The small "edited" mark under a call is read by most followers as proof the channel moved the entry after the fact. Across 100 re-read signal messages, 38 carried the mark and not one of those 38 said anything different from the version stored when it was first seen. The mark is a prompt to compare, and the comparison needs a copy you kept yourself.

You take a call, the trade goes against you, and a day later you scroll back to the message. Under the timestamp there is now a small grey word: edited. The entry in front of you is the entry you traded, as far as you remember. But the mark is there, the loss is real, and the obvious conclusion arrives on its own — they changed it.

That conclusion is usually wrong, and the reason it is wrong matters more than the accusation, because the same weakness that makes the mark useless as proof is what makes the real thing so easy to miss.

What does the marker look like across a hundred calls?

Published signal messages in our index get re-read later and compared, word for word, against the text stored when the message was first seen. As of August 2026, in a run of 100 messages: 38 carried Telegram's "edited" marker, and 0 of those 38 had text that differed from the stored version.

Not one of those thirty-eight had a moved entry, a shifted stop or a quietly dropped target. Thirty-eight messages announced that something had been changed. Nothing that mattered to a follower had been.

That is the number worth carrying around, and it is worth stating the limit on it just as plainly. It does not mean channels never change published calls. They do, and the wider count says how often: across 3,448 published signals in 15 channels, re-read as of August 2026, three messages no longer said what the stored copy said. One of the three had moved an entry price six days after it was published. The other two had gained a hashtag.

So both halves are true at once. The marker fires constantly and points at nothing. Real changes exist, are rare, and mostly are not the kind that costs anyone money — and the one that was is invisible to anybody reading the marker, because a hashtag and a moved entry wear exactly the same badge.

Why does a marker fire on a message nobody rewrote?

Because the marker is not a record of what changed. It is a flag attached to a timestamp.

In Telegram's own API description, the field behind it is edit_date, documented as "Last edit date of this message". That is the whole of it. No previous version is stored and no comparison is offered, so nothing separates a rewritten entry price from the ordinary housekeeping a channel does to its own posts — a fixed typo, a swapped image, a button added or taken away, a bit of formatting tidied up. To that field they are one and the same event, and every client renders them all with the same six-letter word.

Even the timestamp is softer than it looks. A user report on Telegram's own bug tracker, filed in November 2024, describes the edit timestamp on an already-edited message moving to the moment somebody added a reaction — an event that involves no author and no text at all.

So the mark tells you an event occurred in a category that includes both "corrected a spelling" and "moved the entry a long way". Reading it as the second one is a guess wearing the costume of evidence.

There is a limit on our side of this too, and it points the same way. A comparison can only start from the first moment the wording is recorded. An edit made in the seconds after publication, before anybody wrote the text down, sits inside the baseline rather than showing up against it. That is not a quirk of one archive; it is the shape of the problem for anyone, including you.

Can a call be changed without the tag appearing?

Yes, and this is the half most followers never hear.

The same message structure carries a second flag, edit_hide, documented as "Whether the message should be shown as not modified to the user, even if an edit date is present". In plain terms: a message can be edited and still be displayed as though it never was.

Put the two facts next to each other and the marker collapses as an instrument. Present, it usually means nothing. Absent, it does not prove nothing happened. A signal you are looking at right now, with no mark under it at all, is not thereby the message that was published — it is the message as it currently stands.

That is a different claim from "channels are lying to you". It is a claim about what the interface can and cannot certify, and the answer is: very little. Which is precisely why a channel's own feed cannot be the record you lean on, the same structural problem behind a monthly results post that adds up only the green rows.

What does an accusation built on the tag cost you?

More than people expect, and the cost lands on the wrong person.

Post "look, it says edited, they moved the entry" in a public group and you are, on the evidence of that run, most likely accusing a channel of rewriting a call whose words are exactly what they always were. The operator can say so, truthfully, and the room will believe them, because they are right. You spend your credibility, and the next time you point at something — including the time you are actually correct — you are the person who cried edit.

There is a quieter cost as well. Someone who has decided the marker is the tell stops looking for the thing that would have shown up in a comparison, because they believe they already have a detector. A channel that genuinely does rewrite calls does not need to defeat anything sophisticated; it only needs its subscribers watching the wrong indicator.

The distinction is the same one that separates the three unrelated things a badge can mean, only one of which is about results: a sign that something was checked is not the check.

