A ChainRated channel page answers eight questions in an order that works as a reading route, and this guide puts the route first, then explains what each block measures and how the blocks change each other's meaning.
How do you read a channel page in five minutes?
Top to bottom, one block and one question at a time. A channel's rating page on ChainRated is built from the channel's own posts: every call is stored as it was first read, replayed against price history where its levels allow it, and anchored on a public blockchain. The page then reports what that record shows, in eight blocks.
| Block | The question it answers | How to read it |
|---|---|---|
| Name and badges | What kind of channel is this, and what is already known about it? | Signal Verified means the record was measured. Look for a feed, repost or Mostly promo label, and for a warning about a look-alike name. |
| Stop-loss line | How many calls came with a stop-loss? | The lower the share, the more the accuracy further down rests on a 10% fallback stop. |
| "Edited after publication" (only when something was found) | Did the channel change or remove calls after they were recorded? | Compare the count with the number re-read and check the date. A missing line means nothing was found, or nothing was re-read yet. |
| Composite score | How does the channel score out of 100 on Return, Risk and Honesty? | Read the three tiles before the total. A Return of 0 means the as-published replay ended below zero. A "Not on the leaderboard yet" note means the sample is still small. |
| One-take accuracy | How often did calls move 2% their way before the stop? | Read the count of verdicts first, then how many were scored on the fallback stop. |
| As-published record | What would following the calls exactly as posted have returned? | Check whether the total ends above or below zero. |
| Latest calls | What do the ten most recent replayed calls look like, and was each stored copy anchored? | Read them for behaviour. Ten rows are too few to judge a hit rate. |
| Signal stream overlap (only where a repost link exists) | Are these calls the channel's own? | If the page names a source channel, judge the record there. |
Some blocks are best read in pairs:
- High one-take accuracy beside a Return of 0. The entries tend to move the right way first, and the calls followed as posted still lose: one ranked channel read 67% on one-take and -316.2% as posted.
- A high score with a "Not on the leaderboard yet" note. An early reading on a small sample, worth checking again after 30 answered signals and 60 days.
- Accuracy on a repost. The calls behind it first appeared on another channel, a few seconds earlier, so the source's page is the one to read.
Not every page has every block. A channel that posts nothing that can be replayed has no score and no track record. The sections below take the blocks in page order, and the catalogue figures in them are as of 7 October 2026.
What do the badges under the name tell you?
What has been established about the channel before any number is read. The first badge is either Signal Verified or Unverified. Signal Verified means the calls were replayed against real price data and the numbers below were computed from them. It says the record was measured, and says nothing about whether the record is good; the line under the track record puts it as "Verified means measured, not recommended". What the badge settles is set out in what Signal Verified means.
Other badges in the same row are conditional:
- Aggregated feed or Automated feed marks a channel whose calls come from a bot or a collection of other sources rather than one trader. Feeds are scored the same way and ranked apart, as explained in automated and aggregated signal channels.
- Repost or Reposted by N marks a link to another channel's signal stream; the overlap block lower on the page explains it.
- Mostly promo marks a channel where trade signals are under 5% of at least 100 posts in the last 30 days.
- A warning under the name appears when the name is easy to confuse with another listed channel's, as described in the channel name that reads like another.
Private feed open and Proof of trades belong to the trust ladder. The first means the channel opened its paid feed, so its closed calls are replayed like public ones; the second means the owner connected a read-only exchange key, so a real trading account is visible behind the posts. Both badges are about transparency, cannot be bought, and come off by themselves when the access behind them stops.
What does the stop-loss line say?
How much of the channel's published risk you can actually see. The line reads "Stop-loss published: with a stop X% (a of b parsed signals) · without a stop Y% (c)". The count covers every call that could be parsed, and a stop counts whether it was given as a price or as a percentage.
A call without a stop leaves the exit to you, and it also leaves the record without a defined loss. Across the catalogue, 76.1% of the 36,681 parsed signals carried a stop. Outside it the share can be much lower: a DayTrading.com study of seven stock-signal groups found that "74% of signals had no stop loss at entry", on a sample of 112 signals.
When a call has no usable stop, the one-take replay uses a 10% fallback stop, and the page says how many calls were scored that way. A wide stop makes +2% easier to reach first, so a high accuracy built on fallback stops partly reflects the distance to the stop. The full argument is in why half of all crypto signals come without a stop-loss.
