Channels show you their winners. We show you the average.
Paste a channel's messages and we score every call it made, not just the screenshots, on one fee-accurate bar. It's the bar behind our published panel: 15 channels so far, and none shows skill yet.
Free if you add the channel's signals to our independent index. A measurement of past signals, not financial advice.
What its screenshots can truthfully show
19.3% of calls touched +60%true.
What they don't show
−3.65% per call
Average return 72 hours after each call, after fees
The winners are real. The average call still lost.
Failed the barHow an audit works
About a minute from paste to verdict. You see what we read before anything is scored.
Paste or upload
Copy messages from Telegram Desktop, or upload a chat export or a CSV. What you paste is read in memory and never stored.
Check what we found
We list every signal we extracted, with its coin, direction and time, so you can confirm we read the channel correctly.
Get the verdict
Return after fees, the t-stat with and without the five best calls, best and worst price touched, the stop it would have needed, and a plain-English verdict.
The published panel
15 channels, scored daily on the same bar an audit uses. Channel names are never published: we publish the method and the measured result.
| Channel | Calls | Days | Avg per call | t-stat | Verdict |
|---|---|---|---|---|---|
| Channel O | 3953 | 120 | +0.53% | 2.09 | Could be luck |
| Channel G | 113 | 59 | +2.31% | 1.80 | Failed the bar |
| Channel C | 321 | 109 | +1.38% | 1.11 | Failed the bar |
| Channel E | 48 | 7 | +3.44% | 1.09 | Too little data yet |
| Channel K | 696 | 122 | +0.61% | 1.03 | Failed the bar |
| Channel B | 461 | 44 | +0.53% | 0.84 | Failed the bar |
| Channel H | 791 | 112 | +0.43% | 0.67 | Failed the bar |
| Channel J | 195 | 95 | +0.50% | 0.57 | Failed the bar |
| Channel L | 916 | 89 | +0.11% | 0.39 | Failed the bar |
| Channel W | 160 | 69 | −0.19% | -0.19 | Failed the bar |
| Channel M | 312 | 85 | −0.58% | -0.40 | Failed the bar |
| Channel P | 2122 | 34 | −0.23% | -0.56 | Failed the bar |
| Channel I | 135 | 64 | −3.65% | -1.54 | Failed the bar |
| Channel X | 952 | 89 | −0.47% | -1.69 | Failed the bar |
| Channel U | 240 | 99 | −2.04% | -2.02 | Could be luck |
0 of 15 show skill.
Generated 2026-10-01. Each call is measured 72 hours on, with 0.107% round-trip fees charged. Repeated calls on one coin in one day count once, and the verdict is corrected for testing every channel at once. “Could be luck” means significant on its own but not after that correction.
We ran the same test on ourselves
Our own trading scanners are scored on the identical bar. Nothing gets a pass for being ours.
Is the channel you pay for any good?
Everything you need to run an audit, and exactly what comes back.
What you can use
- Telegram (best): copy messages from Telegram Desktop (the copy carries each time), or upload its chat history export.
- Discord, X, Facebook (beta): exports that carry exact times. Copied posts show times like “3h” and can't be scored.
- A CSV with one row per signal and a header row:
symbol,side,time BTCUSDT,long,2026-03-04 14:30 ETH/USDT,short,2026-03-05T09:12:00Z SOL,buy,1772631120
- symbol: BTCUSDT, BTC/USDT, BTC-USDT.P or $BTC. Any Bybit USDT perpetual (about 900 coins).
- side: long or buy, short or sell.
- time: year first, ISO 8601, or unix seconds. No time zone means UTC. Day-first dates are rejected, not guessed.
What we need
- At least 25 independent ideas (distinct coin, side and day), and about 30 days of history for a full verdict.
- Signals from the last 3 days can't be scored yet: each needs 72 hours of price after it.
- Up to 20,000 rows or 1 MB.
What you get back
Return after fees, the t-stat with and without the five best signals, best and worst price touched, edge ratio, the stop width the channel would have needed, and a verdict in plain English. The bar is never easier than the published panel's.
Private audits need an access code. Join the list to hear when they open. If we can't produce a score, you aren't charged. See the terms and how we handle uploads.
Help set the price
The panel stays free. Leave your email to hear when paid audits open, and tell us what one would be worth to you. Your answer sets the price. Nothing is charged until then.
Questions
Is this the same as Sentinel Bots?
No, it’s a separate product. Sentinel Bots is an NFT pass to an automated bot. Channel Audit measures signal channels, including our own scanners, on one fee-accurate bar. Use the switch at the top to move between them.
What does an audit cost?
It’s free if you add the channel’s signals to our independent index (up to 3 a day). A private audit, where nothing is added to the index, needs an access code.
What if you can’t score my data?
If there are fewer than 25 independent ideas, the signals are too recent, or we don’t cover the coins, we tell you exactly why and no code is used. If you paid, you get a full refund.
What happens to what I paste?
It’s read in memory and never stored. Only the extracted list (coin, direction, time of each signal) is kept until your score is ready. If you choose to contribute, that list joins the index. Channel names are never published.
Why are the panel’s channels anonymised?
We publish the method and the measured result, not accusations. Every figure is a measurement of past signals, and anonymising keeps the focus on the statistics.
Is this financial advice?
No. It measures what a channel’s past signals did. It isn’t a forecast, and it never recommends buying, selling or following anything.