The ZAR carry trade is a spread problem before it is a rates problem. Hear me out. I pulled the broker sheet this morning and the receipt looks like this — Exness quotes a 1.0-pip average on the majors on its standard book and 0.1 pip on Pro; FBS shows 0.7 average and 0.0 on Pro; HF Markets shows 1.2 and 0.0. Nothing on that sheet is a ZAR quote. That is the entire point. Before you touch a Societe Generale carry-and-gold thesis on the rand, you are already trading through a cost stack that the thesis never mentions and most retail write-ups quietly refuse to price.
What the Numbers Actually Say
Look at the sheet the way a desk analyst reads a term sheet — line by line, not headline by headline.
Exness was founded in 2008. Minimum deposit one dollar. Maximum leverage 2000:1 on the offshore book. The standard account quotes an average 1.0-pip spread on majors. The Pro account quotes 0.1 pip. The regulator stack is FCA at the top and then a sprawl of secondary licences — CySEC, FSCA, CBCS, CMA Kenya, FSA, FSC BVI, FSC Mauritius, JSC Jordan. Withdrawals are marketed as instant. The MetaTrader 4 and 5 rails are the default.
FBS: founded 2009. One-dollar minimum. Leverage 3000:1 — the highest number on the sheet. Standard spread 0.7 pips average, Pro spread 0.0. Tier-1 supervision through ASIC; secondary through CySEC and FSCA. Withdrawal marketed as instant to one day. MT4 and MT5, plus their own FBS Trader.
HF Markets: founded 2010. Five-dollar minimum, 1000:1 leverage. Standard 1.2 pips average, Pro 0.0. Regulator stack: FCA, CySEC, FSCA, DFSA, FSA. One-day withdrawals. MT4, MT5, their own app. The company positions itself around 1,200-plus instruments and Islamic accounts.
FXTM: 2011, ten-dollar minimum, 2000:1 leverage. Standard 1.5 pips average — the widest on the sheet — and 0.1 on Pro. FCA at the top of the regulator stack, then FSCA and FSC. One-to-three-day withdrawals.
AvaTrade: 2006, hundred-dollar minimum, 400:1 leverage — the most conservative number on the sheet, and not by accident. The regulator list is ASIC, FSCA, ADGM, CBI, FSA. Withdrawals one to three days. AvaOptions sits alongside MT4 and MT5. The desk's stated weakness on its own factsheet is that scalping is prohibited.
None of these entries publishes a USD/ZAR spread on the sheet you were just handed. Every ZAR carry-trade write-up assumes the reader is trading it. That mismatch is the whole story.
The Societe Generale angle in retail summaries always reduces to two nouns — carry and gold. The rand pays yield. South Africa exports gold. Add them together and you get an "outperformer" caption. What the summaries do not say is the execution cost between the caption and the trade you actually put on. The FX majors on the broker sheet are the cheapest instruments any of these firms quote. USD/ZAR is not on the majors sheet, and when it appears on the extended sheet the spread is not 1.0 pip. It is not 1.5 pips either.
What Nobody Mentions
Here is what an Exness Standard 1.0-pip EUR/USD average and an Exness Pro 0.1-pip EUR/USD tell you about the ZAR quote you will actually receive: almost nothing directly, and almost everything indirectly.
The commission-model history matters here, because the reader inherits it whether they know it or not. There were two lineages that converged into the modern retail broker.
The first lineage was the transparent commission desk. You paid a fixed number per lot — typically three to seven dollars per side per standard lot — and the broker quoted you an interbank spread, thin. IC Markets and Pepperstone built their institutional-facing books around this model. On EUR/USD the raw feed might show 0.0 to 0.1 pips; the commission was disclosed, itemised, and testable against the tape.
The second lineage was the zero-commission book. XM zero commission. Exness zero commission. FBS zero commission. The cost did not vanish — it moved. The spread you saw quoted was the spread the broker sold you plus a mark-up. On the majors the mark-up was small and often defensible. On the minors and exotics it was neither disclosed nor small. And the ZAR quotes live in that second bucket.
The historical compliance record on this is clear enough. When the FCA in the UK and CySEC in Cyprus tightened disclosure rules through the mid-2010s, the firms that had built their marketing around "zero commission" did not renounce the model — they added a Pro or Raw account tier alongside it, quoted 0.0 to 0.1 pips on majors on that tier, and let the retail Standard account continue to run the wider spread. Look at the sheet again. Exness Pro 0.1. FBS Pro 0.0. HF Markets Pro 0.0. FXTM Pro 0.1. The Pro tier is the transparent-commission model living inside a zero-commission brand.
