The same fixed stake produces wildly different risk from one signal to the next, because the stop distance changes underneath it. In our corpus the quartiles sit at 2.47% and 10.16%, which is a factor of four.
Why is size the only part of a signal you control?
Because everything else was decided before you read the message. The pair, the direction, the entry, the targets and the stop all arrived as text. What you choose is how much of your account stands behind them.
That makes sizing the whole of your risk management when you follow someone else's calls, and it is the part of the process almost nobody publishes, discusses or checks. Channels advertise their hit rate. Nobody advertises the size their subscribers used, and the same call produces a survivable month for one follower and a closed account for another purely on that number.
This article is about the arithmetic of that choice across a real feed, not about which number is right for you.
What does a signal actually tell you about size?
Nothing, and the field people mistake for it is the leverage line. Leverage sets how much margin the exchange asks for a given position. It does not tell you how large that position should be.
The two get conflated constantly, and the conflation is expensive. A call reading "LONG 20x" is not an instruction to commit twenty times your usual stake. It is an instruction about margin efficiency, and the reason it appears at the top of the message is that it converts a 2% move into a headline number.
Sizing from the stop rather than from the leverage line is the standard answer, and Following crypto trading signals: how it actually works works through the calculation with an example. What that article does not cover, and what turns out to matter more across a whole feed, is what happens when you apply one rule to calls whose stops are nothing like each other.
Why does a fixed position size mean inconsistent risk?
Because the stop distance is what converts a position into a loss, and stop distances vary enormously between calls. Fix your stake and you have not fixed your risk, you have handed it to whoever wrote the stop.
The spread in our own data is wide. Across 1,462 signals with a price stop, as of 7 August 2026:
| Percentile | Distance from entry to stop |
|---|---|
| 25th | 2.47% |
| Median | 4.74% |
| 75th | 10.16% |
| Mean | 10.17% |
Read what that does to a constant stake. The same position size loses about four times as much when the stop sits 10.16% away as when it sits 2.47% away, and both kinds of call arrive in the same feed, in the same format, often on the same day.
Now read it the other way, which is the direction that empties accounts. If instead you fix the loss you are willing to take, the position size has to move inversely: a call with a 2.47% stop supports a position roughly four times larger than one with a 10.16% stop, for identical risk. Most followers do neither. They take a habitual stake, and their actual risk per trade is set by a number in someone else's message.
The tails are worse than the quartiles suggest. 255 of those stops sit closer than 2% to the entry, and 377 sit beyond 10%, so the two ends of the range differ by a factor of five in stop distance. The full distribution is in Why half of signals come without a stop loss.
How do you turn a stop distance into a size?
By deciding the loss first and letting the size follow. The relationship is one line of arithmetic and it runs in the opposite direction to intuition.
position size = amount you accept losing / stop distance in percent
An account willing to lose $50 on a call with a 5% stop supports a $1,000 position. The same $50 on a 2.5% stop supports $2,000, and on a 10% stop supports $500. The loss stays constant; the position moves.
Two properties of that formula are worth noticing, because they are where followers get hurt.
- It ignores leverage entirely. Leverage determines the margin the exchange holds against the $1,000, not whether $1,000 is the right number. A follower who sizes from the stop and then picks leverage for margin efficiency is doing it in the right order.
- It breaks when the stop is missing. With no invalidation level, there is no denominator, and the formula returns nothing. That is not a technicality, and it is the situation half the time.
We are describing arithmetic here, not recommending a figure. What fraction of an account anyone should put at risk on a stranger's trade idea is not a question this site answers.
What do you do with a call that has no stop?
You cannot size it, which is the honest answer and also the useful one. Of the 3,227 signals we have parsed as of 7 August 2026, 1,647 - 51% - publish no stop at all.
For those calls the denominator in the formula above does not exist. You can invent one, and inventing one is the only defensible move if you are going to take the trade: pick an invalidation level yourself, before entering, and size from that. What you cannot do is take the call at your usual stake and tell yourself you have managed the risk, because nothing in the message bounds the loss.
There is a second consequence that arrives later. A call with no stop never resolves as a loss on the channel's record either, so the trades that hurt a follower most are also the ones least visible in the record they used to choose the channel.
How does a target ladder change your effective size?
It shrinks the position that reaches the far targets, and it changes the arithmetic you did at entry. A ladder is a series of partial exits, and after the first one you are no longer in the trade you sized.
Ladders are the norm rather than the exception in our corpus. Among the 3,227 parsed signals as of 7 August 2026, 1,259 carry six targets, 499 carry seven, 458 carry three and 352 carry five.
