The State of Telegram Signal Channels: October 2026

Of the 1,820 channels we could read, 241 post at least three complete signals: asset, direction, stop loss and a target with its price. Those are the only ones anyone can check against the market. Another 110 (6%) only post results (“TP hit,” “+80 pips”) without ever posting the signal first, so there is no way to know the trade existed. The rest post analysis, promotions, signals without a stop, or aren’t trading channels at all.

Among the 898 channels we read in full, the share rises to 26.8%. The other 922 we read on their first page only, which showed no readable signal, so the true figure most likely sits between the two. Another 633 channels have no public view: they can only be read by joining, and they are not part of these numbers.

Of the 254 channels listed in our directory, 235 mostly post gold (XAUUSD). Counting any asset that makes up at least a quarter of a channel’s signals, gold shows up in 241 channels, forex in 23, crypto in 5 and indices in 4.

This says as much about our crawler as about Telegram: we started out looking for gold channels, and the sample reflects it. It is not the mix of every signal channel out there.

We counted the messages that announce a target hit (“TP1 hit,” “TP reached ✅”) and the ones that acknowledge a stop (“SL hit,” “stop loss hit”) across 478 channels. There are 6,121 of the first and 165 of the second: 37.1 to one. The channel with the most announcements accounts for 10.5% of the targets; without it, the ratio is still 33.2 to one, so this isn’t about a single channel.

Of the 114 channels with at least ten announcements, 83 (72.8%) never acknowledged a stop loss in everything we read. That doesn’t mean their signals never hit the stop: when we test them one by one against the real price, a large share ends there (see our gold signals study). What isn’t announced simply doesn’t show up in the channel.

We fully audited 16 channels: we read their history with a read-only account, including what they later deleted, and tested 1,864 trades against the real price, tick by tick and with the spread. For each channel we worked out the range its total result falls in 95% of the time, by resampling its trades. We only call a channel “profitable” when that whole range sits above zero and it has 30 trades or more.

As of September 30, no channel was proven profitable. This number moves: a channel near the line can change after a few trades. And not being proven profitable isn’t the same as losing: many channels need hundreds more trades before their result can be told apart from luck.

We read 7,436 gold signals from 147 channels with at least five signals that include a stop. For each signal we took the distance between entry and stop; for each channel, its median; and then the median across channels, so a channel that posts a lot doesn’t weigh more than another. The typical channel sets its stop $11 from the entry (110 pips); the middle half of channels, between $10 and $13.

With stops this tight, the spread and a few seconds of delay when copying matter more: we measured it here. In forex, indices and crypto we didn’t reach ten channels with enough signals, so we don’t give a number.

We spread the 14,461 signals of 107 channels with at least twenty of them across the UTC hour they were posted, giving each channel the same weight. The busiest hour is 07:00 UTC: between 07:00 and 10:59, the London open, 27% go out; between 13:00 and 16:59, the New York open, 16.2%. After 17:00 UTC posting drops off, and only 2.7% come on Saturdays and Sundays.

For someone copying from the US, the London open is 3 a.m. in New York. If you can’t be at your screen then, it’s worth knowing before you pick a channel.

Telegram labels a message as “edited” when it was changed after posting, and the channel’s public page shows it. Across 106 channels with at least twenty signals, 58 (54.7%) have at least one edited signal; overall, 900 of 14,440 signals (6.2%). The median per channel is 1%: most edit little and a few edit a lot.

Editing isn’t cheating in itself: sometimes it fixes a typo or adds “TP1 ✅” to the signal. But a signal changed after the price moves is no longer the one that was posted. That’s why, when we audit, we always test the first version we saw. This number doesn’t say what was changed in each case.

The channels. Our crawler finds them on Telegram, starting with gold, and we review them by reading their public page. A channel counts once even if it has two names; messages forwarded from another channel and repeated messages don’t count.

Signals and announcements. A complete signal is a message with an asset, a direction, a stop loss and a target with its price. A target or stop announcement is recognized by the usual phrases in English and Spanish, and is counted per message.

Why other Verpips pages give a different ratio. In early October, our gold signals study and our glossary put it at about 48 and about 56 targets announced per stop loss. Those pages also count messages forwarded from other channels and repeated texts, and add up everything read up to the day they’re calculated. Here, forwards and repeats are removed and only what was posted before October 1 counts: 37 to 1 is this edition’s number.

The verdicts. They come from full audits: the history read with a read-only account, every signal rebuilt against the real price, and a 95% confidence interval by resampling (2,000 repetitions).

Rules before data. The method was written before computing any number, in the header of the program that computes them, and anything below the set minimum (20 channels for a share, 500 messages, 10 channels per market) was left out. Cutoff: everything posted through September 30, 2026. Our method, step by step.