How we test
Real accounts, real money
Every broker listed is in our test program: tested through a real, ordinarily-opened, funded retail account. We do not use demo accounts, press accounts, or broker-arranged access, and brokers are not notified of test timing. Until a broker's first timed round completes, its row shows 'Testing in progress' and the table displays that broker's own published claim, clearly labeled as a claim rather than a measurement — we never present an unmeasured figure as measured.
Withdrawal timing
One real withdrawal per broker per rail per month, in varied amounts on varied dates to reflect ordinary usage. We record the request timestamp and the arrival timestamp (bank credit or on-chain confirmation). The published figure is full request-to-arrival time. Evidence screenshots are archived with personal details redacted; an evidence hash is published per entry.
Islamic account audit
We open each broker's swap-free variant, hold overnight positions past the published grace period on commonly-held instruments, and log every fee actually charged. Published terms and audited outcomes are shown side by side.
License verification
Regulator entries list the license or reference number as printed by the broker, linked to the official register where it can be checked. We do not list a license we could not trace to a primary source.
Measured rank beats commercial interest
Some brokers pay us commissions. Measured tables are ordered by measured results; where no measurement exists yet, tables are ordered by our standing presentation order and labeled accordingly. We publish worst-in-table results for partners without exception — the site's value is the honesty of its numbers.
Corrections
Errors are corrected in place with a note in the changelog. Datasets are versioned; superseded numbers remain in the archive.