Gold rejected the $4,400 area and is pressing back toward $4,300 — and the question we keep getting is which broker, which account, which execution model. That is the wrong first question. The right first question is a flowchart, and we are going to walk it in prose. Three forks. Each fork has a yes branch and a no branch. At the end there is a table that maps the eight combinations to a concrete recommendation drawn from the five operators in our grounding set — AvaTrade, Exness, FBS, FXTM, and HF Markets — plus the raw-spread institutional context (IC Markets Raw, Pepperstone Razor, FXCM Active Trader, Tickmill Pro) that frames the comparison. We assume you already know what a pip on XAU/USD is worth and that spreads on gold are not the same object as spreads on EUR/USD. Let's go.
Question 1: Is Your Average XAU/USD Hold Time Under 90 Minutes?
This fork matters because the economics of a markup spread and the economics of a raw-plus-commission account are not the same curve. They cross somewhere. Where they cross depends almost entirely on how long you hold the position and how often you re-enter the book.
Here is the thing nobody says out loud. Before the ECN model crept into retail gold around 2008-2010, every XAU/USD quote a retail trader saw was a markup. The dealing desk took the interbank bid-ask — in that era five to eight pips wide on a quiet London afternoon — slapped a 30-50 cent cushion on top, and that was your execution. There was no commission line. The spread was the fee, and the fee was invisible in the sense that it was embedded in the price. Post-2001, as electronic matching reached the metals book and ECN liquidity providers (which is the institutional lineage behind IC Markets Raw, Pepperstone Razor, FXCM Active Trader and Tickmill Pro) started quoting raw to retail, gold spreads on the top venues compressed to 10-25 cents on quiet sessions. The commission reappeared, broken out, honest.
If Yes — Hold Times Under 90 Minutes
Scalpers and intraday reversion traders eat spread on every entry and every exit. If you turn the book over six times a day on XAU/USD, you pay the gold spread twelve times a day. The markup account kills you here. Not because the quoted spread looks wide — it is often advertised as "20-30 cents on gold" — but because a markup account is paying for execution infrastructure that an active trader is subsidising for the slow retail book sitting next to them.
The right answer in this branch is a raw-spread execution model. Exness Pro, with its 0.1 pip average on EUR/USD and its genuinely competitive gold book, lives here. HF Markets' zero-spread tier does too, with the caveat that the commission schedule needs to be checked against your actual turnover. FBS runs a zero-spread option as well, but the regulatory posture matters when we hit Question 3.
AvaTrade is explicitly out of this branch. Their own operational posture — scalping prohibited — removes them from the question. That is not a criticism. It is a design choice. Their book is built around options and medium-term directional positioning via AvaOptions, and gold scalpers were never their customer.
If No — Hold Times Over 90 Minutes
Now the math flips. If you hold XAU/USD for four hours, overnight, or several days, you are paying spread twice — once in, once out — spread over a far larger P&L surface. A 25-cent spread on a position you ride for 300 points is a rounding error. A 25-cent spread on a position you ride for 15 points is 1.6% of your P&L envelope before you even start.
Here, the markup account is not punitive. It is often preferable. AvaTrade's 0.9 pip average on EUR/USD telegraphs the shape of their XAU/USD book: wider than Exness Pro, narrower than most banked gold desks pre-2010. For a swing trader, that is fine. You are paying for the AvaOptions platform, the ADGM licensing stack, the 1-3 day withdrawal operational tempo — all things that matter more at a lower trade frequency.
FXTM's 1.5 pip standard account sits in the same conceptual zone. Wider than raw. But paired with the FCA entity, the education stack, the rupee-denominated account support — if you are a positional trader based somewhere that cares about those operational details, the extra spread is a feature, not a bug.
Question 2: Are You Trading More Than 20 Lots of XAU/USD Per Week?
The second fork is volume. Specifically, lots-per-week. Twenty is not a magic number — it is a rough empirical threshold where the commission-plus-raw model's math starts to dominate the markup model's math at retail-realistic position sizes. Below twenty lots per week, the fixed costs of maintaining a raw account (commission minimums on some ECN tiers, the operational overhead of monitoring commission-per-side as a separate P&L line) stop being worth it. Above twenty, every basis point of spread compression compounds.
The historical texture here is worth a minute. The commission-plus-raw model was institutional for most of the twentieth century. Retail did not see it because retail did not have the connectivity. The dial-up era of FX made commission-per-lot operationally impossible to display in a way the client could understand, so markup was the only workable model. Electronic matching, and specifically the FIX protocol penetration of retail OEMs in the 2003-2008 window, is what made the raw book presentable to the retail trader. The reader who already knew this can skip to the recommendation.
