as an investment adviser”

17 CFR 275.202(a)(11)(G)-1(b)

Read the scope list against the first condition whenever it changes. Adding a co-investor, a friend’s capital, or an executive’s own account puts the exclusion in play, and the policy dates when it happened.

Permitted assets, written as a test. Four questions do the work: will the named custodian hold it, does an independent price exist at the policy’s observation time, does any lockup or unbonding period apply, and does holding it require network participation. A frozen list of names falls behind what the office already owns.

The custody standard, stated as a standard. Describe the arrangement rather than the provider: titling in the legal owner’s name, segregation, who at the office is an authorized person, and what qualifies a location to hold assets. Then keep every permitted location on one dated inventory and make adding a location an amendment.

Limits, with a denominator, a date, and a consequence. Whatever number the family and its advisers arrive at, four things travel with it: the quantity measured, what it is measured against, the observation date, and what a breach triggers. The fourth is the one left out, and a breach that triggers nothing describes the portfolio rather than governing it. Say whether it forces a review, a trade, or a signed memo explaining the decision to hold. This page takes no view on what any limit should be; the allocation policy belongs to the family’s own advisers.

Transaction authority, by size and by destination. On-chain settlement is final, so approval has to be recorded before the signature. Write tiers: a routine level with a stated ceiling, a level needing a second named approver, and a level needing the investment committee. Then add a destination rule, since a whitelisted address and a fresh counterparty differ at the same size.

Valuation, tax, and what the policy commits to report

Valuation is where a policy usually inherits somebody else’s answer without saying so. The federal standard is method-based:

“If a virtual currency is listed on an exchange and the exchange rate is established by market supply and demand, the fair market value of the virtual currency is determined by converting the virtual currency into U.S. dollars … in a reasonable manner that is consistently applied.”

Notice 2014-21, A-5, modified on other grounds by Notice 2023-34

Consistency across entities is what fragments in a family office. Name the source, the time of day, and the time zone, then apply that one method to the trust report, the LLC books, and the consolidated statement. For trusts the obligation is statutory: a Wyoming trustee sends qualified beneficiaries a report at least annually listing the trust assets and, “if feasible, their respective market values” (W.S. 4-10-813(c)). Here “if feasible” is doing real work, and the policy is where an office settles what it means.

Three tax facts set the reporting the policy is promising. Basis follows the location, since Rev. Proc. 2024-28 allocated unused basis to a wallet or account as of January 1, 2025 and tracks it wallet by wallet after that, so every internal transfer the policy permits is a records event. Income arrives on the protocol’s schedule, because Rev. Rul. 2023-14 puts validation rewards in gross income in the year the taxpayer gains dominion and control, so permitting staking means naming who books the entry. And Form 1099-DA now carries broker gross proceeds, with a digital asset acquired after 2025 for cash treated as a covered security, so the office’s numbers get read against a form it did not prepare. Set the reconciliation cadence here.

Delegation, exceptions, and the version in force

Where trust assets are involved, the policy performs a job the statute already describes. A delegating trustee must exercise reasonable care, skill and caution in “Establishing the scope and terms of the delegation, consistent with the purposes and terms of the trust”, and in “Periodically reviewing the agent’s actions” (W.S. 4-10-909(a)). The first of those is the policy itself. The second is a cadence, and a delegation with no stated review interval fails on its face. If the office engages an outside manager, scope, terms, and review interval belong in the document rather than in the engagement letter alone.

Exceptions need a route: who may authorize a deviation, how long it runs, where it is recorded, and who reviews it next. Amendments need who proposes, who approves, on what vote, and where superseded text is kept. Both matter because compliance with the prudent investor rule sits on that record:

“Compliance with the prudent investor rule under this article is determined in light of the facts and circumstances existing at the time of a trustee’s decision or action and not by hindsight.”

W.S. 4-10-908

An office keeping only the current text has thrown away its proof of what was in force when the decision was made.

What I actually see

The first pattern is the policy written around today’s relationships. The operative clauses name the current adviser and the current custodian by company name, so one provider change leaves half the document pointing at a firm nobody uses.