What does the "Edited after publication" line mean, and what does its absence mean?
When the line is there, it counts recorded calls that the channel later changed or removed. When it is missing, either no mismatch was found among the calls re-read or nothing was re-read yet, and the page does not show which. Telegram lets channels rewrite history. Its channels guide says "You can easily edit your messages in channels, regardless of how long ago they were posted." Calls on ChainRated are re-read at the source while they are less than 90 days old and compared with the stored text. When the check finds mismatches, the line reads, in part: "X of N signals we re-read at the source no longer say what they said when we recorded them, and Y are no longer in the channel." It ends with the date of the last check and a link to how the check works.
Every number on the page uses each call as first published, so an edit changes only the count. The count leaves the reason open, because channels also fix typos and delete duplicates.
Changes are rare. Of the 18,771 signals published in the 90 days up to 7 October 2026 and re-read at the source, 29 no longer matched their stored text and 4 were gone from their channel. The page line counts the same 90-day window, so the figure moves every day. What a cleaned feed looks like, and what the line can and cannot catch, is in what a channel's feed looks like after the losing calls disappear.
How is the score out of 100 put together?
From three axes with fixed weights: Return 40%, Risk 35%, Honesty 25%. The block is headed "Composite score", shows the total "out of 100", and gives each axis its own tile; the methodology has the full rules.
- Return, "Per unit of drawdown". The as-published replay result divided by the deepest drop along the way. A ratio of 3 or more scores 100. A replay that ends below zero scores 0 on this axis, so losing channels are not ranked against each other by how much they lost.
- Risk, "Stops, leverage, worst fall". The largest part, 60% of the axis, is the share of scored calls that carried the channel's own stop. The rest is the median published leverage (up to 3x costs nothing, 25x scores 0) and the worst drawdown of the replay.
- Honesty, "Edits, unfilled, reposts". Edits and deletions make up 40% of the axis, entries that were never reached 25%, reposting another channel's stream 20%, and whether each call names symbol, direction, entry, stop and target 15%.
Working from the published weights, with every part measured: a channel that never publishes a stop gives up about 21 points of the 100 (35% × 60%). Heavy edits cost up to about 10 points (25% × 40%), and being a repost about 5 (25% × 20%). These are calculations from the weights; the methodology does not state them as figures.
The Risk axis can also be marked down for how the trades behaved: how deep a winning trade went into the red before it came good, and how long a losing trade waited for its stop. A line under the tiles says when this applies. Each of the two can take at most a quarter off the axis, needs at least ten trades of its kind and can only lower a score. The thresholds are under how this is counted.
A part that could not be measured drops out, and its weight moves to the parts that were. It does not count as zero. A channel with no replay at all has no score.
Honesty carries the smallest weight on purpose: it measures hygiene, and a clean channel can still be useless to follow. Of the 72 channels with a score, the median score was 48.2, and 42 of them scored 0 on Return.
Why does a channel with a score have no place on the leaderboard?
Because a score appears before the sample is large enough to rank it. A channel enters the leaderboard at 30 answered signals and 60 days between its first and last. Until then the page shows its score with a note: "Not on the leaderboard yet: ranking starts at 30 answered signals and 60 days of history. The score above is real, the sample behind it is small."
One channel scored 94.2 on 25 answered signals and was not ranked. Of the 72 scored channels, 36 had cleared both thresholds: 28 were ranked on the leaderboard, and 8 automated or aggregated feeds were listed on their own tab. A high score on a small sample is a reason to come back later, since a score from 200 calls carries more weight than the same score from 25.
What does one-take accuracy measure?
How often a channel's calls moved 2% in their direction before hitting the stop, under one rule applied to every channel. The block sits under "Track record we computed" and shows a large percentage with a line such as "38 of 54 signals reached +2% before the stop", followed by how many were stopped out and how many reached neither within 30 days.
Under the rule, every call is entered at the posted price with the whole position and no leverage, has a single take-profit at +2% from entry, and keeps the channel's own stop. Calls are replayed on 1-minute Binance futures candles with exchange fees included. Each call gets one of three verdicts:
- Reached +2% before the stop. Counts for the channel.