If Yes — Over 20 Lots Per Week
High-volume gold traders should be on a raw book. The question becomes which raw book. In our grounding set, Exness Pro and HF Markets' zero-spread tier are the live options. The institutional lineage — IC Markets Raw, Pepperstone Razor, Tickmill Pro, FXCM Active Trader — frames what the standard looks like at the top of the retail ECN segment: sub-0.2 pip majors, commission around $3-$3.50 per side per lot, and gold spreads routinely under 20 cents in London hours.
FBS deserves a mention here with a caveat. Their 1:3000 leverage ceiling is the loudest marketing claim in the retail space, and for a disciplined high-volume trader it is irrelevant — nobody serious runs 1:3000. What matters is whether the zero-spread account's commission schedule is competitive against Exness Pro and HF Markets Zero at your specific turnover. Pull the ticket data. Do the arithmetic. The marketing is not the math.
If No — Under 20 Lots Per Week
Low to moderate volume traders are better served by the markup account's operational simplicity. One number per trade. No commission reconciliation at month-end. No per-lot minimum friction. AvaTrade, FXTM standard, and the Exness Standard tier (as distinct from Exness Pro) all live here. The spread premium versus raw is real but immaterial at this turnover.
The subtle point: at low volume, the quality of the surrounding stack — platform, withdrawal speed, education, support in your language — outweighs 0.3 pips of spread compression. FXTM's rupee account support matters more than ten cents on gold if you are banking in India. AvaTrade's AvaOptions matters more than fifteen cents on gold if your edge is options structuring around the metals book.
Question 3: Do You Need Tier-1 Regulation on the Entity That Holds Your Margin?
This is the fork that resets the whole decision tree. Not because the first two forks were wrong — they are correct within their own domain — but because tier-1 regulation (FCA, ASIC, specifically) materially constrains leverage, overnight financing structures, and the client-money regime that governs what happens if the operator fails.
Every one of our five grounding operators holds at least one tier-1 licence. AvaTrade and FBS carry ASIC. Exness, FXTM, and HF Markets carry FCA. But — and this is where traders consistently misread the regulatory map — holding a licence and routing your account through the licenced entity are not the same thing. The entity that holds your margin determines the regime that governs your money.
If Yes — Tier-1 Entity Required
If you need the FCA or ASIC entity on your account, the leverage ceiling drops. Hard. ESMA-aligned retail limits in the FCA regime cap XAU/USD at 1:20 — a very different animal from the 1:1000 and 1:2000 numbers advertised on the offshore entities. The 1:3000 FBS headline exists because FBS's offshore entity exists. The ASIC entity does not offer that leverage. Neither does any FCA entity offer it. There is no path through the regulatory map that gives you 1:2000 and FCA client-money protection on the same account.
So the practical implication: if the answer here is yes, your realistic grounding-set options narrow to AvaTrade (ASIC), Exness UK (FCA), FXTM UK (FCA), or HF Markets UK (FCA), with the leverage, platform, and spread economics of each respective tier-1 entity — not the headline numbers you see on the global marketing site.
If No — Offshore Entity Acceptable
If the answer is no, the full offshore stack opens up. FBS at 1:3000, Exness at 1:2000, AvaTrade at 1:400, HF Markets at 1:1000. The regulatory posture here is a mix of CySEC (which is a serious regulator but not tier-1 in the client-money protection sense), FSCA, FSA, FSC, and the various Caribbean regimes.
This is where the primary document cross-reference matters. The FCA's handbook on client money (CASS 7, in operative form since the 2017 overhaul) treats client funds as a trust-law obligation on the operator — segregated, auditable, with a specific administration protocol if the operator fails. The CySEC CIF regime, by contrast, operates under a Markets in Financial Instruments Directive framework that is comparable in principle but has produced a very different track record in practice: the 2018 FXTM Cyprus restructuring and the various offshore-entity collapses in the retail CFD space between 2015 and 2022 are the historical texture that justifies the fork. Both regimes claim client-money segregation. Only one has the enforcement archive the FCA has.
If You Answered Everything: The Combo Table
Here is the eight-row map. Q1 is hold time under 90 minutes. Q2 is above 20 lots per week. Q3 is tier-1 entity required.
| Q1 | Q2 | Q3 | Recommendation |
|---|---|---|---|
| Yes | Yes | Yes | FCA Exness UK or HF Markets UK; accept the capped leverage. |
| Yes | Yes | No | Exness Pro offshore or HF Markets Zero — raw spreads, full leverage. |
| Yes | No | Yes | Exness UK Standard; low-volume scalping under FCA client-money regime. |
| Yes | No | No | FBS zero-spread offshore; small-ticket scalping with light operational load. |
| No | Yes | Yes | AvaTrade ASIC or FXTM UK; swing-sized positions, tier-1 margin protection. |
| No | Yes | No | AvaTrade offshore or Exness Standard; wider spreads amortise over longer holds. |
| No | No | Yes | FXTM UK Standard; the full tier-1 education and operational stack. |
| No | No | No | AvaTrade with AvaOptions; swing exposure plus options structuring on gold. |
Two notes on the table. First, the raw-plus-commission institutional lineage — IC Markets Raw, Pepperstone Razor, FXCM Active Trader, Tickmill Pro — is the reference point for what a tier-1 raw book looks like at the top of the retail market. Our grounding set names Exness Pro and HF Markets Zero as the closest commission-plus-raw executions available among the five operators we can cite directly. Second, every row of this table collapses the moment your trading pattern crosses one of the three forks. Re-walk the tree quarterly.
