Spotting fake results
Fabricated and curve-fit results share a small set of tells. Learn them and you can discard most untrustworthy services before reading a single testimonial.
Every one of these is a version of the same problem: the result cannot be independently checked. Spot two or three together and the headline number stops mattering.
- Only winning trades are ever shown; the losing stretches simply never appear.
- Entries are vague — “buy around here” — so almost any later price can be scored as a win.
- A large win-rate number sits on the page with no count of calls beside it.
- There is no drawdown figure anywhere, only the upside.
- The whole record is a backtest presented as if it were live results.
- The backtest is suspiciously perfect — a smooth, near-vertical equity curve — the signature of curve fitting.
- The record lives only in a chat that scrolls away and cannot be re-checked later.
- The income flows from broker affiliate links, which pays the service for your sign-up rather than for good calls.
- No named person and no verifiable credential stands behind the calls.
- Nothing is timestamped, so any call could have been written after the move.
Why these tells cluster, and how to weight them
The tells are not random — they group by how a result is faked. Fabrication hides the losers and avoids precise entries, so that nothing can be pinned down. Curve fitting produces the opposite tell: a backtest so flawless it could only have been tuned to the data it is shown on. Unverifiability is the quiet one — a record that lives in a scrollback or a screenshot, where there is simply nothing for a stranger to re-check. The grid below ranks the kinds of evidence by how many honest questions each can survive.
Two tiers of tell
Not every flag is equal. Treat them in two groups. The disqualifying tells defeat verification outright: nothing is timestamped, the record lives in a chat that scrolls away, a backtest is passed off as live, or a win-rate number has no count behind it. Any one of these is enough to walk, because the central claim cannot be checked at all. The cautionary tells — vague entries, a missing drawdown figure, no named author, affiliate revenue — rarely sink a service alone, but two or three together describe a culture of telling you as little as it can. The working rule: one disqualifying tell ends it; a cluster of cautionary tells means go looking for the disqualifying one you have not found yet.
The curve-fit tell, specifically
Curve fitting deserves its own note because it fools careful people. A backtest is a simulation on past data, and it is genuinely useful for designing a strategy — but a strategy can be tuned so tightly to that past data that it looks perfect there and falls apart on anything new. The tells are a backtest equity curve that is implausibly smooth, returns that have no losing months at all, and a refusal to show live, real-money results since. The honest counter is a continuous live record with the losers in it, which is precisely what the verification screen insists on. A clean backtest is a hypothesis; a checkable live record is evidence.
The clean way to act on all of this is the positive screen rather than the negative list: run the five-question verification, and a service either survives it or does not. These tells are simply the fast version — the patterns that tell you a service will fail the timestamp gate before you bother running it.
The takeaway: fabrication hides losers, curve fitting hides nothing but is too perfect to be live, and unverifiability hides the record where no one can reach it. One disqualifying tell is enough to walk away.