Consider what that does. You size a position against a stop, take a third off at the first target, and the remaining two thirds now face the same stop distance with two thirds of the stake behind it. Your maximum loss has fallen, which is fine, but so has the payoff you were sizing for, and the trade that was designed to return the ladder's full run now returns a fraction of it.
The failure mode is the reverse case. A follower who takes the first target and then adds back, or who moves the stop to entry and re-enters, is running a different position from the one they sized, with an invalidation level that no longer bounds anything. Whether that is a good idea is a strategy question; that it invalidates the sizing arithmetic is not.
What happens when you follow more than one channel?
Your effective size is the sum across all of them, and the correlation between the calls is usually higher than it looks.
Two mechanisms drive that. The first is ordinary. Crypto pairs move together, so three long calls on three altcoins are close to one leveraged bet on the same market direction. Sizing each at your per-trade limit means running three times that limit on a single idea.
The second is specific and measurable. Three of the seventeen publishing accounts in our index are one feed under three brands, which is visible in the data as three channels with 298 outcomes each and hit rates within a fifth of a percentage point of one another. A follower subscribed to all three receives the same call three times, sizes it three times, and believes they have diversified. The count of what is actually independent in our index is in Only 14 of the 47 signal channels we index still post.
The practical version: your per-trade limit has to be a limit on the idea, not on the message. Two channels calling the same pair in the same direction is one position.
What does leverage actually change here?
The margin, the liquidation distance and the funding bill, not the size of the bet. The bet is the notional, and the notional is what the market moves against.
That distinction gets lost because the exchange screen shows returns as a percentage of margin. A $1,000 position at 10x uses $100 of margin, and a 2% move shows as 20% on that screen. The gain and loss in dollars are identical to the same $1,000 position at 2x. What changed is how close the liquidation price sits and how much of the account is tied up.
Our corpus carries the extreme end of this. 2,965 of the parsed signals state a leverage multiple as of 7 August 2026, and the highest is 200x, where the entire margin behind a position is a move of one two-hundredth, or 0.5%. At that setting the stop-based sizing question becomes secondary to a simpler one: whether the position survives the noise between the entry and the first target. What happens to an account when it does not is set out in What account blow-ups really mean.
Funding is the quieter one. It is charged on the notional rather than on the margin, so a leveraged position pays the same funding as an unleveraged position of the same size while tying up a fraction of the capital. Held across settlements, that is a drag on a trade that is doing nothing except waiting for a target.
Where does the arithmetic go wrong in practice?
In four places, and none of them is exotic.
Fees are charged on the notional too. A ladder with several partial exits plus a stop means four or five fills, each one charged on the position rather than on the margin. On a 2% target that is a meaningful share of the result before anything goes wrong.
The fill is not the published price. A stop is an instruction to the exchange, and instructions fail. The CFTC's consumer advisory puts it plainly: "a stop-loss order might be executed at a better or worse price, or not be executed at all" (CFTC). Sizing assumes the stop caps the loss. In a fast market it caps it approximately.
Minimum sizes and step sizes round you up. On a small account, the position the arithmetic asks for may be below the exchange's minimum, and the available position is larger than the calculated one. The honest response is to skip the call rather than take a size you did not choose.
The entry you got is not the entry that was published. Sizing from the published entry while filling somewhere else changes the real stop distance, usually by making it wider. In our corpus the most common outcome of all is that price reached the target before the published entry ever filled. 1,554 of 3,167 outcomes are cancelled that way as of 7 August 2026, one channel supplies 1,197 of them, and across the other sixteen publishers the rate is 21.9%. The breakdown is in Half the trades never happen.
What this does not prove
Nowhere above is there a recommendation about how much to risk, on what, or whether to follow signals at all. It describes arithmetic and the shape of one dataset. Nothing here is financial advice.
The stop-distance distribution describes the 1,462 signals in our index that published a price stop as of 7 August 2026. Those come from 17 publishing accounts out of 50 indexed, which is not a random sample of the market, and the figures move as the replay catches up. A different feed would produce a different spread, and the point survives either way: whatever the spread is in the feed you follow, a constant stake does not produce constant risk across it.
Sizing correctly does not make a bad signal good. A well-sized position in a feed with no edge loses money more slowly, which is a real benefit and not the one people hope for. Whether a feed has an edge at all is the separate question worked through in Win rate: why 90% accuracy can still lose you money.
Nothing here alleges misconduct by any channel. Publishing a wide stop is not wrongdoing, and a channel is not responsible for a follower's sizing.
The practical read
Three questions convert a message into a position, and only the third one is about the channel.
What is the distance from this entry to this stop? What position size makes that distance cost what I have decided to accept? And have I already taken this same idea from another channel today?
Where the channels we can score currently stand is at Signal Providers, and the rules our replay applies are on the Methodology page.