These are the rules we work to. They are written so that you can check whether we followed them, which is the only kind of editorial policy worth publishing.

On this page

What Kind of Site This Is

We research retail forex and CFD brokers as they operate in African markets, and we publish what we find: how a broker is regulated, which company actually contracts with you, what trading with it costs, how money gets in and out, and what the law in your country says about any of it.

Two things follow from that and they govern everything below.

We separate what we verified from what we concluded. That a broker holds a named licence on a named register is a fact, and it carries its source and the date we read it. That a broker is a reasonable choice for a first account is a judgement — ours, argued, and open to disagreement. We write the two differently so you can tell them apart without having to trust us.

We do not write about a broker from its marketing. Every claim we publish traces to something else: a regulator’s own register, the broker’s governing contract, a statute, a payment provider’s published terms, or our own account with the broker.

How We Verify a Broker’s Regulation

We check the register, not the claim. A licence number in a website footer is a claim. We look it up on the issuing authority’s own register — the FSCA’s FAIS and OTC Derivative Provider registers, CMA Kenya’s licensee categories, SEC Nigeria’s registered-operator search and its illegal-operator alerts, Bank of Ghana and SEC Ghana, the Bank of Tanzania and CMSA, CMA Uganda, Egypt’s FRA, and the offshore authorities where the group’s other entities sit.

We enumerate a register rather than searching it. A name search returns a hit and tells you nothing about which category it sits in. Where a regulator publishes separate lists per licence class, we read the lists — because the class is usually the part that matters.

We establish which legal entity will contract with you. A large broker is a group of separate companies, and the one that takes your account depends on where you are and how you signed up. A group-level licence that does not cover your entity is worse than no claim at all, so we name the entity and say what its regulator actually covers.

We check the licence category, not just its existence. Some local licences are agency licences: the regulated local company is not the counterparty to your trades. The licence is real; what it protects is not what most readers assume, and we say which kind it is.

We record what a regulator does not do. Compensation schemes, dispute resolution and the difference between a statutory scheme and a private ombudsman are not decoration — they are the part that matters if a broker fails.

Which Documents We Read

We read the documents in full, not the page that summarises them.

  • The client agreement — the contract that actually governs your account, including the clauses on payments, withdrawals, conversion, dormancy, account closure and the company’s discretions.
  • The fee schedule and pricing pages — spreads, commissions, swaps, conversion charges, and the per-instrument specifications where they exist.
  • The key facts or key information document, where one is published, and whether it agrees with the contract it refers to.
  • Bonus and promotion terms — the turnover conditions, the withdrawal restrictions and the expiry rules, which is where a bonus is either real or not.
  • Deposit and withdrawal pages, per country, including which methods are named and which figures are published.
  • The fine print — the parts set smaller than the rest. A meaningful share of what we publish that nobody else does comes from there.

Where documents contradict each other, we say so and quote both. A marketing page promising no charges and a contract reserving the right to pass charges on are two published positions of the same company, and the contract governs. We show the pair rather than resolving it silently.

What We Test Ourselves

Documents tell you what a broker promises. They do not tell you what happens when you try to use it.

We open an account with the brokers we cover and work through registration and verification as a reader in that country would — recording which identity documents are demanded, what the process asks for, and where it stops.

We go to the deposit screen in each country and record what is actually offered there. This is one of the few places where reading is not enough: the methods a broker lists publicly and the methods that appear once you are inside frequently differ, and the country you registered from changes the answer. What the deposit screen shows is a fact about that market.

We use the platform on the kind of device our readers have — a mid-range Android phone on mobile data, not a desktop on fibre. We record how it behaves there, not how it behaves in a screenshot.

We use the demo account and the platform’s functions: the instruments actually available, the order types offered, how leverage is presented and what it is set to by default, the spreads shown, and the tools a beginner is steered towards.

What we publish about onboarding, deposits, platforms and costs comes from those two places — the account we opened and the documents themselves. Where a figure exists in neither, we say so.

How We Handle User Reviews and Complaints

Almost everyone in this category either repeats a review score or ignores reviews entirely. We do neither: we check whether the complaints are justified.

Here is the arithmetic that makes the difference. A broker processing tens of thousands of withdrawals a day will generate withdrawal complaints as a matter of certainty — not because anything is wrong, but because that is what happens at volume. A complaint count with no denominator therefore tells you nothing at all, and a five-star average tells you nothing either.

So we work the other way round:

We look for the pattern, not the volume. “Slow withdrawals” is the weather; every broker has it. “Accounts closed immediately after a first withdrawal request” is a specific, repeatable behaviour, and that is the kind of thing worth investigating.

We check the complaint against the broker’s own terms. A share of blocked withdrawals are traders who breached a condition they agreed to — a bonus turnover requirement, a third-party payment, a verification step left incomplete. Where the contract explains the outcome, that is the honest answer, and it is more useful than outrage.

