How we calculate Accuracy
Accuracy is the one-take metric: the percentage of replayed signals that reached a +2% price move from entry before hitting their stop, based on real 1-minute exchange candles (highs/lows, not just closing price). The stop is the channel’s own published stop; when a channel publishes no usable stop we score the signal with a wide 10% fallback stop and disclose how often that happened. Signals whose entry was never reached, or where we have no candle data, are excluded from the percentage. The full rules, including every edge case, are on our Methodology page.
Every channel publishes a different ladder of targets, which makes "hit its first target" numbers incomparable between channels — a channel with a target 0.5% away trivially "wins" more often than one aiming at 5%. One fixed +2% take, full position, makes every channel answer the same question.
What "Verified" means
A channel is Verified when ChainRated has replayed its published signals against historical 1-minute exchange candles and computed the resulting track record — what following each signal exactly as posted would actually have returned. Verified does not mean "good"; it means the numbers shown are measured by us from real price data, not self-reported. A channel can be verified and losing, and its page will say so. Verification cannot be bought: everything we sell to a channel — a listing tier, the section of its own page it writes, and an advertisement marked Promo — buys visibility, and none of it changes a figure we computed.
Edits and deletions after publication
We record every signal message as first seen ("snapshot"), and every metric is computed from that snapshot — so editing or deleting a message afterwards changes nothing in a channel’s statistics. Deleted signals stay in them; they don’t disappear.
Separately, we re-read recent signal messages at the source — signals from the last 90 days — and compare each one against the snapshot we stored when we first read it. Two things get counted: the text no longer matching that snapshot, and the message being gone from the channel. Where we find something, the channel’s page shows the count and the date we last looked; where we find nothing — which is the ordinary case — it says nothing, because a clean note on almost every page is noise rather than information. The counts cover the same 90-day window we re-read rather than the channel’s whole history: a mark we can no longer re-check is a mark we could never withdraw.
Telegram stamps a message as edited for changes that have nothing to do with what it says — buttons, attached media, a link preview. On the first channel we checked, 38 of 100 messages carried that mark while their text was identical to our snapshot. Counting those as edits would have been an accusation dressed as a neutral number, so the mark only tells us where to look, and the comparison decides.
A changed message is a fact; why it changed is not ours to state. Channels fix typos, correct a ticker or add a note. We publish that the text differs from what we recorded, and the number of signals it applies to, and leave the conclusion to the reader. We do not keep a copy of the new version or a diff — only the original snapshot, which we already had.
The four kinds of feed we list
Not every channel in the catalogue is trying to do the same thing, and one label for all of them would hide the difference that matters most. We sort every listed channel into exactly one of four kinds, and the tabs on the Signal Providers page follow that sorting.
- Signal channels — publish entries we can replay. These are the ones that can become Signal Verified.
- Analytics channels — publish analysis, commentary or results, and no entries. Nothing on their feed could ever be replayed, so they carry no metrics. They are listed, not hidden, and the tab says "Unverified analytics channels" because that is exactly what they are: unverified by nature, not by neglect.
- VIP storefronts — the free feed is a shop window for a paid group. Same position as analytics channels: reviews and reports, no metrics, until the owner opens the paid feed to us.
- Promo funnels — publish neither entries nor analysis; the posts route readers into a third-party platform or bot the channel earns a commission from. Their own shelf, because the difference from an analyst is not cosmetic: an analyst is paid when readers stay, a funnel is paid for the traffic either way.
Separately from all four, some channels post entries in a format our parser does not support yet. Those are listed too, with no metrics and no activity claim, under "Feed formats not supported yet". That heading is about our coverage, not about them — they may be posting daily.
How a channel can prove more
A channel that starts out unverified is not stuck there. Two voluntary steps make its page carry more, and each one is a step the channel takes, never a verdict we hand down.
- Private feed open — the channel gives our reader access to its paid feed, and we replay those signals exactly as we replay public ones. What this settles is the words: there are no signals behind the paywall that we do not see.
- Proof of trades — the owner connects a read-only exchange API key, and we can see the account actually trading rather than only the posts about it. What this settles is the money: a channel can publish one thing and trade another, and a key closes that gap.
Both badges mean transparency, not performance. A channel with either one can be losing money, and its page will say so — that is the point of measuring instead of recommending. Neither badge is for sale: everything we sell — placement, the owner-written section of a page, the Promo banner — buys visibility and nothing else, and no package includes, implies or discounts a badge.
Both badges are held by a live fact and drop off on their own when the fact stops being true: access withdrawn, or a key disabled or expired, removes the badge without anyone asking us to. A badge that outlives the access it describes would be the page lying on the channel’s behalf.
Duplicate / reposted signals
When a channel reposts a signal from another source (via native Telegram forward, or by copying the same price levels within a short window), we attempt to identify the original source and exclude the repost from that channel’s own Accuracy credit.
Methodology changes
We may refine our calculation methodology over time as we improve our parsing and verification systems. When we do, historical figures may be recalculated to stay consistent — we do not manually adjust individual providers’ numbers outside of this process.