Three things carry nearly all the diagnostic weight: whether the return is described as certain, whether anyone can say where it comes from, and whether getting your money out requires paying something.
This piece names no platform, company or individual, and makes no finding about any of them. What follows is structural, and you are the one who has to apply it.
Tell 1 · The return is described as certain
The problem is not the size of the number. It is presenting something inherently uncertain as fixed. Any return is a distribution of outcomes, not a figure.
This is worth labouring, because most people screen on the rate. By that filter a promised 8% looks prudent beside a promised 80%. But if both are described as guaranteed, both have the same structural problem — the conservative one simply unwinds more slowly, which gives more people time to put in more.
Regulated firms face real constraints on how they describe returns, precisely because certainty language is so effective at misleading. The inference runs the other way too: the more freely something uses guarantee language, the less likely it is operating inside those constraints.
Tell 2 · Where the money comes from is vague, or too beautiful
Genuine asset management can tell you three things: what the money is invested in, where the risk sits, and under what conditions you lose. Any of the three missing is worth pressing on.
Two evasion styles. The vague one — "professional team", "proprietary strategy we can't disclose", "diversified allocation" — provides nothing checkable. The polished one is more dangerous: a return curve with almost no drawdown, or an elaborate mechanism described in dense terminology.
The polished version works better because it satisfies the expectation of expertise. The test is the same for both: ask under what circumstances it loses money, and what the worst historical drawdown was. A real manager answers that readily — it demonstrates competence. Deflection ends the conversation.
Tell 3 · Getting out starts costing something
The late-stage pattern is consistent: money goes in smoothly, and coming out attracts charges. Fees, tax, deposits, tier upgrades, processing charges, risk-release payments.
They share one structure — to receive what is yours, you must first send something new. That is a gate, and it has no end point: each payment produces the next reason, because each one proves you will pay.
Obstruction is not always a charge. Delay works too: maintenance, review queues, the next settlement window. Delay drains patience while providing time to encourage further deposits.
And a preceding signal worth remembering: before anything jams, there is usually one or more smooth small withdrawal. That is not evidence of safety — it is the step that makes people commit, dissected in a stranger offers to make you money.
A specific variant worth naming: small withdrawals continue working while large ones do not. This is more convincing than a flat refusal, because the mechanism appears functional and the problem looks like it is with your particular request. If modest amounts leave without difficulty but a meaningful one attracts conditions, that is the same finding, arriving politely.
Underneath all three: only two places a return can come from
Anything paying returns continuously is drawing from one of exactly two places: real income generated externally, or money deposited by later participants. On the surface they are indistinguishable — the balance rises, withdrawals work — and the only difference is how long it can continue.
With a genuine external source — lending, market making, staking rewards — the return moves with conditions, has an identifiable counterparty or mechanism, and falls in bad periods. Variability is the evidence that the source exists.
Without one, the arithmetic requires continuous net inflow. Everything the operation does organises around that: recruitment, encouragement to add, resistance to withdrawal. Its timing is set by deposit flow, not by markets.
One question that separates them
Do not ask whether it is a Ponzi. Ask: "If no new user joined for three months, would existing users still be paid the same?"
A real external source answers immediately, because the answer is yes and the two are unrelated. Something without one will move the conversation — to scale, to the team, to the sector's prospects, or to why you are asking. The deflection is the answer.
What "audited", "insured" and "licensed" actually cover
These three are often lined up as reassurance. They are not meaningless — they simply cover far less than most readers assume.
| Claim | Establishes | Does not establish |
|---|---|---|
| Audited | Some code was reviewed at some point by some firm, with a published report | That the code running now is that code; that the business model works; that funds are safe |
| Insured | Some cover arrangement exists | The limit, what events are covered, or who the beneficiary is — frequently not you |
| Licensed | An entity registered for some activity in some jurisdiction | That the current activity falls inside it, or that you are protected where you live |
The verification method is the same in all three cases: check with the issuer, not with the claimant. An audit report should be findable on the auditing firm's own channels, with the contract address in it matching the one in use. A licence should be searchable in the regulator's public register by number, with name, number and permitted activities all matching.
Certificates, PDFs and screenshots are trivially fabricated, and registration numbers can be borrowed from unrelated entities. Anything you can only obtain from the party being assessed is not evidence.
What real documentation looks like
Everything above is negative space. Here is the positive reference, because recognising what adequate documentation looks like is faster than cataloguing every way it can be missing.
There is a document, and it is boring. Terms, a prospectus, a product sheet — something written for compliance rather than for persuasion. It reads dryly, it repeats itself, and it spends a lot of space on things that could go wrong. That tone is not a bad sign; it is what obligation sounds like.
Risk is described before return. Regulated material almost always leads with what can be lost. Where the risk section is short, vague or positioned after the projections, that ordering itself tells you which audience the document was written for.
Redemption is specified. How to withdraw, how long it takes, what it costs, under what conditions it can be suspended. That last one matters most and is the one most often absent — a product that never contemplates suspending redemptions has not thought about stress, or is not telling you.
Custody is named. Who actually holds the assets, under what arrangement, and whether that is verifiable independently. "Held securely" is not an answer; a named arrangement you can check is.
None of these guarantees anything. A document can be produced and still be untrue. But the absence of one is much more informative than its presence: an operation that cannot supply plain documentation, or supplies only marketing, has told you something about itself before you read a word of it.
Nine questions, copy them directly
Better than memorising characteristics: arrive with questions. These are all ordinary, and a legitimate operation answers them without strain.
Ask these
- How is this return generated, specifically? What is the money invested in?
- Under what conditions do I lose money? What is the largest drawdown to date?
- Who holds the assets in custody, and can I verify that independently?
- Which jurisdiction are you registered in, under which regulator, and what is the licence number?
- If I want to withdraw my entire principal today, what is the process, how long, and what does it cost?
- If I stop adding funds, does my existing return change?
- Can I see the agreement in full and have someone else read it?
- Can I invest the minimum and redeem all of it in a week?
- Can I take a few days to decide?
The last two work differently from the rest. The eighth tests whether a withdrawal path genuinely exists; the ninth tests for time pressure. Both measure behaviour rather than statements, which makes them much harder to fake.
Reading the answers
Pay less attention to the content than to how being questioned is received.
A legitimate operation responds flatly: material is provided, points are clarified, nobody is in a hurry and nobody argues you into anything. Whether you invest is, to them, a routine commercial matter.
Responses worth treating as findings: turning the question into a judgement of you ("you're overthinking it", "that much caution means missing opportunities"); substituting emotion for information ("ten years and never a problem", "everyone's already in"); introducing a deadline; and the most characteristic one — answering at great length, where not one sentence can be checked.
Two things you can verify without them. Public warning lists: many regulators maintain searchable registers of unauthorised or flagged entities, including the MAS investor alert list, the SFC alert list and the FCA warning list. A hit settles it; an absence proves nothing, because lists lag. And the licence check described above, done at the regulator's register rather than from a supplied image.
To run your specific situation through these points one at a time, the suspicious platform decision tree makes them nine tickable questions and returns a risk-signal summary rather than a verdict. It runs in your browser and sends nothing anywhere.
The short version of all six flaws: each one is the same question about where the money comes from, asked in a different way. When that question has no answer that can be checked, everything built on top of it is decoration.