Investors often begin digital-asset conversations with conviction: Is bitcoin digital gold? Is ethereum a settlement layer? Which narrative will win? Those questions are interesting. They are not first. First are custody questions—because a correct thesis held in an inaccessible, poorly controlled, or single-point-of-failure wrapper is not an investment. It is a hostage situation with a ticker.
Access risk is portfolio risk
Price charts assume you can act. Custody determines whether that assumption holds. Keys, multisig policies, custodian legal structures, exchange withdrawal rules, and operational continuity all sit between you and the asset. Traditional markets hide some of this behind brokers and depositories. Digital assets put more of it in your face. That visibility is a gift if you use it.
I ask who can move funds, under what controls, with what recovery path if a person, device, or vendor fails. I ask how withdrawals behave under stress. I ask whether the custody setup matches the size. A research wallet and a mandate-sized allocation are different animals wearing the same token name.
Conviction without operations is incomplete
Teams sometimes treat operations as a later workstream. In practice, operational design decides what sizes are even ethical to run. If your controls cannot support a position, your maximum conviction is capped by process, not by opinion. That is not conservatism for its own sake. It is refusing to let rhetoric outrun infrastructure.
Counterparty concentration
Parking everything on one venue because it is convenient recreates classic counterparty risk with modern branding. Diversifying venues without understanding each venue’s controls can recreate complexity risk. There is no free lunch—only trade-offs you should name.
What traditional desks already knew
Prime brokerage, collateral schedules, and cash wiring protocols trained generations of investors that access and legal finality matter. Digital-asset allocation should import that seriousness rather than reinventing casual habits. “Not your keys, not your coins” is a slogan. A written custody policy with roles, tests, and incident drills is a practice.
I also separate research holdings from portfolio holdings psychologically. Studying a mechanism can justify a tiny exploratory position. It does not justify skipping custody hygiene. Curiosity is cheap. Negligence is expensive.
A sequencing I recommend
Incident readiness is part of custody
Custody is not only where keys live. It is how you respond when something goes wrong: lost device, compromised employee laptop, vendor outage, suspicious withdrawal pattern. Write an incident path while everyone is calm. Name who can halt transfers, who can escalate to a custodian, and how you communicate without creating new attack surface. An untested incident plan is a wish.
For institutional setups, legal agreements matter as much as cryptography. Bankruptcy remoteness, segregation language, and audit rights are dull until they are decisive. Read them with the same seriousness you apply to a valuation model.
Human factors
Many custody failures are social: rushed approvals, shared credentials, “temporary” policy exceptions that become permanent, and hero operators who are single points of failure. Rotate duties where possible. Avoid monocultures of knowledge. Train people that speed without checks is not professionalism.
Self-custody and delegated custody both can be rational. Mixing them without a map is not. Know which assets sit where, why, and how that choice changes your operational risk budget. Then—and only then—let conviction debates consume meeting time.
A correct market view held inside a fragile custody wrapper is still a fragile portfolio. Sequence the work so infrastructure leads narrative.
Map custody and exit paths. Define sizes consistent with those paths. Then debate the thesis. If the thesis is strong and custody is weak, fix custody or reduce size. If custody is strong and the thesis is weak, you still should not buy. Both gates matter. Only one is fashionable.
Sizing follows the weakest control
Your maximum position should be bounded by the weakest relevant control: key management, vendor resilience, legal clarity, or human backup. A brilliant thesis cannot ethically outrun that bound. If you want larger size, improve controls first. That sequencing keeps ambition aligned with stewardship.
Review custody quarterly even when nothing “happened.” Calm quarters are when drift in people, devices, and vendors accumulates. Conviction debates can continue in parallel. They should not jump the queue.
Markets reward narratives. Risk management rewards boring sequencing. Start with custody. Conviction can wait its turn.