Across every way we tried to parse them, the reward at a published call's nearest exit came out around 0.4 times the risk. A setup that lopsided loses its arithmetic to a delay long before it looks stale.
What changes when the price moves before you fill?
A published call is three distances: entry to target, entry to stop, and current price to entry. Only the third is visible as movement on a chart. The other two change at the same moment, in opposite directions, and nothing in the message updates to say so.
Drift toward the target shrinks what remains to be gained and widens what stands to be lost. A follower reading the post twenty minutes later sees the three numbers that were written at publication, arranged the same way, describing a trade that no longer exists at those levels.
How much that matters depends on how lopsided the two distances were to begin with.
How lopsided is a published setup?
Heavily, and toward the wrong side. Across the archived channels, the reward at a call's nearest exit comes out at roughly 0.4 times the risk: the first target sits about four tenths as far from the entry as the stop does.
That single figure is the one worth carrying, because it is the one that survived being attacked. We parsed the archive several ways, an independent recomputation swept about fifteen hundred variants of the parse, and across all of them the median per-signal reward-to-risk at the first target landed between 0.40 and 0.42. The share of calls whose first target sits nearer than the stop moved more, between roughly two thirds and nine in ten depending on how strictly the fields are read, but it never approached parity.
What did not survive is anything absolute. Median distances to the stop ranged across a factor of two and a half between parses, and median distance to the first target across a factor of sixteen. Those numbers appeared in an earlier version of this article and have been removed, because a figure that moves that much under a defensible change of method is not a measurement.
So the claim here is deliberately narrow. At the nearest exit, a published call offers well under half a unit of reward for a unit of risk, and it does so consistently enough that no reasonable way of counting makes it look otherwise.
Why does that shape make delay so expensive?
Because a late fill takes from the small number and adds to the large one. Reward shrinks by exactly the distance the price has travelled; risk grows by the same distance. When the two start far apart, that trade is punishing.
Work it through on a call whose reward is 0.4 times its risk. Call the stop distance S, so the target sits 0.4S away. Filling after price has travelled a quarter of the way to the target leaves 0.3S of reward against 1.1S of risk, a ratio of 0.27. Filling halfway leaves 0.2S against 1.2S, a ratio of 0.17.
| Fraction of the way to the target already travelled | Reward left | Risk now | Reward-to-risk |
|---|---|---|---|
| None | 0.40 S | 1.00 S | 0.40 |
| A quarter | 0.30 S | 1.10 S | 0.27 |
| A half | 0.20 S | 1.20 S | 0.17 |
| Three quarters | 0.10 S | 1.30 S | 0.08 |
| All of it | nothing | 1.40 S | zero |
The table is arithmetic rather than measurement: it follows from the 0.4 ratio and holds whatever the absolute distances are. That is the point of building the argument on the ratio, and the reason this version of the article does not print a percentage move.
The whole reward is gone once price has travelled the distance to the first target, which on a lopsided setup is a short trip. Past it, the position's nearest exit is a loss, and the post looks exactly as it did before.
How late is the market by the time a follower reads the call?
Frequently past the entry altogether. In the replayed index, 1,554 of 3,167 resolved outcomes are cancelled, meaning price reached the 2% target zone before the published entry ever filled, as of 7 August 2026.
That figure travels with a caveat everywhere it appears, because it describes a concentration and not a norm. One high-volume channel supplies 1,197 of those 1,554, and across the other sixteen publishers the rate is 21.9%, as of 7 August 2026. The breakdown is in half the trades never happen.
Even at the lower rate, roughly one call in five is gone by the time it can be acted on. A follower entering anyway has moved to a different and worse trade that happens to share a ticker with the published one, while the published record is still scored on the original levels. That gap is one reason a channel's percentage and a subscriber's results diverge.
When is a call unusable at any price?
When it never stated where it ends. Without a stop there is no published failure condition, which removes the ability to size the position, to score the call afterwards, or to say later whether it worked.
This is the majority case rather than an edge case: 1,647 of 3,227 parsed signals, 51%, publish no stop at all, as of 7 August 2026. Separately, 104 signals name no target, and the snapshot does not record how far those two groups overlap, so they cannot be added.
Nothing about the current price repairs the omission. A call without a failure condition is not a trade description with a field missing; it is a claim that cannot be graded by anyone, including the person who wrote it. Why channels publish this way is in why half of signals come without a stop loss.
What about a call whose levels contradict its direction?
