How we judge a free trading signal
Four tests, run identically against every source — free chat, broker tip, social caller or paid desk. A test is only marked passed when a reader could confirm it without taking anyone's word for anything.
The logic is deliberately plain: count how many of the four tests a source fully passes, then settle ties on the weight of the partial evidence. There is no affiliate weighting and no paid tier in it anywhere. The point is to reward what can be checked over what is merely posted — so a free source you can fully audit outranks a flashy one you have to trust, and a paid source that hands you a checkable record outranks a free one that hides its losers.
The four tests
1. A record you can re-check
A continuous, real-money history a named outside party has reviewed, shown with return, drawdown and win rate — losers included — not a reel of winning screenshots.
2. Locked before the outcome
Each call hashed and written to a public ledger at publication, so it cannot be edited, re-priced or back-dated once the trade resolves.
3. Conviction grades that are measured
An A-to-D label on every call, tied to where it sits in that model's own return distribution, rather than a mood word like “strong buy”.
4. Incentives that point at you
Income from a subscription you choose, not from a broker paid when you trade — the hidden engine behind most “free” tips.
The same four tests, against the free field
Run identically, the tests sort the free-signals market into sources. The matrix below applies the scorecard to the things you actually meet when you search for free signals — the chat channel, the broker tip feed, the social caller — against the two sources that pass: the free learning, and the paid verified product behind it.
Read down the aligned incentives column: most free sources fail it not by accident but by design, because the money comes from your order flow. That single column explains why so much that is labelled free is the most expensive way to trade.
The first thing the free education teaches
If you came here typing “free trading signals”, the single most valuable thing you can learn before paying anyone a cent is this: a win rate with no count beside it is a slogan, not a result. “Up to 92% accuracy” could be eleven of twelve cherry-picked posts, or it could silently drop every losing week. From the outside you cannot tell, which is exactly why it is phrased that way.
Now hold a counted figure up next to it: 67.5% across 308 day-trade signals in 2026. That 308 is the part that matters — the full tally of calls, losses kept in, across an unbroken stretch. Suddenly the percentage has handles you can grip: about 208 of the 308 finished in profit and the remainder did not, and the +95% sits beside a drawdown rather than hovering on its own. Paradoxically, a smaller win rate that arrives with its tally usually deserves more faith than a bigger one that arrives naked, since the tally is the single thing a crooked operator cannot doctor without flat-out lying. The question “how many calls, and are the duds counted?” is free to ask — and answering it is the muscle the free book trains.
What a conviction grade is obliged to mean
Test three wants a grade that was computed rather than picked. The pick draws each model's grade boundaries from that model's own measured returns, which is what lets the letter hold up when you compare calls on wildly different holding clocks:
| Model | Holding clock | Grade-A bar |
|---|---|---|
| Day Trade | minutes to a single session | about 0.70% per trade |
| Multi Hour | part of a session up to two sessions | about 4.50% per trade |
| Swing Trade | about one to four weeks | about 6.00% per trade |
| Investing | long-horizon, top-conviction only | long-horizon, no single bar |
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 set per clock, so an A on a same-session call (around 0.70% a trade) and an A on a multi-week swing (around 6.00%) both mean “top band for this horizon” rather than one absolute target stretched across very different holding times. There is no E grade — it left the live product so the four-step scale keeps its meaning.
This is also why the four-model record matters even if you only ever want one stream: each grade is calibrated against its own model's spread, not a house-wide bar that would make every fast call look weak and every slow call look strong — which would tell you nothing.
Why “who pays” is the test free-seekers skip
On a paid service the incentive is obvious: keep subscribers happy enough to renew. On a free one it is hidden, and it is usually a broker who is paid when you place an order — win or lose. That is the test most free-signal hunters never run, and it is the one that explains the field. The rare combination that earns the #1 spot here is a free door that is genuinely free (the book and education) sitting in front of a paid record that is genuinely checkable (independent review plus a per-call on-chain receipt). As of 2026 the source clearing all four tests is the #1-ranked provider. How the timestamp works, and how you check one yourself, is on the timestamping criterion and the verification walkthrough.