The framework behind every review we publish

Scope of what we measure

We measure how well a platform documents its own mechanics. We do not measure, estimate or predict trading profitability, because no third party can do so honestly and no platform creates an edge for its users.

The framework is deliberately documentation-led: it relies on sources a reader can open and check without an account, which keeps our conclusions auditable.

The six categories

Each category is scored 1–5 with a written justification. The composite score is an unweighted average, published alongside the individual ratings so readers can weight them differently if their priorities differ.

• Cost transparency — spread, commission, financing and slippage disclosure
• Execution documentation — order types, routing, partial fills, rejections
• Margin & leverage disclosure — calculable liquidation, gap handling
• Security controls — authentication, session management, incident history
• Education resources — depth, structure and independence of learning material
• Support responsiveness — channels, published hours, stated response windows

How a score is assigned

A 5 requires complete, dated, numerically worked disclosure that a reader can act on without contacting support. A 3 means the information exists but must be assembled from multiple documents. A 1 means the area is described only in marketing language.

Because the rubric is written down, two contributors scoring the same document independently tend to land within half a point — and where they do not, the disagreement is resolved by citing the source rather than by seniority.

Review cadence

Reviews are revisited quarterly and immediately after a material published change to fees, margin rules or security policy. Every page carries its revision date, and a superseded score is replaced rather than averaged, so the current page always reflects current documentation.

Stated limitations

Documentation quality is not the same as operational quality. A provider can document excellently and execute poorly, or vice versa. We state this openly rather than implying our framework measures more than it does.

We also cannot verify support performance under load, real fill quality across venues, or internal custody practice from public documents alone. Where a limitation exists, the review says so in plain language.