It cannot be executed as written. Across the parsed corpus, 31 signals carry levels inconsistent with their stated direction as of 7 August 2026: a long whose stop sits above the entry, or a short whose target sits on the wrong side.
Individually these are typos, and everyone produces them.
Collectively they describe a process. A desk reading its own posts before sending catches a stop on the wrong side of an entry immediately, because it is the most visible error a signal can contain. An operation forwarding calls it did not generate has nothing in the loop to catch it.
The post is unusable either way, and the response to a correction request carries information the original did not: whether one arrives, how fast, and whether the corrected version quietly changes anything besides the broken field.
Does widening the stop repair a late fill?
It restores the printed ratio while enlarging the loss, which is a different thing from repairing it. A position filled late, with the stop moved out far enough to show the original reward-to-risk, now risks more money to reach the same exits.
The relationship between stop distance and position size is arithmetic rather than preference: the further the stop, the smaller the position carrying the same money at risk. What that does to sizing is in position sizing for signal followers.
Where a stop belongs is a trading decision and belongs to whoever carries the risk. This article names the trap and stops there.
What conditions describe a stale call?
Four, each a statement about the published call rather than about what anyone ought to do with it. Each is checkable against the post and the current price without an opinion about the market.
- Price has travelled a meaningful fraction of the distance to the first target. The fraction is computable from the post's own two numbers, and the table above shows what each fraction does to the ratio.
- The first target sits behind the current price. The nearest exit has become a loss, whatever the further rungs still offer.
- The call publishes no stop. No price level repairs a missing failure condition.
- The levels contradict the stated direction. The call is not executable as written.
A fifth is softer and about the post rather than the trade: a call whose reward-to-risk cannot be worked out from what is written has not supplied enough to be evaluated. The checks that apply before any money moves are in how to verify a crypto signal channel.
What does a skipped signal cost?
Nothing, and the asymmetry is the whole argument. A call not taken costs a follower zero, while a call taken after its geometry has inverted costs whatever the stop is now worth, and the stop grew wider during the delay that caused the problem.
Channels publish continuously, which is what makes the asymmetry practical rather than theoretical. Another call arrives on the same feed within hours, with its arithmetic unspent. The scarce resource was never the signal.
Most of what separates a follower who loses slowly from one who loses quickly turns out to be arithmetic performed before a position exists. The two numbers are in the post; the only question is whether they still describe what is on the screen.
Sources
The reward-to-risk figure, and how to reproduce it. Computed from the archived public posts of 39 Telegram signal channels, 23,707 messages collected between January 2021 and August 2026, of which 16,694 are unique texts once exact duplicates are removed. The parser is committed to the repository at work/when-to-skip-a-signal/parse_geometry.py and the article publishes what it prints. It reads each field from its own line, requires an explicit separator between a label and its number, drops percent-bearing lines, and discards any post whose stop and targets contradict its stated direction.
Why only the ratio is published. An earlier version of this article printed median distances to the stop and to the first target. Those did not reproduce: across parses they moved by factors of two and a half and sixteen respectively, and they have been removed. The median per-signal reward-to-risk landed between 0.40 and 0.42 across every variant tried, including about fifteen hundred swept by an independent recomputation, and the share of calls with the first target nearer than the stop stayed between roughly two thirds and nine in ten. Those two are what the article claims.
A concentration caveat that applies to this sample. The parsed set is not spread evenly across the archive. Two channels supply roughly three quarters of it, and nine channels supply all of it, so it describes the publishing habits of the channels whose format a parser can read rather than the market as a whole.
The replay figures. From the ChainRated index as of 7 August 2026: 3,227 parsed signals across 459 pairs, 3,167 resolved outcomes of which 1,554 cancelled, one channel supplying 1,197 of those, 21.9% across the other sixteen publishers, 1,647 signals with no stop, 104 with no target, and 31 with levels inconsistent with direction. A different and larger corpus than the archived dumps, overlapping but not identical, and never combined with it.
On reward-to-risk. Distance from entry to first target divided by distance from entry to stop, taken per signal and then medianed. Reporting it the other way round, as a ratio of the two medians, gives a different number; this article uses the per-signal median throughout and says so because mixing the two silently was one of the defects found in review. It ignores fees and funding, both of which make the real figure worse.
Our own definitions. How a published signal becomes a resolved outcome is on the methodology page, and what a complete call contains is in how following crypto signals actually works.
What this does not establish. These are levels channels published, not prices anyone obtained. Nothing here recommends taking or avoiding any trade. Nothing here is financial advice.