Bitcoin Position Sizing for South African Family Offices: A Framework for Large Holdings

A percentage rule is where retail Bitcoin advice starts and where family office advice ends. The 1% to 5% band that suits a private investor tells a R500 million family office almost nothing about how to actually get the position on, how to hold it and how to hand it on. The number is the last decision, not the first. Everything that decides whether the holding survives sits underneath it: execution, custody, structure and governance.

Where the percentage stops being the point

Start with the arithmetic, because it is the last time the arithmetic is simple. A family office carrying R50 million takes 5% and lands on R2.5 million of Bitcoin. That is a straightforward buy, held in one place, managed with barely a second thought.

Now scale it. The same 5% against R500 million is R25 million, and against R2 billion it is R100 million. The percentage did not change. Everything else did.

At R2.5 million you place an order and you are done. At R100 million a single market order would eat through the visible liquidity on the book and fill the rest at steadily worse prices, so the purchase has to be staged over months through an over-the-counter desk rather than dumped onto a public exchange. One custodian becomes a concentration you cannot defend. A trust deed that never mentioned crypto assets becomes a live problem. The question quietly changes from what percentage should Bitcoin be into at what point does this position need real infrastructure to hold it, and in practice that point arrives a good deal lower than most families first expect. This is the same lens I apply in my framework for sizing a Bitcoin position, scaled up to family office numbers.

Execution: buying R100 million without announcing it

The single biggest difference between a private client and a family office is that the family office can move the price against itself simply by trying to buy. Liquidity is generous for a R1 million order and thin, at any single price, for a R50 million one.

The operating model I run for size solves this with an escalation rule rather than a hero trade. Small tickets fill in seconds on exchange. Large tickets get staged, and the moment the size or the time it would take to fill starts threatening the price, the order moves to an OTC desk that quotes a firm price on the whole amount and settles it in one clip. The spread is a little wider than the screen price. In exchange you get certainty on a nine-figure fill and you do not telegraph your buying to every other participant on the book. SimplB executes at any size this way, so the family office never has to choose between the price it wants and the size it needs.

None of this is a one-off event. A family office accumulating a large position spreads the buying over time on purpose, because dollar cost averaging into a volatile asset buys more when the price is low and less when it is high, and it removes the temptation to time an entry that nobody times well. The same discipline runs in reverse when a position is eventually trimmed.

Custody: no single point of failure

Whoever holds the keys holds the Bitcoin, and at family office scale that sentence stops being a slogan and becomes an operating constraint.

A single hardware wallet with one keyholder is fine for a hobbyist and entirely wrong for an institution. The position has to survive the death, incapacity, emigration or plain unavailability of the person who set it up, and it has to resist theft and coercion at the same time. Those requirements pull against each other, which is exactly why the institutional answer is a 2-of-3 multi-signature vault. Three keys exist, any two can move funds and no single key can act alone, so one lost key is a recoverable inconvenience rather than a catastrophe. The keys sit on hardware from different manufacturers and live in different places, because a supply chain problem with one device should never compromise the quorum. That is the principle the Vault I run is built on: the family holds two keys on its own devices, I hold the third as the regulated recovery leg through CAEP Asset Managers (FSP 33933) and nobody moves a coin unilaterally.

Some very large holdings go a step further and split the position across more than one custody arrangement, so that no single institution, jurisdiction or team is a chokepoint. That is a genuine belt-and-braces move and it is not free. Every extra custody leg is another set of keys to generate, back up on steel in a separate location, document and verify, and the golden rule that no two seed backups ever sit together has to hold across all of them. Complexity is the price of resilience here, and the honest advice is to buy only as much of it as the size actually warrants.

The custody design is where I start every family office engagement, well before the number is settled, for the same reason I set out in why Bitcoin for a family office is a governance decision first. A well governed small holding outlasts a poorly secured large one every time.

