The gap between a channel's published results and what lands in your account comes from a short list of places. Each one is a column you can write down.
Why keep a log if the channel already publishes results?
Because the channel's results and your results are answers to different questions, and only one of them is about your money.
A channel's number describes what its posts would have produced under the assumptions it chose: that you got its entry price, that you exited where it says, that you took every call, that costs are somebody else's problem. Your number describes what happened to your balance. The gap between the two is real, it has a small number of identifiable causes, and it is invisible unless somebody writes it down. Nobody else is going to.
This article is about that record. It does not tell you what to trade, and a log is not a strategy.
Where the two numbers come apart
Five places, and each one becomes a column.
You did not get the entry. A signal quotes an entry price or a zone, and price does not always come back for you. Across 3,167 replayed signals in our corpus, 1,554 were cancelled: price reached the target before the published entry ever filled, as of August 2026. The headline rate of 49.1% is misleading on its own, because one publisher supplies 1,197 of those and across the other sixteen channels the rate is 21.9%. Read it as one in five on a typical feed and four in five on the worst one. The breakdown is in half the trades never happen.
You exited somewhere else. Ladders have several targets and the advertised percentage belongs to the last one. Where you actually closed, and what happened to the remainder, decides your result. In the signals we parsed that publish an entry, a stop and a ladder, the median trade needs 60.1% of the position closed at the first target simply to break even if the rest stops out. That arithmetic is in you took the first target and still lost money.
You skipped some. Anyone following a busy feed takes a subset. The moment you select, your results stop being the channel's results and start being yours, and the selection is now part of the strategy whether you meant it or not.
Costs came out of yours and not out of theirs. A round trip on standard fee tiers costs roughly 0.08% to 0.11% of the position before spread and funding, as of August 2026. Against a 2% target that is 4% to 5.5% of the entire move. Advertised hit rates are gross; your account is net. See what fees and slippage do to a 2% target.
You sized them differently. Two people following the same calls with different position sizes get different outcomes from identical signals, and the one who sized the losers larger can lose money on a winning feed.
None of those five require anyone to be lying. They are the ordinary distance between a published record and a real account, and they add up to a number that no channel can compute for you.
The seven fields
Keep it small enough that you will actually do it. A spreadsheet is fine. Per trade:
| Field | Why it earns its place |
|---|---|
| Date and channel | Lets you compare feeds later, and separates them when you follow more than one |
| What was published | Entry, stop and targets, copied as posted. This is the claim |
| Did you take it, and why not | The skipped rows are the ones that reveal your selection |
| Your fill: price and time | The difference from the published entry is a cost you can measure |
| Size and leverage | Without it, no row can be compared to any other |
| Your exits: price, fraction, time | Where the money was actually made or lost |
| Net result in currency | After fees. The only figure that is unambiguously yours |
If seven feels like too many, three will still work: what was published, what you actually did, and the net in currency. The others make the answer sharper. Missing rows are what makes a log useless, so a smaller record you maintain beats a thorough one you abandon.
Two habits are worth more than any field. Write the row when you enter, before you know how it ends, since a log filled in afterwards inherits every bias you were trying to measure. And keep the skipped signals: a log containing only the trades you took cannot tell you whether your selection helped or hurt, which is one of the few things it is uniquely able to answer.
What a month of this can and cannot tell you
It can answer some questions immediately.
Am I getting the published entries? Compare your fill column to the published entry column. A persistent gap in one direction is a cost, and it is measurable in the first week.
What did I actually pay? Add the fees column. People are routinely surprised, particularly on a feed that posts often.
Is my selection helping? Compare the trades you took against everything published. If the ones you skipped would have done better, your filter is costing you money, and that is worth knowing early.
Am I following the plan I thought I had? The exits column answers it. This is the most uncomfortable column and usually the most useful.
Now the limit, and it is a hard one. A month is not enough trades to judge a hit rate. A percentage from a small sample has a confidence band around it wide enough to swallow the difference between a good feed and a bad one. To pin a hit rate to within about 10 points either way takes roughly 93 scored trades; to get to 5 points either way takes around 381. Those thresholds are worked out in what sample size makes a hit rate real, and they apply to your log exactly as they apply to a channel's.
So a month of logging answers the mechanical questions with confidence and the statistical question with none. Treat an early P&L as a description of what happened, and keep going.
One more thing your log will surface that a channel's page never will: how many of its calls have no determinable outcome at all. In our corpus 1,653 signals, about 51%, could not be resolved as published because the channel never posted a stop, as of August 2026. In your log those show up as rows you cannot close, and a feed that generates many of them is telling you something about itself.
The part that is only in your log
Your log holds one thing that exists nowhere else: what following this channel did to you.
That includes the trades you took at 3am and would not have taken at noon, the position you doubled after a loss, the week you stopped following the rules and the week you followed them exactly. None of that appears in any track record, ours included, and all of it belongs to the question of whether this arrangement works for you specifically.
It is also the raw material for a review other people can use. A review saying a channel is bad persuades nobody. A review saying you took 34 of its calls over six weeks, filled at the published entry on 19 of them, and finished down after fees is checkable, specific and hard to argue with. What separates the two is in how to write a review that holds up.
What a log will not do
It will not make a bad feed profitable. It measures; the measuring changes nothing on its own.
It will not settle a small sample. Twenty rows is twenty rows whoever wrote them, and the confidence band does not shrink because the data is yours.
It will not prove a channel is dishonest. A gap between your results and its published ones has five ordinary explanations above before it needs a dishonest one. What a published record can and cannot establish is set out in what a backtest can and cannot prove, and why a percentage alone establishes very little is in why win rate alone means nothing.
It will not tell you how much to risk. That decision comes before the log and is covered in position sizing for signal followers.
The practical read
A channel can publish whatever it likes about its calls. What it cannot publish is what happened in your account, and that is the only figure that decides whether following it was worth doing.
Seven fields, written at entry, kept for the skipped signals too. In a month it will answer whether you are getting the entries, what you are paying, and whether your own filtering helps. In a few hundred trades it will start to answer the bigger question. Until then, it is still the only record of this arrangement in which the numbers are yours.
What channels have published, and what their history supports for the ones we can score, is at Signal Providers. Nothing here recommends any channel or any asset.
Sources
- Cancelled outcomes: 1,554 of 3,167 replayed signals, 49.1%, of which one publisher supplies 1,197; 21.9% across the other sixteen channels.
work/_snapshot-2026-08-07-batch2.md, line categories, as of August 2026. - Unresolvable signals with no published stop: 1,653, about 51%, same source, as of August 2026.
- Breakeven split at the first target, median 60.1%:
work/tp1-then-stopped-out/parse_tp1.pyoverdata/channel_dumps/, 126 posts from 5 channels, as of August 2026. - Round-trip fee range and its share of a 2% target: as published in what fees and slippage do to a 2% target, standard non-VIP tiers, as of August 2026.
- Sample sizes for a 20-point and 10-point confidence band: as published in what sample size makes a hit rate real, as of August 2026.