One concession worth making to the markup-model view: the strongest argument the markup camp has is that commission reconciliation is an operational burden retail traders consistently underestimate. That is true. We concede it fully. What the markup camp then concludes — that markup accounts are therefore better for most retail — does not follow. The right conclusion is that commission reconciliation is a solved problem in any serious trading journal, and the P&L delta at any meaningful volume dwarfs the reconciliation effort.
This piece does not cover XAU/USD swap-rate economics on overnight positions — the overnight financing math is a different argument and deserves its own walkthrough. It does not cover the tax treatment of CFD gold versus spot gold under HMRC, ATO, or any specific regime — we are not licensed to advise. And it does not cover the microstructure of the LBMA fix and how it propagates into retail quotes during the London AM and PM fixings — a fascinating topic for another piece.
FAQ
Why is a 25-cent spread on XAU/USD different from a 2.5 pip spread on EUR/USD?
Different objects. A pip on XAU/USD at the $4,300 level is $0.01 per troy ounce times the contract multiplier, so a 25-cent quoted spread equals 25 "cents" of gold, which on a 100-ounce standard contract is $25 of cost per round turn. On EUR/USD, a 2.5 pip spread on a standard lot is $25 flat. The dollar cost happens to be similar. The percentage-of-move cost is not — gold's daily range at $4,300 is far wider in dollar terms than EUR/USD's daily range in dollar terms.
How much did XAU/USD retail spreads actually compress between 2001 and 2026?
In 2001, retail gold spreads of 50-80 cents were common because retail saw a markup on an already-wide interbank quote. By 2008-2010, as ECN liquidity reached retail gold books, top-tier retail spreads compressed to 25-40 cents. Post-2015, with commission-plus-raw execution reaching the metals desk at firms like Pepperstone, IC Markets, and Tickmill, raw-tier spreads during London hours settled at 10-25 cents — before commission.
Does FBS's 1:3000 leverage actually matter for XAU/USD trading?
For a disciplined trader, no. Leverage above roughly 1:100 is a theoretical ceiling that bears no relationship to responsible position sizing on gold. The 1:3000 headline exists because FBS operates an offshore entity where no regulator caps the number. The practical ceiling you will ever use responsibly is set by your account size and your stop distance — not by the broker.
Is Exness Pro's 0.1 pip EUR/USD spread representative of their XAU/USD book?
Not directly, but it is a signal. The 0.1 pip pro-tier average on EUR/USD indicates that Exness is routing its pro accounts to raw ECN liquidity and taking commission rather than markup. That execution architecture applies to gold too, which is why Exness Pro is in the Yes/Yes recommendation branches above. The specific XAU/USD spread varies by session and liquidity provider, but the model is the same.
What is the practical difference between CySEC and FCA client-money protection?
Both regimes mandate segregated client money. The FCA regime, under CASS 7, operates as a trust-law obligation with specific administration procedures codified since the 2017 CASS overhaul. CySEC operates under a MiFID-aligned framework that is comparable in design but has produced a less tested enforcement archive. For a trader whose risk model includes operator failure as a non-trivial tail event, the FCA regime is the defensible choice. For traders whose risk model does not, CySEC is adequate.
Why does AvaTrade prohibit scalping on gold?
Operational choice, not regulatory. AvaTrade's execution and risk model is built around medium-term directional positioning, options structuring via AvaOptions, and a client base that trades at lower frequencies. Permitting scalping would change their hedging economics in a way their market-making model is not optimised for. It is a legitimate design decision — it just removes AvaTrade from the under-90-minute hold-time branch of the decision tree.
Does the commission-plus-raw model always beat markup on XAU/USD?
No. The crossover depends on turnover. At under 20 lots per week with hold times over 90 minutes, markup accounts frequently win on all-in cost because you avoid the commission floor and the per-side reconciliation overhead. At over 20 lots per week with scalping-style hold times, raw-plus-commission wins decisively. The decision tree above walks the forks explicitly rather than defaulting to "raw is always better" — because it isn't.
Can I run both a markup account and a raw account at the same operator?
With some operators in our set, yes. Exness offers Standard and Pro tiers under one client ID. HF Markets offers multiple account types including the zero-spread tier. FXTM has standard and ECN variants. Running both lets you route different strategies — short-dated scalping to the raw book, swing positions to the markup book — without maintaining relationships with two separate operators. It is operationally cleaner and legitimately common among active gold traders.