Breakout

Reviewed 12 August 2026 · Sourced from two Journal of Finance studies, the SEC and our own TA course

What a breakout is, and the word that breaks it

Two people are looking at the same chart of the same stock at the same minute. It has spent four months between roughly $38 and $42. It just printed $42.35. The first says it broke out this morning. The second says nothing has happened yet. Neither of them is lying, neither is confused, and there is no referee.

A breakout is price moving decisively out of a range, or through a support or resistance level it had been respecting. That is the standard definition, and some version of it appears in every book and on every charting site you will ever open.

Now read it again, because the entire term lives inside one word. Decisively. Nothing defines it. No exchange, no regulator, no standards body and no agreed practice among traders publishes a threshold for how far past a level price has to travel, over what period, measured how, against a level drawn by whom. Every argument about whether a breakout occurred is an argument about that word, and it cannot be settled, because there is nothing to settle it against.

That is not a caveat for the bottom of the page. It is the central fact about the term. Everything else here — the confirmation rules, the false breakout, the failure modes — is downstream of that one missing definition.

The four choices hiding inside “decisively”

Split the word open and you find four independent decisions. Each is free, none has a published default, and changing any single one of them can reverse the answer.

1. How far past the level

A level at $42.00 and a print at $42.01 is technically through it. Nobody calls that a breakout. A print at $45.00 obviously is. Somewhere between one cent and three dollars the word starts applying, and the boundary is set by whoever happens to be talking. There is no published figure — not a disputed one, none.

2. On what timeframe

The same action can be a breakout on a 15-minute chart, noise on a daily chart, and invisible on a weekly chart. The range price just left on the 15-minute view may sit entirely inside a daily range that is still intact. Neither reading is wrong. They answer different questions, and people mix the two in one sentence constantly.

3. On a close or on a touch

Price can trade through a level and finish the session back inside the range, leaving a wick sticking out above it. On a closing-basis rule, nothing happened. On an intraday-touch rule, something did. This is the largest single source of disagreement about breakouts, and it is why a closing rule is the most common convention — not because anyone proved it better, but because it is harder to trip.

4. Whose level

Before any of the above, you need the level itself. As the support and resistance page sets out, a level is an area a few percent wide, drawn by hand, with no standard for its width, its placement, or how many touches it needs to exist. Two analysts marking up the same chart put the line in slightly different places. That difference is often larger than the decisive move being argued about.

Four choices, and they multiply rather than add. Two plausible distances, three timeframes, two measurement rules and two hand-drawn levels give you two dozen defensible answers about one bar on one chart.

Where the word does get defined

There is exactly one place it becomes precise, and it is instructive. Researchers testing a breakout rule have to express it as code, so they are forced to state it exactly. The working-paper version of Sullivan, Timmermann and White's study of technical trading rules puts it this way: “A simple trading rule based on the notion of support and resistance (S&R) is to buy when the closing price exceeds the maximum price over the previous n days, and sell when the closing price is less than the minimum price over the previous n days.”

That is completely unambiguous. It is also unambiguous only because somebody picked a number for n, and nothing tells you which. The same paper evaluated 7,846 rules spanning filter rules, moving averages, support and resistance, channel break-outs and on-balance volume. A universe that size exists because these parameters have no principled values — the researchers had to test them all rather than justify one.

The confirmation rules, and what each one costs

Because the definition is unusable as written, traders bolt filters onto it. Four are near-universal. Every one of them is an improvement on nothing, and every one of them is a convention — a practice that is standard because enough people do it, not because a body published it or a study established it.

The volume one deserves a sentence of its own

Volume is the most-cited confirmation because it is a genuinely independent series from price — it cannot be reshaped by redrawing a line. But as the volume page sets out, it measures activity, not direction. Every share that changed hands had a buyer and a seller. Heavy volume on a breaking session tells you many people transacted at those prices. It does not tell you which side was the motivated one. You will also see a number attached — that a credible break comes on roughly 1.5 to 2 times average daily volume. That is a rough rule of thumb many traders use, and no exchange or standard-setting body publishes that threshold.

The buffer, worked in dollars

Here is where the ambiguity turns into money. The range top is $42.00. Today the stock trades as high as $42.35 and closes at $41.80. The 14-day average true range is $1.10. Two common buffer rules:

Now run it the other way, because the stricter rule is not simply better. The ATR reader who waits for $43.10 pays $0.75 a share more for the identical idea — $150 more on 200 shares. And if both place a stop just under the range low at $37.80, the patient reader risks $5.30 a share instead of $4.55: $1,060 at risk instead of $910. The buffer that filters out noise also enlarges the loss you agreed to take. No setting does both, which is the trade-off, and it is yours to make on purpose rather than by default.