Which entities should hold it

A family office rarely holds everything in one name. The trust might hold the bulk, a company might hold treasury capital that belongs inside the business and individual family members might hold a slice in their own names, each wrapper solving a different problem.

The catch is that each wrapper taxes Bitcoin differently and each hands it down differently, and choosing badly is expensive to unwind because moving appreciated Bitcoin between them is itself a taxable event. A trust buys continuity at the cost of a high capital gains rate. A company suits capital that genuinely lives in a business. Personal name is simplest and cheapest until estate duty starts to bite. I weigh those trade-offs in full in holding Bitcoin in a company, trust or personal name, and the short version for a family office is that the split should be decided before the growth, not renovated after it.

Multi-entity structure earns its complexity only once the position is large enough to justify running custody, documentation and SARS compliance three times over instead of once. Below that threshold a single well built trust is usually cleaner. Above it, the flexibility is worth the administrative weight, and the design of it belongs alongside the estate plan I describe in passing Bitcoin down across generations in South Africa.

The horizon changes the psychology

A private investor tends to think in five to ten year windows. A family office thinks in twenty five to fifty year ones, and that single shift changes what a sensible position feels like.

It helps to be honest about what those decades contain. Since it began trading widely, Bitcoin has suffered four drawdowns from a peak of between 76% and 90%, and a family holding this asset across a generation will live through more of them. The reason the long horizon is an advantage rather than a terror is that anyone who has held Bitcoin for four years or more has historically ended up in positive territory, even having bought at the very top, and over a ten year scale the asset has compounded at extraordinary rates. A family that cannot be forced to sell can treat a brutal drawdown as weather rather than a fire alarm, which is why sizing the position with capital the family will not need for years is the whole game. The one rule that counts for more than the percentage is never to be a forced seller.

Governance: the written policy that removes emotion

At this scale Bitcoin deserves the same governance any serious reserve asset gets, and the tool for that is a written policy the investment committee actually follows.

A proper Bitcoin reserve policy states the purpose of the holding, the maximum allocation relative to total or surplus capital, the expected holding period and the exact trigger conditions for buying more or trimming. It names who may instruct a transaction and at what signing threshold, so a trustee or managing partner can rebalance inside agreed parameters without convening the whole family every time the price lurches. That is what separates a governed reserve from a sideline bet. The multisig structure enforces it in a way paperwork alone never could, because it turns segregation of duties into mathematics: the person who initiates a transaction and the person who approves it are different people holding different keys, and every signature is recorded on chain as an audit trail an auditor can read.

The documentation that sits under all of this is unglamorous and decisive. You need custody agreements and key management protocols on file. You need a succession plan for what happens if the primary keyholder is suddenly unavailable, and a regular verification of the holdings against the chain. It is the file nobody wants to build and everybody is grateful for at the worst possible moment.

What it looks like at three scales

To make it concrete, picture three families walking into the same conversation.

The first carries R50 million and wants a 5% position, so R2.5 million. That is a single licensed South African custodian, most likely a trust, bought over three to six months, reviewed once a year. No multi-entity gymnastics, no OTC drama, no dedicated team. The infrastructure should match the risk and here the risk is modest.

The second carries R500 million and the same 5%, so R25 million. Now the picture thickens: a documented custody arrangement built to survive a single institutional failure, OTC execution for the larger tranches, a written reserve policy, quarterly governance reviews and an annual verification of the holdings. The buying is staged over the better part of a year rather than rushed.

The third carries R2 billion and looks at 5% to 10%, somewhere between R100 million and R200 million. Everything above, plus a custody split designed so no single point can fail, a likely multi-entity structure spanning a trust alongside a company and personal holdings, an accumulation that runs across many months of staged OTC buying, an investment committee that meets quarterly and a succession and transfer file built for the generation that inherits it.

The pattern across all three is the same. The percentage is the label on the box. The structure is what sits inside it, and at family office scale that structure is the work. Design that structure before arguing about the number, then book a family office structure session to build it in that order.