Buying Large Amounts of Bitcoin in South Africa: OTC vs Live Market Execution

A big enough Bitcoin order forces a choice most people never think about until it is in front of them. Do you lock a single price with an OTC desk, or work the order live across the market and take the average you end up with? Both are legitimate. They just optimise for different things, and the assumption that a desk always wins on price is one I have tested and found wanting.

I built the live execution engine before I ever leaned on a desk, so I have watched both sides do their work.

The two ways to fill a large order

Start with what actually happens to the price. A retail buy of R10,000 fills against the best available offers in about thirty seconds and nobody notices. A R7 million buy is a different creature. Sent as one market order it eats the sell orders nearest the current price, then reaches up for the next ones, and by the time the last satoshi lands you have paid a visibly worse average than the number on the screen when you clicked. That gap is slippage, and it is the whole reason the choice exists.

An OTC desk removes that problem by refusing to touch the public book at all. You tell the desk the size, it comes back with a firm price for the full amount drawn off live exchange data, and that number holds for a short window while you decide. No walking the book. No partial fills. The trade clears privately between you and the liquidity sitting behind the desk, and the public price only learns about it afterwards, if at all. I set out the full mechanics of that in how an OTC desk works, so I will not repeat them here. This piece is about the comparison.

Live execution takes the opposite approach. Rather than pay a spread to avoid the book, it works the order into the book carefully. It breaks the order into tranches fed in over time, minutes on a modest size and a couple of hours on something large, using patient limit orders on the maker side rather than sweeping the market in one take. Each tranche captures whatever liquidity is there. The price you finish on is the blend of them all.

Why the desk's price is not automatically the best price

Here is where the common assumption comes apart.

The desk quotes a fixed price set at one instant in time, and that price has the desk's own economics built into it. Its cost of sourcing the coins, its hedging, its own risk buffer and its profit margin all sit inside the number you accept, disclosed as a spread but still real. Even a tight spread is still a spread. It is the price you pay to make slippage and uncertainty someone else's problem, and for the right buyer that is money well spent. But it is not free, and a buyer staring only at how clean and tidy the quote sheet looks can easily miss where the value is actually being captured.

Live execution carries no such upfront spread. What it gives up in certainty it can win back in the market itself. Because the order fills across a window rather than in a single moment, it benefits from price variation while it works. If the market dips halfway through, the later tranches buy lower and drag your average down with them. The number that counts at the end is the volume-weighted average price, the VWAP, which is simply the effective price across every tranche once the trade is done. Not the price when the first order went in. The blend.

When I have measured that VWAP against the firm quotes I was offered on the same size, the live route came out ahead often enough that I stopped treating the desk as the default. The desks were not playing games. Their rates were genuinely competitive. It is just that a locked price cannot benefit from a market that moves in your favour while you buy, and a worked order can. That advantage is not guaranteed. In a market ripping upward through your execution window, the locked quote would have been the better call and the VWAP would have chased the price higher. Honesty demands that I say so.

What each method actually buys you

Think of it as buying different things with the same rand.

The desk buys certainty and discretion. You know your price before you commit, which weighs enormously if you are reporting a treasury decision to a board, striking a fund at a fixed valuation, or simply cannot stomach the position drifting while an order works. It also buys silence. A large buy landing on a thin local book is legible to anyone watching order flow, and on a market this size that information alone moves people. A desk keeps you invisible until the trade is done.

Live execution buys price efficiency and gives up the guarantee. You do not know your final average until the last tranche fills, and that uncertainty is precisely what lets the method capture movement the desk has priced out. It suits a buyer whose first priority is the lowest honest cost of acquisition and who can tolerate not knowing the exact number for the couple of hours it takes to get there.

There is a settlement wrinkle worth naming, because in South Africa the clock is set by rand, not Bitcoin. If your funds move between accounts at the same bank the clearing is near instant and the buy can trigger the moment the money reflects. If the rand comes from a different bank you are into a T plus one cycle, and the purchase only executes the following business day once the funds actually land. That is a plain operational fact whichever execution route you choose, and it is why proof of payment, not intention, starts the clock on any serious order.

The record-keeping difference, and why 2026 sharpens it

The two routes leave different paper trails, and for a South African that is not a footnote.

An OTC fill gives you one clean number. A specific quantity, at a specific rand price, on a specific date, through a single documented transaction. When your tax practitioner asks for your cost basis, you hand over one line. Live execution produces a blended cost basis stitched from every tranche, so the effective average is what counts and a decent provider hands you documentation of all of them. Slightly more to present. Entirely workable. I keep that record either way because SARS taxes a disposal on the difference between your rand acquisition cost and your rand proceeds, and a cost basis you cannot evidence is a cost basis you may not get to claim.

That weighs more from here on than it did a year ago. From 1 March 2026 South Africa moves under the Crypto-Asset Reporting Framework, which pushes detailed transaction data straight from licensed providers to SARS, tax residency, units transacted and the rand amounts, ready to be cross-referenced against whatever you declare. Providers and buyers alike are expected to hold the underlying audit trail for five years. Whichever way you execute, the documentation is no longer optional housekeeping. It is the thing that makes your position defensible. The same logic runs through everything I have written on Bitcoin and FICA compliance in South Africa.

When the desk is the right call

Some situations point clearly at OTC.

Where confidentiality is the priority, for a Stellenbosch family office moving a slice of a business sale into Bitcoin without broadcasting it to the market, the desk earns its spread on discretion alone. Where a fixed price counts more than the best possible price, for a board paper or a governance framework that demands a known number, certainty is the product you are buying and the desk sells it cleanly. Where the size genuinely outruns available exchange liquidity, in a thin market or a volatile stretch, working the book simply is not safe and the desk becomes the responsible route. And where a hard deadline is in play, a fund that must strike its price before a four o'clock valuation cannot afford an order that is still filling at ten to, so it locks the quote and moves on.

None of that makes OTC a poor product. It makes it a scalpel. The point is only that "my order is large" is not by itself the reason to reach for it.

When working it live is worth the patience

For a buyer whose single priority is the lowest effective acquisition price, and who does not need a locked number for reporting, a well-run live execution often deserves the first look. That is truer where the market is likely to be choppy through the window rather than trending hard one way, and where the execution infrastructure behind you is actually good rather than a person clicking buttons.

It is most obviously true for accumulation rather than a single event. Buying steadily over months through a systematic auto-buy is itself a form of live execution, averaging your price across a far longer window than any single trade could. Most real Bitcoin positions in this country were built that way, a monthly plan quietly converting rand while the owner got on with life, and I will point a client there before I point them at a desk if their horizon allows it.

The comparison that actually counts is between a professional live process and a good OTC deal. Not between a clumsy market sweep and a polished desk. Framed clumsily, the desk always wins. Framed fairly, it is a genuine contest.

The questions to answer before you transact

So the useful question is never "should I use OTC?" in the abstract. It is a handful of specific ones. Do you care more about a guaranteed price or the best achievable average? Do you need a locked number for a board or a fund strike? What is your time horizon, a single acquisition or years of accumulation? And how is the market behaving right now, drifting or running? Answer those honestly and the right method usually names itself. Where the answers pull in different directions, that is exactly the conversation I have with clients before a large purchase, and it is the same one I walk through in detail in buying large amounts of Bitcoin in South Africa.

You decide what you are optimising for. I make sure the execution actually delivers it.