Why the false breakout is built into the plumbing

The price clears the level convincingly, the volume looks right, and within two sessions it is back inside the range and heading for the other end of it. That is a false breakout, or a fakeout. It happens constantly, and there is a plain mechanical reason for it that requires nobody to be plotting anything.

Start with where orders sit. A range top at $42.00 is obvious to everyone looking at the chart. Traders who are short the stock put protective stops just above it. Traders who own it and want out of a failing position put stops just below the range bottom. That is not a secret strategy; it is the standard use of the level, and the stop-loss order page explains what those orders do when they fire. The short version, and the U.S. Securities and Exchange Commission is blunt about it: a stop order becomes a market order the moment the stop price is reached, and the stop price is not the guaranteed execution price.

So the area just above an obvious level holds a cluster of resting orders that convert to market orders if price touches them. Two consequences follow, and neither requires anyone to be hunting anybody.

• Price is drawn toward the cluster. Once triggered, those stops become market orders that push price further in the same direction, triggering the next ones. A small poke past the level can produce a burst of movement out of pure order mechanics, with no new information about the company at all.
• That burst is the deepest liquidity available. Unloading a large block requires buyers, and the most eager buyers appear precisely when a breakout is triggering stops and pulling in traders who want in on the move. A large holder filling into that is not manipulating anything — they are doing the obvious thing, which is transacting where the bids are. The effect on the chart is a sharp move up that gets absorbed and reversed.

Put those together and the false breakout is not an anomaly. It is a normal output of how orders are placed and where liquidity lives. The confirmation rules above exist entirely to filter for it.

What the research measured, and what it did not

Breakouts have been tested academically, which puts this term ahead of most chart concepts. Two papers in the Journal of Finance carry the story, and they have to be read together, because the second one is the point.

Brock, Lakonishok and LeBaron (1992). They examined two families of technical rules on Dow Jones Industrial Average data from 1897 to 1986, and one of them — the trading range break — is a mechanized breakout. Their finding, in their own framing: the returns from these strategies are not consistent with four popular null models, namely the random walk, the AR(1), the GARCH-M and the exponential GARCH. Buy signals produced higher returns with lower volatility than sell signals. That is a real result in a top journal, and more than most chart concepts have behind them.

Sullivan, Timmermann and White (1999). They rebuilt the test. Rather than a handful of rules they evaluated 7,846 — the full universe such rules get drawn from, including support-and-resistance and channel break-out families — using a bootstrap method to quantify how much of any apparent success is data-snooping, the statistical shadow cast by trying thousands of things and reporting the best one.

Their first conclusion was generous: the earlier results “appear to be robust to data-snooping, and indeed there are trading rules that perform even better than the ones considered by BLL.” The 1897–1986 finding survived the harder test.

Their second conclusion is the one to carry with you: “However, we also find that the superior performance of the best technical trading rule is not repeated in the out-of-sample experiment covering the 10-year period 1987–1996.”

What nobody measured

Notice what those studies report: returns, volatility, significance against null models. Neither reports a failure rate for breakouts, because a failure rate requires you to define a breakout first, and that definition does not exist.

Which brings us to the most repeated unsourced claim in technical analysis. “Most breakouts fail.” It appears on thousands of pages, usually with a number attached — 70 percent, 75, 80. We could not trace any of those figures to a study, and we are not going to reprint one. It goes in the ledger below as Unverified, named rather than quietly dropped.

What is fair to say is qualitative: breakouts fail often enough that essentially every trader who uses them has built a confirmation rule to deal with it. Nobody constructs elaborate filters for a rare problem. The existence of the confirmation conventions is itself the evidence that failure is common. That is an inference about practice, not a percentage, and it is the honest ceiling on what this page can tell you.

One name belongs here for completeness. Thomas N. Bulkowski's Encyclopedia of Chart Patterns (John Wiley & Sons) is the best-known attempt at systematic per-pattern performance statistics from a large database. It is one practitioner applying his own identification rules to his own sample — not peer-reviewed, and not comparable across authors who define the patterns differently. It is also not the source of the ambient claim. If someone quotes you a hit rate, ask which book, which edition, which pattern definition and which sample.

A breakout is only a breakout afterwards

Here is the structural problem the definition cannot fix, even if you settled every parameter above.

What people mean by a breakout is not a bar on a chart. It is a change of regime — price stops oscillating in a band and starts trending. But a range ending and a trend beginning are two different events, and only the first is visible at the time. A wide bar past a level is range expansion: volatility increased. Whether it becomes a trend is a fact about the sessions that follow, and those have not happened yet.

