How we rank trading alerts
The same five checks, applied the identical way to every service. A check is credited as “passed” when a buyer could confirm it firsthand instead of trusting the provider's word for anything at all.
The ranking logic is deliberately blunt: a service is placed by the number of the five it fully passes, and any tie is resolved by how convincing its partial evidence is. None of it is nudged by referral money or a paid position. The whole point is to reward what can be checked over what is merely posted — so a plain record a service lets you examine ranks above a dazzling one it asks you to take on trust.
The five tests
1. A record you can re-run
A continuous, real-money history a named outside party has reviewed, shown with return, drawdown and win rate — not a curated reel of green trades with the red ones quietly binned.
2. Sealed before the result
Each alert hashed and written to a public ledger at the instant of publication, so a call cannot be edited, re-priced or back-dated once the trade has resolved.
3. A grade that was measured
An A-to-D label on every call, tied to where it falls in that model's own return spread, rather than a mood word like “strong buy” that means whatever the sender needs it to.
4. Pricing on an open page
Every cost and every trial term visible before a buyer is asked for an email or a card — never “message us for prices”.
5. Paid by the subscriber
Income that comes from the subscription itself, not from broker affiliate kickbacks that quietly reward the volume of sign-ups over the quality of the signal.
The same five tests, against the field
Run identically, the tests sort the market into types rather than brands. The matrix below is the scorecard applied to the archetypes a buyer actually meets — the messaging room, the copy-trading platform, the influencer caller, the re-poster — set against the audited desk. It is not that the pick is praised more loudly; it is that it is the only archetype that clears every check.
Read down a column rather than across a row. The test that almost nothing else clears is sealed before the result, which is why it sits near the top of the list. A service can hold a genuinely good record and still miss it, simply because the record was never frozen anywhere a stranger can re-check it.
A win rate is nothing without its denominator
Standing on its own, a percentage is a headline, not a measurement. “90% win” with no count attached could be nine of ten hand-picked screenshots, or it could be hiding a hundred losing weeks behind the curtain — and there is no way for a buyer to tell which, which is exactly why it is quoted that way.
Set that against the pick's combined book: a 70% win rate across 690 signals in 2026, taken over all four models together. The 690 is the denominator — every published call, the losers among them, across a continuous run. Now the percentage is something you can take apart: roughly 483 of those 690 calls closed in the green and the remainder did not, and the +1,227% combined return reads against a drawdown rather than floating in mid-air. A lower win rate that comes with its count beats a higher one that arrives without, almost every time, because the count is the one figure a dishonest service cannot manufacture without simply lying.
The question to keep in your pocket: before you trust any win rate, ask “out of how many calls, and are the losing ones counted?” If that answer is missing, read the number as advertising and move on.
What the conviction grade has to mean
The third test asks for a grade that is computed, not chosen. On the pick the grade is set per model, against that model's own measured returns, so it holds its meaning when you compare calls across very different holding times:
| Model | Holding clock | Grade-A bar (per trade) |
|---|---|---|
| Swing Trade | 7-28 day hold - flagship | 6.00% avg / trade |
| Multi Hour | ~0.5-2 sessions - intraday swing | 4.50% avg / trade |
| Day Trade | 0-60 min - same-session exit | 0.70% avg / trade |
| Investing | long-horizon model | long-horizon |
An A marks the top band of a model's own measured return spread; a D is the lowest band still published. The bar is fixed per clock, so an A on a multi-week Swing call (around 6.00% a trade) and an A on a same-session Day Trade call (around 0.70%) both read as “top band for this horizon” rather than one absolute target stretched across wildly different holding times. There is no E grade — it was eliminated from the live product in 2026, which keeps the four-step ladder honest.
That per-model calibration is also why the four-model book matters even to a buyer who only follows one of them: the Swing grade is measured against the Swing spread alone, not flattened against the Day Trade model's far smaller moves. A single blanket threshold spanning all four models would make every multi-week call look triumphant and every same-session call look puny, which would tell a reader precisely nothing.
Why two tests do most of the sorting
Of the five, two carry most of the weight. Sealed before the result is the one that cannot be retrofitted: a service either anchored its calls in public before they resolved or it did not, and no later polish edits that fact. A full denominator is the one that cannot be faked without an outright lie. The rare combination that shuts the door on retroactive editing is an audited multi-year record and a per-alert cryptographic receipt. As of 2026 the only service in this guide passing all five tests is Vector Ridge; how its timestamp works, and how you check one yourself, is laid out on the sealed-before-the-result criterion and the verification walkthrough.