Where the Stop Goes
On one trade, the stop that buys the most shares, shows the best reward to risk at 3.00 and needs the lowest win rate to break even at 25 per cent is the one that is certainly wrong. It sits forty cents inside the price that would prove the trade wrong, on an instrument whose ordinary bar is forty cents. The stop that is right needs 37.5 per cent to break even, and 44 per cent once costs are in. Every column that made the first one look better was measuring your account instead of the trade.
Prerequisites: Lesson 20, which took the stop distance as given and said this lesson would supply it, and lesson 17, for b, which is where the cost of a wider stop shows up.
Lesson 20 ended with the stop distance sitting in the denominator of the sizing formula, an input it took on trust. It is time to pay for it. Everything below decides one number — how far from your entry the stop goes — and the argument is that the number is not yours to choose freely, because it is a statement about the trade rather than about you.
Two jobs, routinely done with one number
A stop answers one question: is the reason I took this trade still true? Position size answers a different one: what does being wrong cost me? They are separate questions and they have separate instruments, and the bad stop this lesson is about comes from using the first to do the second’s job.
“I’ll risk 2 per cent” sounds like an answer to the first question and is an answer to neither. It sets a distance — two per cent of the entry price, or of the account, depending on who is speaking — from a fact about your balance. The market has never seen your balance. Price moves the distance it moves for reasons that have nothing to do with how much money you brought, and a level set from your side of the screen will land where it lands.
Which sometimes means it lands somewhere perfectly sensible. That is the part worth being clear about: a percentage stop is not always in the wrong place. It is in a place chosen without reference to the trade, so when it happens to be right, it is right by coincidence, and you have no way to tell the coincidences from the rest.
What an ordinary move is, as a number
To put a stop outside ordinary movement you need to know how big ordinary movement is, and that is a measurement rather than an impression. The usual one is average true range.
The true range of a bar is the largest of three distances: high minus low, the high minus the previous close, and the low minus the previous close, the last two taken as positive numbers. The first is the bar itself. The other two exist because price does not always start where it stopped — lesson 15’s subject — and a bar that opens away from yesterday has travelled further than its own high and low admit.
An ordinary bar makes the point quietly: yesterday closed at $49.80, today ran $49.60 to $50.30, and the three candidates are $0.70, $0.50 and $0.20. The range is $0.70, the bar’s own. Now a bar that gaps: yesterday closed at $49.80 again, today ran $48.40 to $48.90, and the candidates are $0.50, $0.90 and $1.40. The bar looks like a quiet fifty cents and actually moved $1.40. Average that quantity over the last fourteen bars and you have ATR.
Two things it is not. It is not a forecast, and nothing about it says the next bar will be that size. And it is not a boundary. It is an average, so bars larger than it are not unusual; they are part of what puts the average where it is. What it gives you is a unit: a way to say “a bit more than one ordinary bar” in dollars, on this instrument, this week.
Beyond the level, not on it
Now the part the percentage cannot do. Every trade has a price at which its reason stops being true — the low that had to hold, the high you faded, the level you entered above. If you cannot name that price before you enter, you do not have a trade yet; you have a direction. Which level it is depends on why you are in. This lesson only insists that one exists and that you know its number.
The stop goes beyond it, not on it. On the level, the ordinary probe that tests the level takes you out — and the level matters precisely because it gets tested, so you are guaranteeing that the thing you are waiting for will remove you before it resolves. Beyond it by roughly one ordinary bar, and the test can happen without you.
That is the whole construction. The level decides the place and comes from the trade; ATR decides how much further, and comes from the instrument. Neither comes from your account. Deciding what being wrong costs was the whole of your account’s job, and lesson 20 has already done it.
Why tighter is not safer
The intuition against all this is strong and worth answering directly: a tighter stop loses less when it is hit, so a tighter stop must be safer.
It loses less per stop-out and it is hit more often, and those two move in opposite directions. Whether the total improves depends entirely on how many of the extra stop-outs were trades that would have worked, which is not a number you can reason about from the armchair — but you can measure it on your own record, and the first problem below is how.
The deeper answer is that risk per trade was never the stop’s job. Lesson 20 fixed it with the position size, and the amount at risk is the same whichever of the three stops below you take, because the share count moves to keep it there. Tightening a stop to reduce risk changes two things at once and controls neither.
The same trade, three stops
You are long at $50.00. You entered because $48.60 held, so $48.60 is the price at which the reason is gone. ATR is $0.40. Your risk budget is $250, from lesson 20. The next level in your favour is at $53.00, so the target is $3.00 away. Three candidate stops:
| Where the stop goes | Distance | In ATR | Shares at $250 risk | Reward to risk, b | Win rate to break even |
|---|---|---|---|---|---|
| 2% below the entry, $49.00 | $1.00 | 2.5 | 250 | 3.00 | 25.0% |
| On the level, $48.60 | $1.40 | 3.5 | 178 | 2.14 | 31.8% |
| One ATR beyond it, $48.20 | $1.80 | 4.5 | 138 | 1.67 | 37.5% |
Read across the top row and it wins every column. It buys the most shares, it has the best reward to risk, and it needs the lowest win rate to break even. On paper it is plainly the best of the three, and it is the only one of the three that is certainly wrong.