So at the moment you are looking at it, a breakout is a hypothesis. It becomes a breakout retroactively, once a trend confirms it, and it becomes a fakeout retroactively if one does not. The word describes an outcome using a bar that occurred before the outcome. That is a genuinely strange thing for a piece of vocabulary to do, and it is worth noticing, because it explains something that otherwise looks like bad luck.

Which is why every example you have ever seen worked

Open any tutorial on breakouts. Every chart in it shows a level, a clean break, and a trend afterwards. Look at ten tutorials and you will see fifty of them. Not one shows a break that failed.

That is not dishonesty by the author. It is how the examples get chosen. The writer searched charts for breakouts, and since a breakout is identified by the trend that followed, the search can only return the ones that worked. The failures were never candidates — they did not qualify as breakouts under the definition being used. This is survivorship, and here it is baked into the term rather than into anyone's motives.

The practical consequence: the pattern looks dramatically more reliable in every lesson you read than on the live chart in front of you, and the gap is not your skill. Markets · Technical shows levels and breadth forming live rather than selected after the fact, which is the cheapest way to calibrate that for yourself.

Where breakouts fail worst

Failure is not evenly distributed. Three conditions make it substantially more likely, and all three are checkable before the fact rather than after.

Thin liquidity

In a stock that trades a modest number of shares a day, a level can be cleared by an order that would not register on a large-cap chart. The breakout is real in the sense that price moved, and meaningless in the sense that it reflects one participant rather than a change in how the market values the company. It cuts the other way on the exit too: the same thinness means the price you get when you leave can sit well away from the price you saw, which is exactly the SEC's warning about stop orders in a fast market. A wide bid-ask spread on a small name can be larger than the buffer you used to define the breakout in the first place.

Earnings and other gaps

A company reports and the stock opens four dollars above yesterday's close. It has cleared the range top, and the level is now irrelevant — not broken, irrelevant. The range described where buyers and sellers had been meeting under one set of assumptions about the business. New information replaced the assumptions. Treating the gap as a technical breakout applies a framework about order flow to an event that was about information, and the chart cannot tell you which one you are looking at. Checking the earnings calendar first is how you avoid a category error.

Index-level breakouts

This is the subtlest one. When an index clears a level, what broke? An index is a weighted sum of hundreds or thousands of unrelated prices, so its “range top” is an artifact of arithmetic, not a price anyone ever placed an order at. Nobody holds a resting order at S&P 4,800 the way they hold one at $42.00 in a single stock, because 4,800 was never a price — it was an output. The order-flow story that makes a single-stock level intelligible does not transfer, and neither does volume confirmation, because index volume is millions of prints in unrelated names. The site's breadth measures exist for this reason: the percentage of S&P 500 members above their 20-day, 50-day and 200-day moving averages tells you how many components are participating, which is a different and far more answerable question than whether the index broke a line.

What is confirmed, what is convention, and what could not be traced

This section looks thin next to a page about a statute, and that is the correct result rather than a shortcoming. Almost nothing about a breakout is established by an authority. Here is the accounting.

Five confirmed, two untraceable, four conventions. Read the confirmed column closely and notice what it contains: two studies, one of which says the effect did not persist, plus a fact about order types and an absence. None of it says a breakout will work. That is the actual state of the evidence.

What trips people up

• Arguing about whether it was a breakout. The argument has no resolution, because the word has no definition. State your rule — timeframe, close or touch, buffer size — and the conversation becomes possible. Skip that and you are two people using one word for different things.
• Switching timeframes mid-thought. Spotting the break on a 15-minute chart and then justifying it with a weekly range is the most common version of this, and it feels like confirmation when it is just two unrelated observations.
• Treating volume as a verdict. It measures activity, not direction. Heavy volume says many shares changed hands, and every one of them had a buyer and a seller.
• Believing a hit rate. If a page gives you a percentage of breakouts that fail, it almost certainly cannot tell you where the number came from. Ask.
• Learning from examples that all worked. Every breakout chart in every tutorial was selected after the outcome was known. Your live chart is not filtered that way.
• Confusing an expansion with a regime change. A wide bar past a level means volatility rose. Whether a trend follows is information you do not have yet.
• Using a level right through an earnings date. New information does not break a level, it retires it. Check the calendar before the chart.
• Assuming a stop caps the loss. It sets a trigger. On a thin name during a failed break, the fill can land well away from it — see stop-loss order.
• Importing intent. Stops cluster above obvious levels and liquidity follows them. You do not need anyone to be hunting you for the fakeout to make sense.

The figures on this page are checked against the source that publishes them, and dated. Published rates move after the release named above — the linked source always carries the current number. This page explains a term; it does not recommend a product.