Method
The point of writing the method down is that you can run it yourself. Nothing on this page needs a subscription, a press pass or a contact in the industry. It needs a browser, patience, and a willingness to record a result that says nothing was found.
Step one: collect the launch material
We start with whatever put the name in front of people. That is usually a landing page, often a press release distributed through a wire service, and sometimes a set of near-identical articles on sites that accept paid placements. We save copies of all of it, with the date of capture, because promotional pages are rewritten frequently and a claim that mattered last month may simply be gone.
If the same paragraphs appear under several platform names, we note that too. Shared copy does not prove shared ownership, but it does tell you that the text was not written to describe one specific product.
Step two: separate claims from statements of fact
Then we go through the material sentence by sentence and sort it. A claim is anything the operator asserts about itself. A fact is anything a third party has recorded independently and that we can open and read.
Most launch material is claims almost end to end. This is not by itself sinister, because every company describes itself in its own words. It becomes a problem when the material is written so that claims read as facts: a licence mentioned without a number, a jurisdiction mentioned without a registrar, a partnership mentioned without a counterparty.
Step three: search the public registers
This is the part Piet does the same way every time, in the same order, and records whether each search returned anything.
- FCA Financial Services Register and the FCA Warning List, for firms touching the United Kingdom.
- BaFin Unternehmensdatenbank, for firms touching Germany.
- CySEC, for Cypriot investment firms, which is a common base for this category.
- IOSCO I-SCAN, which aggregates investor alerts published by supervisors worldwide.
We search the platform name, then any company name the material gives, then any trading name we find in the archive. All three searches go into the notes. An empty result is a finding and gets reported as one.
What an empty result means is narrow and worth stating precisely: the name we typed is not listed in the database we searched, on the day we searched it. It does not mean the operator is unlicensed somewhere else, and it certainly does not mean the operator is safe.
Step four: work the archive
Salome takes the domain and the copy. Registration age, whether the promotional page existed before the announcement claimed the platform launched, whether earlier archived versions carried different promises, whether the press release predates or postdates the site. Where a claim has been quietly deleted between versions, we say so and give the dates.
This step is the one that most often produces something concrete, because archives are indifferent to marketing. A page that appeared three weeks ago cannot have a multi-year track record, whatever the copy says.
Step five: write the verdict in words
Four words are available, and the criteria are fixed:
| Verdict | When we use it |
|---|---|
| Avoid | A supervisor has published a warning or an enforcement document naming this platform, and we link to it. |
| High risk | Concrete, documented problems short of a formal warning: a register entry that contradicts the marketing, a company filing that has been struck off, a confirmed rewrite of a material claim. |
| Unverified | The searches came back empty and the material is unsupported. This is the default and it covers most files. |
| No red flags found | The checks ran, the register entries matched what was claimed, and nothing contradicted the material. It is not an endorsement and it is not a safety rating. |
There is no score out of ten, no star rating and no weighted index. Any of those would suggest we measured something we did not measure.
What this method cannot do
It cannot tell you whether your money would come back. It cannot detect a platform that is correctly licensed and still badly run. It cannot see into private agreements, and it cannot confirm identity claims made about people who may not exist.
It is good at one narrow thing: showing you the gap between what was announced and what leaves a public trace. Read the bulletins for that gap, and get the rest of your information somewhere that has it.