The second is the limit whose breach has never been written down. On somebody’s arithmetic it has been crossed more than once, and the file records no observation, no discussion, and no reason for holding. Each of those calls may well have been right. The office cannot show that anyone made them.

The third is the version problem: one file in a shared drive, overwritten repeatedly, with no adoption date, no signature, and no trace of what it replaced. When a beneficiary or an incoming trustee asks what the policy required in a given year, nobody can answer.

The check I would run takes twenty minutes and needs one person from outside the family. Hand them the policy alone, with no meeting notes and nobody to ask, and have them answer five questions from the text: which legal owners does this bind, who may move assets out of the largest holding location and above what size, which price source at which time, what happened the last time a limit was crossed, and who signed the version in force. Every question that sends them to the phone is a clause that does not exist yet.

Where this goes wrong

The document describes a governance structure the office never adopted.

The specific failures: a policy circulated by email and never adopted by a recorded vote, so it has no moment of birth. Providers named as parties rather than as roles, which turns every relationship change into an amendment nobody executes. A permitted asset list frozen at the drafting date while current holdings sit outside it. Limits measured against a total that includes pledged or locked units the office cannot sell. A staking permission with no owner for the income entry. Delegation with no stated review interval, which fails W.S. 4-10-909(a)(iii). Coverage extended to someone outside the family client definition, which puts the office’s Advisers Act position in question. And the version control failure that makes every other defect unprovable: one file, overwritten, undated.

The decision rule

1. Fix the scope first. List every legal owner the policy binds, and confirm each sits inside the family client definition the office relies on.
2. Write an eligibility test for permitted assets, so a new asset is decided by the test rather than by an amendment.
3. State the custody standard as a standard, park provider names in an appendix, and treat adding a location as an amendment.
4. Give every limit a denominator, an observation date, and a consequence, and name who records the breach.
5. Tier transaction authority by size and destination, with approval recorded before the signature.
6. Name one price source, one time of day, and one time zone, then apply that method across every entity in scope.
7. Put the delegation scope and the review interval in the document, because the statute asks for both and an engagement letter answers one.
8. Sign and date every version and retain the superseded ones, since compliance is judged on what was in force at the time.

Where this sits

The policy is the join between decisions this library treats separately. Governance sets who decides and how often they meet. Custody settles who can move assets and what evidence exists that the entity controls them. Trusts supply the fiduciary standard the document is measured against. Start with the scope list and the authority tiers, because those two sections make the rest enforceable.

These questions cross professional boundaries, and the join is what fails. The attorney drafts the trust language and the entity documents, the CPA owns basis and the filing calendar, custody is set up by whoever solved the operational problem first, and the investment adviser writes the allocation section. Each is right inside their own frame, and nobody owns the seam, so the policy ends up naming a valuation method the CPA never saw, a delegation the trustee cannot evidence, and an authority tier the custodian’s file contradicts. Someone has to read all four documents against each other, and in most families that person has to be appointed rather than assumed.

Sources

• 17 CFR 275.202(a)(11)(G)-1, Family offices
• Wyoming Uniform Trust Code and Uniform Prudent Investor Act, Wyo. Stat. Ann. §§ 4-10-813, 4-10-908, 4-10-909
• IRS Notice 2014-21 (I.R.B. 2014-16), modified by Notice 2023-34
• IRS Rev. Proc. 2024-28 (I.R.B. 2024-31), digital asset basis allocation
• IRS Rev. Rul. 2023-14 (I.R.B. 2023-33), staking rewards
• IRS, About Form 1099-DA
• IRS, Digital assets

Related

• Crypto governance for family offices
• Family office crypto allocation policy
• How should a family office investment committee review crypto?
• Crypto due diligence for family offices
• Digital asset reporting dashboard for family offices
• Crypto family offices

Last updated: 3 August 2026.

This article is general education, not legal, tax, or investment advice. It recommends no allocation, position size, permitted asset, or course of action, and takes no view on any digital asset. A written policy can reduce certain governance risks but does not eliminate them. Talk to a qualified attorney, CPA, and investment adviser about your own situation.