Here is why. At $49.00 the stop sits above the level by forty cents, which on this instrument is one ordinary bar. It is inside the region the trade explicitly expects price to visit. Being taken out there tells you nothing at all, because $48.60 has not been touched and the reason you entered is exactly as true as it was when you entered. You will pay $250 to learn nothing, and then watch the setup resolve without you.
The last column is the honest cost of fixing that, and it should not be hidden. Moving the stop beyond the level takes b from 3.00 to 1.67 and lifts the win rate you need from 25 per cent to 37.5 per cent. Lesson 17’s cost figure of 0.17R pushes all three up, by four points at the narrowest stop and six and a half at the widest — to 29 per cent, 37 per cent and 44 per cent — and it pushes the widest one furthest, because a wider stop is a smaller b and a smaller b is more sensitive to costs. A correct stop is not a free upgrade. It is a worse-looking trade that is a real one.
And now the sentence that indicts the percentage properly. Suppose the level had been $49.40 instead of $48.60, everything else unchanged. The same 2 per cent stop at $49.00 would then sit forty cents beyond the level rather than forty cents inside it, which is exactly right — one ordinary bar past the price that kills the thesis. The rule that produced it has not improved. It has been handed a different level and got lucky, and from inside the trade the two cases look identical.
What this does not settle
Which level is the right level. Everything above assumes you can name the price at which your reason dies, and naming it is a reading skill this course has not taught you yet — Module 4 is where liquidity, structure and the levels that actually hold get their lessons. Until then the rule still bites in its negative form: if you cannot name the price, the trade is not ready, and putting a stop at a percentage is a way of not noticing that.
Which multiple of ATR. One ordinary bar beyond the level is a starting point and not a law. Tighter is hit more often for less each time and wider less often for more, which moves risk between frequency and size without removing any of it, and the honest way to choose is the measurement in problem 1 rather than a number from a book. What is not a matter of taste is the direction: the buffer goes beyond the level, never inside it.
That the stop is the loss. It is the loss if the stop is filled where it is placed, and lesson 15 was a whole lesson on the circumstances where it is not. A stop is an order, and an order needs someone on the other side at that price.
That the target is a free choice. The last column of the table is only readable because there is a target in it, and this lesson took $53.00 as given the way lesson 20 took the stop as given. If the nearest level in your favour is closer than the stop is behind you, the arithmetic in that column stops working and there is no trade — which is the useful half of the convention that a target should be at least twice the risk. Lesson 17’s table is where that convention comes from and what it actually costs.
And that a stop may be moved. It may be moved closer as the trade proves itself, which is trade management and comes later. Moved further away it is not a wider stop; it is the absence of a stop, arrived at one decision at a time: the level that defined it has not moved, so the only thing that changed is your willingness to be wrong. Lesson 24 is about the state of mind that makes that feel reasonable at the time.
Two questions, in this order, and the second is not allowed to change the first. Where does this trade stop making sense? And how many shares can I hold so that being wrong there costs what I decided it would cost?
Problems
- Count your noise stop-outs. Take your last twenty losing trades and for each one answer a single question: when the stop was hit, was the reason you entered still true? Not “did it come back” — whether the price that would have invalidated the trade had actually traded. The fraction that were stopped with the reason intact is the only number that tells you whether your stops are too tight, and it is a fact about your own record rather than an opinion about stops.
- Compute the unit. Take ten bars on the timeframe you actually trade — not the one you like looking at — and work out the true range of each by hand, taking the largest of the three distances. Average them. Now put that number next to the stop distance you habitually use, in the same currency. That comparison is the entire content of this lesson in one line.
- Price one trade properly. Take a trade you would take now. Write down the price at which its reason dies, put the stop one ATR beyond that price, and carry the distance into lesson 20’s formula for the share count. Then compute b to your target. If b has fallen below what your win rate needs — lesson 17’s table, with costs — you have not found a worse stop. You have found out that the trade was not there, at a cost of nothing.
Sources. J. Welles Wilder, New Concepts in Technical Trading Systems (1978), which is where true range and its average come from, and which introduced them for exactly the purpose used here: sizing a stop to the instrument rather than to the trader. Charles LeBeau and David Lucas, Computer Analysis of the Futures Markets (1992), whose chapters on exits are the earliest systematic treatment of the point that an exit rule can be tested separately from an entry rule, and usually matters more. Perry J. Kaufman, Trading Systems and Methods (5th edition, 2013), for volatility-scaled stops and the trade-off this lesson prices: what a wider buffer buys in fewer stop-outs and what it costs in reward to risk.
You now have the number lesson 20 borrowed, and the two lessons together turn a level and a volatility into a share count. The next lesson asks what a run of those trades can do at the extreme: not how deep the fall is, but how likely it is to end the account outright.
Educational only. Trading involves substantial risk of loss. Not financial advice. Past performance does not guarantee future results.
💬 Discussion (0 comments)
Loading comments...
Ready to Trade with Signal Pilot?
Apply your trading education with professional indicators and real-time market analysis tools.
Back to Signal Pilot →