You do not need to know anything about trading to check a trading call. You need the message timestamp, a chart with history, and four questions asked in the right order.

What can a candle actually testify to?

Four prices per interval, and nothing about the order in which they happened. A candle on any exchange chart is a summary: the first trade of the interval, the last one, the highest and the lowest. Binance's market data returns exactly that per kline - open, high, low, close (Binance spot API, kline endpoint), and Bybit's equivalent carries the same four fields (Bybit v5 API, Get Kline).

That summary is evidence of a useful kind. If a price level sits between a candle's low and its high, the market traded at or through that level during the interval. If it sits outside, it did not. The extremes include the thin spikes at the top and bottom of the candle, so a level touched only for a second still shows.

What the summary destroys is the path. Inside one candle you cannot see whether price went down first or up first, how many times it crossed a level, or when. Everything below follows from that one property, because checking a signal is entirely a question of order: entry before target, publication before the move, stop before the take.

This is not a piece about reading charts to trade. It is about reading them backwards, as a record, to check a specific claim someone else made - the narrow skill of an investigator rather than a trader.

What do you need before you start?

Three things, all free.

The signal message itself, because a Telegram message carries a visible timestamp and that timestamp is the anchor for every check. Note the timezone your app displays it in.

A chart with history for the exact market the signal names. The exchange's own chart works, TradingView works, anything that lets you select a 1-minute or 5-minute interval works. Exact market matters: a perpetual contract and the spot pair of the same coin print different prices, and a level that was touched on one may have been missed on the other.

And the four numbers from the message: entry, stop, targets, direction. If some of them are missing, that is a finding in itself before any chart is opened.

Set the chart's timezone to match the one you noted on the message, or convert both to UTC. An hour of mismatch is enough to make an honest call look post-dated and a post-dated call look honest.

Question one: did price reach the entry after the message?

Find the candle that contains the message's timestamp, then look only rightward. The question is not whether the entry price traded that day. It is whether it traded after publication, and a daily candle cannot answer that.

Here is the trap. The signal says enter at a level, the daily candle's range includes that level, and the recap will happily count that as an entry. But the day is twenty-four hours, and the message went out somewhere in the middle of it. If the only trades at the entry price happened in the morning and the message was posted in the afternoon, the range still includes the level and the trade still never existed.

So drop the interval until the answer becomes visible. On a 1-minute or 5-minute chart, start at the publication minute and scan forward: the first candle whose low-to-high range contains the entry is the earliest moment a follower could have been filled. If no candle after publication ever contains it, nobody who read the message traded that call, whatever the recap says.

One honesty note in your own favour: a candle range containing the level proves price got there, not that your order would have been filled there. Everyone reading the same message queues at the same price, and a touch can be too brief to serve the queue. Which order type gets served, and how each one fails, is its own subject - order types for signal followers. For a hand check, treat a touch as a fill and know you are being generous. Our own replay makes the same generous assumption, and says so.

Question two: what came first, the target or the stop?

This is the question the daily chart is structurally unable to answer, and the reason this article recommends minutes.

Take a hypothetical signal - the numbers are invented for illustration. Long, entry 61,750, stop 60,850, first target 62,400, posted at 14:32 UTC. Now suppose the day's candle prints an open of 62,900, a high of 63,250, a low of 60,700 and a close of 61,900. The stop is inside that range. So is the target. The daily candle certifies that both levels traded at some point that day, and is silent about which traded first.

Both stories fit the same candle. Price filled the entry, dropped to the stop, and only then climbed to the target: the follower lost. Or price filled the entry and rose straight to the target: the follower won. One candle, two opposite outcomes, no way to choose between them from that timeframe. Any recap that scores this trade off a daily chart is asserting something the evidence cannot carry.

On the minute chart the same hypothetical resolves. Price touches 61,750 at 15:04, keeps sliding, trades through 60,850 at 16:11, and reaches 62,400 at 09:40 the next morning. The entry filled, the stop was hit, and the target arrived hours after the trade was already dead. A channel scoring itself on the daily candle reports a winner. The chart, read at the right resolution, records a loss.