Now the ZAR question. USD/ZAR on a retail Standard book at any of these firms is not going to trade like EUR/USD. The instrument is thinner, the interbank spread is wider, and — critically — the mark-up rules that apply to the majors do not extend down the pair list at the same ratio. If a firm marks up EUR/USD by roughly 0.9 pips over its Pro book, it does not mark up USD/ZAR by 0.9 pips. It marks up by whatever the desk risk model justifies against the underlying volatility, which is a materially different number.
The retail carry-trade thesis walks straight past this. It quotes the SARB policy rate against a G10 funding rate, subtracts the funder, calls the number the "carry", and moves on. What it does not subtract is the round-trip spread cost, applied to the notional, over the holding period. On a pair where the round-trip is 30 to 60 pips at retail — versus 3 to 8 pips on a Pro book with a fixed commission — the carry that survives the friction is not the carry the thesis quoted.
The Real Cost
Let us do the math with the sheet in front of us. This is the block you can reproduce with a calculator.
Assume a one-lot USD/ZAR position. Standard lot size is 100,000 units of the base currency, so 100,000 USD notional. Assume USD/ZAR is quoted at 18.00 for pip-value arithmetic — one pip of movement, defined as a change in the fourth decimal, is 0.0001 of ZAR per USD, so on 100,000 USD the pip value is 100,000 × 0.0001 / 18.00 = 0.556 USD per pip. Call it 0.55 USD per pip for rounding.
Now the retail spread. Retail Standard books at the firms on the sheet quote USD/ZAR at round-trip costs that historically sit in a 30-to-60-pip band depending on session and volatility. Use 40 pips as the working number — that is a conservative Standard-book figure. Round-trip cost per lot: 40 pips × 0.55 USD = 22 USD.
Now the Pro alternative. Take an Exness-style Pro book at 0.1 pip on the majors and extrapolate the exotic mark-up conservatively — call USD/ZAR round-trip 10 pips inclusive of a 3.5-USD-per-side commission. Round-trip cost per lot on the Pro book: 10 × 0.55 + 7 = 5.5 + 7 = 12.5 USD.
The gap per lot per round trip is 22 minus 12.5 = 9.5 USD. That is the delta the commission-model choice buys or costs you before you have a view on the rand at all.
Now the carry side. Assume the annualised carry on a long-ZAR/short-USD position, net of swap costs, is 4.5 percent — a defensible working number given SARB has kept its repo rate materially above US short rates through the cycle the Societe Generale note is written into. On 100,000 USD notional, 4.5 percent annualised is 4,500 USD per year, which is roughly 12.3 USD per calendar day.
Compare the two costs. On the Standard book, a single round-trip eats 22 USD, which is 1.79 days of carry. On the Pro book, the same round-trip eats 12.5 USD, which is 1.02 days of carry. A trader who is turning the position over — closing and re-opening around news, rebalancing hedges, rolling — pays that difference every time. Ten round-trips a year is nine days of carry surrendered to the Standard book that would have stayed in the P&L on the Pro book.
Now stretch it. If the round-trip on your desk is not 40 pips but 60 — which is not unusual on a Standard book during a New York session when a G10 print is due — the number becomes 60 × 0.55 = 33 USD per lot, which is 2.68 days of carry per round trip. Over twenty round trips a year — a mild active-management pace — you have surrendered 53 days of carry to the spread. On 4,500 USD of annualised carry that is 650 USD. On a 100,000-USD notional that is 0.65 percent of the trade, gone. The carry that the retail thesis quoted was 4.5 percent. What you keep, on a Standard book at that turnover, is closer to 3.85. The Societe Generale-style outperformance argument has to survive that haircut before it survives anything else.
The gold leg does not save you either. The commission-model reading of XAU/USD across the same firms shows the same pattern — a wider retail Standard spread than the Pro-account spread — and the spread on the gold leg widens fastest during exactly the sessions the thesis relies on for the correlation. When gold and the rand move together on a risk-on print, the desk widens both books. That is when your round-trip cost is 60 pips, not 30.
If You Only Remember One Thing
A retail carry-trade thesis is a claim about a yield differential net of every friction between you and the balance sheet. Nothing on your broker sheet — the standard-book pip, the pro-book pip, the leverage number, the withdrawal speed — appears in the Societe Generale summary. That absence is not neutral. It is where the yield goes when you are not looking.
