- Brands checked
- 153
- Official pages captured
- 1,238
- Facts with a source quote
- 1,658
- Account prices recorded
- 736
- Register lookups
- 235
- Registers used
- 16
- Trustpilot profiles read
- 148
- Archive histories read
- 151
What the team does
The Data Research Team runs the brand checks behind Fintrev. It visits each brand’s official site, records what that site says about itself, searches regulator and company registers, and supplies the data used in our reviews, rankings and scores. The current checks are dated November 2026 and cover 153 brands, including prop firms, forex and CFD brokers, banks, card issuers and crypto exchanges.
The other desks write the reviews. This team produces the record they write from.
One brand check, in order
The site visit
We open the brand’s official site in a real browser from a server in Tallinn, Estonia, in the EU. We load it at desktop and phone screen sizes.
What we record
We log what we opened and clicked. We keep the page text and screenshots. We also note the legal entity names and license numbers the site shows about itself.
Prices, rules and fees
These come from the brand’s own pages. A language model pulls out each fact together with an exact quote. Code then accepts the fact only if that quote appears word for word in the captured page text. If it does not, the fact is dropped. Where a price appears only after login or after configuring an account, we say so and record no price.
Sign-up and checkout
Where a sign-up or checkout start is reachable without an account, we click through and record whether an account is required first. We stop there.
Register searches
We search the relevant regulator or company register in a browser and keep a screenshot of each step. So far that means 235 lookups across 16 registers. Each lookup ends in one of four outcomes: the entity is confirmed, our search did not return the entity the brand names, the register shows it as inactive or dissolved, or the register could not be searched at the time.
Trustpilot, archive and domain record
We note the brand’s Trustpilot score and review count, with the date we read them. We look up the first Wayback Machine capture of the domain, plus a snapshot from about a year earlier where one exists. We also take the domain registration date from the public registration record.
What the team does not do
We do not open accounts, deposit money, buy challenges, trade, request withdrawals or contact support. Our pages never claim we did. A check describes what a brand’s public site and the public registers show on the date of the check. Nothing more.
No brand can pay for a check, a score or a ranking position. See our advertising disclosure.
How results reach the pages
Each checked review carries a section titled “How we checked” followed by the brand name, written from this record. The page shows the date of the check, and that date changes when we run the check again. Automated checks compare the text with the record: every number and link in that section must appear in the record, or the text is rejected and rewritten.
Scores are computed by code from the same data. Editors do not adjust them by hand. If an input is missing, it counts as neutral, at half marks, and the review says the figure was not found. The weights are set out in our methodology.
FAQ
Is the Data Research Team a person?
No. It is the byline for the checking process described on this page. We do not publish staff names, and we would rather say so than invent one. Our other bylines are listed on the authors page.
Why check from Estonia?
Our server is in Tallinn, Estonia, which is in the EU. That is where the browser runs, so the brand sites we record are the versions served to a visitor from that location.
What happens when a register is offline?
We record the lookup as one where the register could not be searched at the time, and we do not treat it as confirmed. The FSCA search in South Africa, for example, was offline during part of our checks. If you spot a result that looks wrong, tell us through the contact form and see how we handle corrections.