This is why our replay engine works through 1-minute candles, highs and lows rather than closes, minute by minute in sequence - the rules are published on the Methodology page. And even a minute is not zero: a violent minute can contain both levels, the same ambiguity one notch down. The order inside that minute is unrecoverable from candle data, so a replay has to pick a rule, and ours resolves the tie against the trade rather than for it. When you meet the same situation by eye, do the same, and for the same reason: the benefit of the doubt belongs to the follower's money, not to the record being checked.

Question three: was the call published after the move?

Everything so far assumed the message came first and the market answered. Check that assumption, because it is the cheapest one to fake.

Scan leftward from the publication minute. If the entry level was trading freely in the minutes before the message, and after the message price only moved away from it, the call described an entry that existed in the past. Sharper still: if the move toward the first target was already underway when the message went out, the channel called a move that was already happening. On the minute chart this is unmistakable - the level sits behind the timestamp, not ahead of it.

Our replay treats these mechanically. A signal whose entry is never reached after publication is excluded from Accuracy: it produced no trade, so it can produce no verdict. A signal whose target zone is reached before the entry ever fills is recorded as cancelled, overtaken by its own thesis.

There is a nastier cousin this check cannot catch: the message whose numbers were edited after the outcome was known. The timestamp survives an edit; the original text does not, and you are reading the revised version. What an edited entry does to a record, and what protects against it, is covered in the entry price that changed after you bought. The chart check and the edit problem are complementary: the chart tests the numbers you can see, and your own saved copy of the message tests whether those numbers are the ones that were published.

Question four: what is the channel's "profit" a profit of?

Once you have run the first three checks on a handful of calls, reread the channel's recap posts with fresh eyes. A recap that says "+3.2% on the long" is making a compound claim: the entry filled, the position survived to the target, and the arithmetic runs from the published entry. You now know how to test all three parts.

The pattern to look for is profit claimed on entries that never filled. The market moved in the called direction, the entry sat below the move for a long or above it for a short, price never came back to it, and the recap counts the full ride anyway. The move happened. The trade did not.

At scale this is not an edge case. Across 3,927 replayed outcomes in our index as of 17 August 2026, 1,958 - 49.9% - were cancelled: price reached the target zone before the published entry ever filled. That headline carries a caveat wherever we quote it, because one high-volume channel supplies 1,501 of those cancels, and across the other twenty-one parseable channels the rate is 22.8%. The full breakdown is in Half the trades never happen. Roughly one call in five being untradeable is the ordinary case even away from the extreme, and every one of those calls is available afterwards for a recap to claim - no stop was risked, because no position existed.

A channel is under no obligation to score itself the way you would. That is exactly why the scoring rules it uses, stated or unstated, are the thing your four checks reverse-engineer.

What this does not prove

A by-eye replay of a few signals is a spot check, and it inherits every limit of the data it reads.

Candle ranges are generous to fills. Price touching a level is weaker than your order being served at it, so a hand check overstates what a real follower would have got - and so does our engine. The assumptions and their direction are laid out in what a backtest can and cannot prove.

Your chart is one venue's tape. The signal's followers may have traded elsewhere, at prices that differ by enough to flip a marginal touch. Checking the market the signal actually names removes most of this, never all of it.

A handful of calls settles structure, never rate. Finding one post-dated call or one claimed profit on an unfilled entry tells you how the channel keeps its books. It does not tell you the channel's hit rate, and five clean calls do not certify the other five hundred.

And the worked example above is hypothetical. The prices, the times and the outcome were invented to show the mechanics; no channel in our index is being described there. Nothing in this article alleges misconduct by any channel, and nothing in it is financial advice.

The practical read

Timestamp first, then rightward for the entry, then order of stop against target on minutes, then leftward for the head start, then the recap against all of it. Four questions, one chart, maybe fifteen minutes for three calls.

The point of the exercise is not to catch every channel at something. Most checks come back boring, and boring is information: a channel whose calls survive the minute chart is a channel whose record can at least be argued about. Where this fits in a fuller vetting sequence - sampling thirty calls, counting the ones that never resolved - is set out in how to check a track record in 10 minutes.

For the channels we can parse, the replay you would do by hand has already been run, signal by signal, against 1-minute candles - the results are at Signal Providers. For the ones we cannot, the chart is public, the timestamps are public, and the method above needs nothing else.

Sources