If you take the ZAR carry thesis at face value, run it on a Standard-book cost stack, and turn the position over at a modest pace, the arithmetic above tells you what happens: you keep three-and-a-bit percent out of four-and-a-half. If you run it on a Pro book at a fixed disclosed commission, you keep meaningfully more. If you cannot get the pip cost and commission structure on your platform in writing, the trade does not exist as the thesis describes it — it exists as whatever the mark-up desk decides on the day.
We would reverse the framing of this article if the firms on the sheet — Exness, FBS, HF Markets, FXTM, AvaTrade — began publishing round-trip USD/ZAR costs on the Standard book, session-tagged, alongside their EUR/USD marketing numbers. Until any of them does, the retail carry thesis is a spread problem first and a rates problem second, and this desk will keep writing it that way.
FAQ
Does the Societe Generale carry-and-gold thesis on the rand assume institutional or retail execution?
Every published sell-side note on ZAR carry — and the Societe Generale angle is no exception in this respect — is written against interbank execution, not retail. The implicit spread is tight, the commission is negotiated, and the rebalancing cost is absorbed in prime-broker fees. When a retail write-up quotes the same thesis without adjusting for round-trip spread on a Standard book, it is comparing an interbank P&L to a retail cost stack. That comparison does not survive contact with the ticket.
What does "commission model" mean for a broker that markets itself as zero-commission?
Zero-commission means the fee was moved from an itemised line into the spread mark-up. Exness, FBS and XM all built their retail brands on this framing, then reintroduced a Pro or Raw tier that quotes a tight interbank spread plus a disclosed commission. On the sheet, that shows up as the Standard versus Pro split — for example Exness Standard 1.0 pip on EUR/USD versus Pro 0.1. The cost did not vanish. It moved.
Why does USD/ZAR trade wider than EUR/USD on the same broker?
EUR/USD sits in the deepest interbank pool with the tightest underlying spread, so the retail mark-up sits on a very thin base. USD/ZAR is thinner interbank, more volatile intraday and — for a broker with an African client base — often carries a specific desk risk model. The Standard-book spread reflects all three. This is why extrapolating the 1.0-pip EUR/USD figure into a ZAR expectation gives you a wrong number by an order of magnitude.
How much of the carry am I actually surrendering to the spread on a Standard account?
Using the working figures in the piece — a 40-pip round-trip on USD/ZAR at 0.55 USD per pip on one standard lot — a single round-trip costs 22 USD, or roughly 1.8 days of a 4.5-percent annualised carry. Ten round-trips a year is around eighteen days of carry gone. Twenty round-trips at a 60-pip cost is roughly 53 days, or about 0.65 percent of the notional. That is the retail friction the thesis does not price.
Which regulator posture should I care about for a ZAR-specific execution question?
For rand-denominated retail flow, the FSCA in South Africa is the direct supervisor. Three of the firms on the sheet — Exness, FBS and HF Markets — hold FSCA registration alongside their top-tier licences. The FCA and ASIC memberships matter for solvency and dispute resolution, but the FSCA relationship is what determines how a ZAR-side complaint is handled onshore. It does not, on its own, tighten your spread.
Does a higher leverage number change any of this arithmetic?
Not in the way retail marketing suggests. Leverage of 1000:1, 2000:1 or 3000:1 alters the margin you post, not the round-trip spread cost. If you double the notional, you double the round-trip dollar cost. FBS quoting a 3000:1 headline number does not compress USD/ZAR spread; it only reduces the deposit required to control the same lot. The carry-versus-friction arithmetic is unchanged.
Is a Pro or Raw account always the better choice on a carry trade?
For a position held with meaningful turnover — closing around scheduled prints, rebalancing a gold-leg hedge, or rolling — the Pro tier's transparent commission plus tight spread almost always wins. For a pure buy-and-forget position held for months with no turnover, the Standard book's wider spread is a one-time entry-and-exit cost and matters less. The break-even sits at surprisingly low turnover — the piece walks it through above.
What primary source should I read before touching a sell-side ZAR call at retail?
Read your own broker's own written execution documentation for USD/ZAR — the round-trip cost, the swap rate on both sides, the session-specific spread widening policy, and the commission schedule on your account tier. If your broker does not publish these numbers in writing, the trade you are about to put on is not the trade the sell-side note described. That gap is where the carry goes.