What actually shows that a call was changed?

Two texts, side by side. The message as it reads now, and the message as it read when it was published. Nothing else does the job, and with both, nothing else is needed.

That comparison is what makes the real cases visible, and they have recognisable shapes:

  • An entry that has moved toward where the market actually went, turning a trade that never filled, or filled badly, into one that reads as clean. The arithmetic of what a single changed number does to a published record is worked through in the entry price that changed after you bought.
  • A stop relocated to the other side of the price that stopped you out, so the loss in the record was never a loss.
  • A target removed after price failed to reach it, leaving only the targets the move happened to touch.
  • The whole message gone, which is the same manoeuvre with a bigger eraser and its own tell, covered separately in why a channel deletes a losing call.

Each of those is a difference in the words. Each is invisible in the current text, in statistics computed from the current text, and in the marker — and each is obvious the second you hold the earlier version next to the later one.

What does that mean if you have no archive?

It means the copy you take at the moment you act is the only version of that call you will ever be able to check, and once the moment passes it cannot be recreated. Not by scrolling back, not by asking, not by screenshotting later — a screenshot taken after an edit shows the edited text, and it looks exactly like a screenshot taken before one.

So the habit is narrow and cheap.

Capture the call when you commit money to it, not when something goes wrong. By the time it goes wrong, the useful version may no longer exist.

Capture the whole message, with the timestamp in frame. A cropped photo of two numbers proves nothing about when they were on screen or which message they came from. The channel name, the message time and the full body are what turn a picture into a record.

Write the same values into your own log as you enter them — entry, stop, targets, size, and the time you acted. Text you typed yourself is searchable, comparable, and does not depend on finding one image among thousands six months later.

Keep it where you keep receipts, not in your camera roll. If a payment ever becomes a dispute, the material that gets asked for is dated and boring: the US fraud complaint centre's guidance is to keep all original documents in a secure location, and it lists "printed or preferably electronic copies of web pages" and cryptocurrency receipts among the record types to retain. "The message says edited" is on no such list, and it should not be — it is not a fact about what the message contained.

The asymmetry is worth sitting with. The operator can revise the published record at will. You cannot revise theirs, and you cannot recover what it used to say. The one document neither of you can rewrite is the one you made at the time.

What this does not prove

The run described here is a hundred messages from our own index, re-read as of August 2026 against the text stored when each was first seen. Three limits belong with it.

It is a scope, not a census. The comparison covers messages younger than 90 days; older posts sit outside the window, and a mark on a post nobody re-reads can be neither confirmed nor cleared.

It is a statement about a run, not about an industry. Zero differences among those thirty-eight is a finding about those messages at that moment. It is not evidence that published calls are never rewritten, and nothing here clears any particular channel — that would take checking that channel's messages, not this count.

And it does not make the marker meaningless. It makes the marker a reason to compare, which is a real function. If a call you are holding suddenly wears one, that is a fine moment to check the wording against whatever you captured. What it is not, ever, is the finding itself.

The practical read

Three things survive from this.

The mark under a message reports a change in a category so broad that it covers the trivial and the fraudulent alike, and the same documentation shows a change can also be displayed as no change at all.

The manipulation people fear is real, but it is only visible as a difference in text, which means somebody has to have kept the earlier text.

For a follower, that somebody is you: once per call, at the moment you act, in less time than it takes to place the order — and it is the only part of this that is inside your control.

Current ratings for the channels we can score are at Signal Providers. Nothing here is financial advice.

Sources

  • Telegram API, message constructor — edit_date ("Last edit date of this message") and edit_hide ("Whether the message should be shown as not modified to the user, even if an edit date is present")
  • Telegram bug tracker, report 45276 (November 2024) — a user report describing the edit timestamp moving when a reaction is added
  • Telegram API, messages.editMessage — documented editing errors, including MESSAGE_EDIT_TIME_EXPIRED ("You can't edit this message anymore, too much time has passed since its creation")
  • IC3 (FBI Internet Crime Complaint Center) FAQ — what to retain when filing a complaint about online fraud
  • Our own figures come from re-reading published signal messages against the text stored when each was first seen, as of August 2026, with a 90-day comparison window: a run of 100 messages (38 marked, 0 of those 38 changed) and a wider pass over 3,448 signals in 15 channels (3 changed).