We compare the clause against the rest of the industry. Every retail broker’s contract permits prices to move, trades to be voided on manifest error, accounts to be closed and dormant accounts to be charged. A clause is evidence when it differs from what everyone else writes — an unusual discretion, a missing time limit, a condition no comparable broker imposes. Quoting a standard clause as though it were a discovery is building a case, not reporting.

Where we can establish the pattern but not its base rate, we report what we saw and draw no conclusion from it. Reported honestly, that is still useful. Dressed as a verdict, it is the most common failure in this whole category.

User reviews are labelled as what they are — dated, attributed, and never the basis of a rating.

How We Treat Country Law, Exchange Control and Tax

We read the instrument, not a summary of it. A regulator’s site search is not the regulator’s corpus. When we report that a law does not mention contracts for difference, the text of that law was searched.

We check two authorities per country, not one. The securities regulator decides whether a broker can be licensed. The central bank decides whether your money may leave the country — and in several African markets that is the rule that actually binds a retail trader. Checking only the first produces a confident article that misses the point.

We say who a rule binds. Most online-trading rules put the obligation on the broker, not on you. Some, notably in South Africa, also bind the resident. That distinction is the single most useful thing in a legality article and it is usually missing.

We cover tax where the record establishes it, and say so where it does not. If a revenue authority’s own publication does not state under which head a trading profit falls, we report that rather than supplying a plausible rate. A wrong tax figure is worse than no figure.

We read absence differently in different countries. Where a regulator publishes a named list of unauthorised firms, a broker’s absence from it is real information. Where no such list exists, absence means nothing — and saying “not on any warning list” in that market would be technically true and actively misleading.

How We Compare, Rank and Score

We compare brokers on the same basis or not at all. A cost comparison built from figures measured on different bases is arithmetic dressed as precision. Where a comparison cannot be made like-for-like, we say what is being compared rather than publishing a number that looks authoritative.

A rating is an opinion built on verified facts, never a fact itself. Every rating shows what it is made of, and our methodology explains the reasoning behind it.

Our rating model is being rebuilt, and we would rather say so than describe a method we are changing. The reason is worth stating because it shows what we are trying to avoid: our first version rewarded brokers for carrying particular platform software and for offering telephone support, neither of which is what a reader on a phone in Nairobi is choosing between. A scoring rule can be perfectly consistent and still measure the wrong thing. This section will be completed before any rating appears.

One yardstick for every broker. The same questions, the same sources, the same standard, whether or not we earn anything from the company being rated.

What We Do Not Publish

  • Anything we could not verify. If a figure is not in our records with a source and a date, it does not appear — including where a plausible number would read better.
  • Guarantees. Not “ensures”, not “your money is protected”, not “you cannot lose”. No broker can promise that and we will not imply it.
  • Complaint counts without a base rate, or a contract clause presented as unusual without checking whether the industry writes the same clause.
  • Facts obtained by defeating access controls. Where a broker has put its limits and fees behind a login, we report the withholding rather than circumventing it.
  • Anything from an archived snapshot of a page that no longer exists. Archives are useful for finding a live source. They are not a source.
  • Invented credentials. Nobody on this site claims experience, employers or qualifications, because those would be the only unverifiable statements on a site whose whole value is that its claims can be checked.

Dating, Re-checking and Corrections

Every figure carries the date we read it. A correct number from a year ago is a wrong number now, and in these markets rules move.

Records are re-checked on a schedule, more often where the regulatory position is moving. When a source changes, every page standing on it is flagged for review rather than left to age quietly.

When we withdraw a claim, we withdraw it visibly. A page that silently loses a sentence teaches nobody, including us. Substantive corrections are marked as corrections.

If you find something wrong, tell us. A specific correction with a source is the most useful message we receive, and we would rather hear it from a reader than not at all.

Independence and Funding

We earn commission from some of the brokers on this site, and not from others. It costs you nothing and changes nothing about your account. The disclosures page sets out the arrangements in full.

The people and processes that produce our verdicts do not have access to the commercial terms. Our editorial records contain no commission rates and no commercial priorities — that information is kept deliberately separate, so a rating cannot be shaped by a number nobody writing has seen. We cover brokers we cannot earn from, by the same standard, and where a broker that pays us is the wrong choice we say so in its own review.

Who Approves What

Research, writing and approval are separate functions. Nothing publishes until a claim has been checked back to its source — the licence to the register, the clause to the contract, the figure to the page it came from and the date it was read. Work goes back when a claim cannot be traced, when a verdict runs ahead of its evidence, or when a contract term is presented as unusual without the industry comparison behind it.

We are not lawyers and we are not financial advisers. Nothing here is advice about your circumstances, and local professional advice is worth taking before acting on anything published here.