- Hit the stop first. Counts against.
- Reached neither within 30 days. Also counts against: a call that goes nowhere for a month has given its answer.
A call whose entry was never reached, or whose coin has no candles on that exchange, is left out of the percentage. In the catalogue listings, a hit rate on fewer than ten verdicts shows as a dash. The channel's own page prints the percentage anyway, with the count beside it, so a page can read 100% on a single call. The case for one fixed target across all channels is argued in how accurate are crypto signals.
Below the percentage, "What the signals asked of you" gives three medians: the wait to +2%, the dip before the target and the wait to the stop. They describe what holding the calls felt like. Two of them can mark the Risk axis down, as described above, and each appears only when at least ten calls of that kind exist.
Can a channel be right two times in three and still lose as posted?
Yes, and the page shows both numbers so the difference is visible. Below the one-take block, four tiles (Total per trade size, Win rate, Closed trades, Average trade) report the as-published replay: every call followed with its own targets and stops, trade by trade. One-take accuracy asks how often a channel is right. The as-published record asks what following it would have returned.
One ranked channel showed 67% one-take accuracy, 536 of 806 calls. Its as-published record over 749 closed trades read -316.2% in total, and its Return tile read 0.0. Its entries usually moved the right way first, and what happened afterwards, with the channel's own targets and stops, lost more than the wins made. The other ways a high percentage misleads when read alone, from target distance to sample size, are in why 90% accuracy can still lose you money.
What does the Anchored badge in the latest calls prove?
That the stored copy of a call existed, in exactly that form, by a certain date. The table lists the ten most recent replayed calls with the symbol, LONG or SHORT, the posting date, the outcome and the result. A badge next to a call shows its anchor state, and a call without one has no anchor this page can confirm.
- Anchored. The call's snapshot is part of a daily batch whose root was written to a contract on Arbitrum One. Hovering over the badge shows how many days after publication the call was anchored and the batch number. The batch's own page gives the date and the transaction. In a very large batch, the single-call verification page skips the check and leaves it to the batch page.
- Anchor pending. The snapshot is recorded and goes into the next daily batch.
Across the record, 99.7% of signals were in a confirmed batch. The hover text gives the limit in the site's own words: "Proves these bytes existed by then, not what the channel posted." How the anchoring works and how to check one call yourself is in how we anchor every signal on Arbitrum. A long gap between the posting date and the anchor date has its own meaning, covered in why a channel's old history proves less.
Read the table for behaviour: whether calls carry stops, how outcomes are spread, whether recent calls look like the older ones. With 7 hits out of 10, the plausible range for the true rate still runs from about 40% to 89% (95% Wilson interval). The arithmetic is in how many trades a hit rate needs before it means anything.
What does the overlap block mean on a repost?
That the calls on this page first appeared somewhere else. The block, "Signal stream overlap", appears only on channels with a public repost link. It opens with the rule: "Identical trades published seconds apart, over and over, mean one feed — not two independent track records." On the repost's page it then names the source channel, the share of calls that appeared there first, the number of shared trades and the typical lag. On the source's page it lists the channels that repost it.
The repost's page still shows its own one-take accuracy, computed from its own posts. One repost read 65% while all 309 of its signals had appeared on the source about 12 seconds earlier. Reposting another stream accounts for a fifth of the Honesty axis, and the page tells you where to look instead: "Judge its record on the source channel."
One limit is stated on the page itself: "first" means first among the channels tracked, and the whole set may be relaying a closed group the site cannot see. Across the catalogue, four channels carried a Reposts line. Why copies of channels exist and how to spot them without the site is in clone signal channels.
What this page does not prove
Every number on the page comes from a replay of published calls against price history, and a replay is not a trade. US commodity rules require simulated results to carry a warning that applies here as a principle, even though those rules were written for registered advisers:
"Unlike the results shown in an actual performance record, these results do not represent actual trading." - 17 CFR § 4.41(b)(1)(i), Cornell Legal Information Institute
Your fill can be later and worse than the posted entry. Your size and leverage are your own. A channel can change its style tomorrow, and nothing on the page forecasts what it will post next. The page measures a record and recommends no channel. A channel that is not in the catalogue can still be checked by hand, as described in how to check a signal channel's track record in 10 minutes.