On exits
How to set a stop loss on a breakout
Published October 13, 2026
Your stop loss has nothing to do with how much you can afford to lose.
That sentence is the whole article, and it's worth sitting with, because almost every stop in retail trading is set the other way round. Someone decides they're willing to lose a few hundred dollars, divides by their share count, and puts a line there. The number comes out of their bank account. The stock has never heard of it.
There are two common ways to set a stop on a breakout, and both of them start in the wrong place.
The two wrong starting points
The account-balance stop. "I can risk $300 on this." That's a real constraint and it matters — but it's an answer to how many shares, not where the exit goes. Using it to place the line means the exit sits wherever your position size happens to put it, which is to say: nowhere in particular.
The fixed-percentage stop. A widely taught convention, associated with the CAN SLIM school, is to cut any position at roughly 7 to 8% below the buy point. As a discipline for someone who would otherwise hold a loser forever, it's a meaningful improvement over nothing, and that's why it's taught.
But it's the same number for every stock in every market, and stocks aren't the same. A quiet, heavily traded large cap might never move 7% against a correct entry. A faster-moving name might do it twice in a week while the pattern stays intact. A single percentage applied to both is loose on one and tight on the other, and the trader can't tell which they're holding.
Here's a working trader on a public forum making exactly this point to someone whose stop kept getting hit:
"Problem is you're using one single close to get you out of the trade… What you should've done is use volatility atr… place your stop outside of the noise."Trading forum, public post
Outside the noise. That's the idea the percentage misses.
The question that actually sets the level
There's a better question, and it isn't about money at all: at what price is the reason I bought no longer true?
That price is the invalidation level — the point at which the thing you identified has stopped existing. Above it, your reasoning still holds. Below it, you aren't in a trade that's going badly. You're in a different trade, one you never agreed to, built on a pattern that is no longer there.
This reframing does something useful: it turns the exit from a feeling into a fact about the chart. "I'm down and I don't like it" is a feeling. "The base has broken" is an observation anyone looking at the same chart would make.
Where the level sits on a contraction base
For a breakout from a volatility contraction pattern, the structure hands you the answer.
A contraction base is a sequence of pullbacks, each shallower than the last, tightening toward a ceiling — the pivot. The whole premise of buying the breakout is that the sellers have been absorbed: each wave of selling got smaller, and the one that matters most is the last one, because it's the most recent evidence of how much supply is left.
So the level that matters is the floor of that final contraction. It's the most recent place buyers stepped in and held. If price comes back through it, the specific claim the pattern was making — the sellers are finished here — has been answered, and the answer is no.
Notice what this gives you that a percentage doesn't. The level is derived from the stock in front of you. A very tight final contraction produces a close invalidation level; a loose one produces a distant level. The chart sets the distance, not a convention and not your account.
It also means two stocks breaking out on the same day will have stops at different distances, and that's correct. They're different patterns making different claims.
Noise versus invalidation
Every stop placement has the same enemy: ordinary, meaningless movement.
A breakout almost never goes straight up. It clears the pivot, drifts back, tests the area, then goes — or doesn't. That first pullback is normal, and a stop placed too close gets taken out by it, which is the single most frustrating outcome in trading: you were right about the direction and you got removed from the position anyway.
The forum quote above names the fix: the stop has to sit outside the range of normal movement for that particular stock. How much a stock typically travels in a day is measurable — average true range is the common tool for it — and a level inside that band isn't a stop, it's a coin flip on the next session's wiggle.
But there's a tension here that honest writing has to admit, because the two pressures point in opposite directions. Too tight, and noise ends the trade before the idea gets tested. Too wide, and a single loss is large enough to undo several wins.
There is no setting that escapes this. Anyone who tells you they've found the stop distance that never gets hit by noise and never costs much is selling something. What you can do is place the level where the structure says the idea failed, and then make the position size absorb whatever distance that turns out to be — which brings us to the part most people have backwards.
The stop sets your size, not the other way round
This is the mechanical consequence of everything above, and it's where the account balance comes back in — in its proper place, at the end. The order of operations that works:
- Find the level where the pattern fails. The chart decides this.
- Measure the distance from your intended entry to that level. That's one unit of risk.
- Choose a share count such that losing that distance costs an amount you've decided in advance you can lose.
The order most people actually use is the reverse: pick a share count that feels right, then look for somewhere to put the stop. That's how you end up with a line that has no relationship to the stock.
Done in the right order, a wide base automatically gets a smaller position and a tight base gets a larger one — the risk is held constant and the structure does the sizing. No judgment call required.
It also makes the arithmetic of targets work, which is what the win plan is built on: if the distance to invalidation is one unit, then gains can be measured in the same units, and the same number means the same thing on every trade you take.
The market changes the answer
One more factor, and it's the one most often left out: the same pattern does not deserve the same treatment in every market.
In a deteriorating tape, breakouts fail more often, and they fail faster. A stop that was reasonable in a strong market is doing a different job in a weak one. Our own scan responds to this by tightening as conditions degrade — a breakout has to clear a larger volume surge to count as confirmed, and the floor for where a stop can sit narrows. The rules for that are on the methodology page.
The general point stands regardless of whose system you use: the condition of the broad market is part of the stop decision, not a separate topic.
Decide it before the breakout, not during
Everything above can be worked out in a quiet hour with the market closed. None of it requires watching a live chart.
That matters, because the moment you'll actually need the level is the worst possible moment to be calculating it. Price is moving, the position is open, and the two loudest voices available to you are fear and the urge to be right. A trader on the same forums put the problem plainly:
"if that trade starts going against you, your gut feel is going to tell you to stick with it, while your set up will tell you to get out"Trading forum, public post
The level written down on the calm evening is the setup talking. The renegotiation that happens at 10:15 the next morning is the gut talking. A stop decided in advance and placed as a resting order in your broker is simply the arrangement that lets the first one win.
And a stop that gets moved down while a trade goes against you was never a stop. It was an opinion with a number attached.
A level that's never reached isn't a loss
Worth saying, because it confuses people looking at any honest record: a setup whose price never climbs to its pivot was never a trade at all. Nothing was bought, nothing was risked, and no stop was ever live. On our public record those rows appear as never triggered — neither wins nor losses, because treating them as either would be dishonest in one direction or the other.
That's a normal outcome, and a record that quietly dropped those rows would look much better than it should.
How we handle it
The Coil Report is a research letter for swing traders in US stocks. After every close, a scan works a defined universe down to the names showing a genuine contraction, confirmed against the company's fundamentals and the broad market's condition, and grades what qualifies A through F.
The free weekly Letter carries the market read, the watchlist by name and grade, and the pivot on the setups sitting closest to one. Signals, the paid tier, adds the invalidation price for each setup, profit-target references and the trailing rule — the full exit framework described above, written out for each name before the move rather than after it.
Every graded setup is fingerprinted the night it's published and revealed once its measurement window closes, losses included, measured on one fixed exit rule rather than one chosen afterward.
What none of it does is tell you what to buy, or what to do with your account. The levels are published identically to every reader; the decisions are yours.
Keep reading
The Coil Report is a financial publication, not an investment adviser. Nothing published is personalized investment advice or a recommendation to buy, sell, or hold any security. Trading carries substantial risk, including loss of principal. Quotations are from public trading forums, anonymized; the posters are not subscribers and their words are not endorsements. References to the Volatility Contraction Pattern and related concepts describe publicly taught methods; The Coil Report is not affiliated with Mark Minervini, William O'Neil, or Investor's Business Daily.