What Percentage of Breakouts Fail? Data From 1,611 Signals

Ask the internet how often breakouts fail and you'll get confident percentages with no dataset in sight. Seventy percent. Eighty. "Most." Numbers that circulate through trading forums the way ghost stories circulate at campfires — repeated forever, sourced never.

We're in a position to count. Every signal our engine fires is tracked from entry to exit and published weekly, winners and losers. So we pulled every completed breakout-family signal in our database — 1,611 of them, February 24 to August 7, 2026, across 15 instruments — and did the arithmetic.

75.5% failed. The sample still made +337R.

Both halves of that sentence are the article.

The number, precisely

Every signal in the sample is a breakout entry — breakout continuation or Asian range breakout — with a defined entry, stop, and target at fire time. This is the pattern most of the failures die as:

Three things can happen after a signal fires:

So the citable numbers, defined strictly:

• 75.5% of breakouts failed to follow through to their measured objective.
• 66.7% — two in three — reversed all the way through the entry to the stop.

Two honesty notes, because a stat is only worth citing if you know what's inside it. First, we count strictly: 79 signals expired in profit but never tagged the target — they're in the failure column anyway. Second, these were filtered breaks. Every signal here had already cleared momentum and level-quality checks before it fired. The naked level pokes that never made it past the filters — the ones most traders chase manually — fail more often than this.

Why a 75% failure rate still paid +337R

Here's the table the failure-rate ghost stories never include:

The average loser costs just under one unit of risk — that's the stop doing its job. The average winner pays back three. At that ratio you break even winning 24% of the time, and the sample won 29.4%. That five-point gap, compounded across 1,611 signals, is +337R.

This is the entire economics of breakout trading in one paragraph: the failure rate is the price of admission, and expectancy is the product. You don't get the +3R runners without paying the −1R tolls on the way. Anyone selling you a way to skip the tolls is selling you the ghost story.

And it puts a number on a mistake we see constantly: taking profit at +1R "to be safe." At a 29.4% win rate, capping winners at one unit of risk turns +0.21R per signal into roughly −0.39R — the same signals, the same failure rate, now a losing system. The 75% failure rate makes letting winners run mandatory, not stylistic.

Failed breakouts die fast — and that's useful

The market grades a breakout almost immediately. Half of all failures were dead within two hours; winners took three times longer to mature. A break that's going to work starts working — a break that loiters around the entry, going nowhere, is usually telling you which column it's headed for.

Practically: the fast failure is a feature. Each loser costs −1R and gets you an answer in under two hours. That's cheap information — as long as the stop is where the system put it, not where it stopped hurting.

The session paradox

Bucketing the 1,611 signals by the session they fired in (same windows as our best-time-to-trade study):

Read that twice, because it dismantles the obvious strategy of "trade the sessions where breakouts fail least."

The lowest failure rate on the board — off hours at 66.7% — lost money. The two most profitable windows, Asia and the London morning, failed at nearly the highest rates in the sample. And New York managed the only clean sweep: the most failures and a negative average.

Failure rate doesn't rank sessions. Expectancy does. When an Asian-hours or London-morning break holds, it travels far enough to pay for a crowd of failures; when a US-afternoon break holds, it usually doesn't go anywhere worth the wait.

Timeframes: same failure rate, very different paychecks

(Daily and above: only 42 completed signals, nearly all deadline expiries — too few to say anything honest, so we won't.)

The failure terrain is roughly flat — call it 70–78% everywhere. The economics aren't: H1 breakouts paid +0.63R per signal, the best single cut in this entire dataset, while M15 churned near breakeven despite the biggest sample. Faster charts don't fail much more often; they just pay less when they work — one more reason timeframe selection matters more than most traders think.

By instrument: silver is the fakeout king

A few standouts from the per-instrument cut (minimum 30 signals):

• JPY crosses held best. EUR/JPY had the lowest failure rate in the book — 68.0% — and paid +0.63R per signal, with GBP/JPY and USD/JPY right behind. If you want breakouts that stick, the yen board was the place.
• Silver was the fakeout king: 91.1% of XAG/USD breakouts failed. Fewer than one in eleven reached target. Cite that the next time someone calls a silver break "clean."
• The Dow proved the whole thesis in one row. US30 failed 76.9% of the time — worse than average — and still averaged +0.95R per signal, the highest payout of any instrument. Its rare winners were enormous.
• Cable and crude leaned trap-heavy: GBP/USD failed 80.8%, WTI 80.6%.

Same engine, same rules — the levels behave differently by market. Worth knowing before you treat every chart the same.

Longs, shorts, and one honest regime note

Long breakouts failed 74.1% of the time and averaged +0.30R. Shorts failed 80.6% and averaged −0.15R. In this stretch of 2026, breaking down was a meaningfully worse trade than breaking up.

We'd resist tattooing that one anywhere: it's five-plus months of a broadly risk-on tape, not a law of markets. But it's a live reminder that "the setup" and "the tape it fires into" are separate inputs — and only one of them is printed on the chart.

Is 75% just one bad month? No.

Monthly failure rates across the sample: 71.7% (March), 71.1% (April), 82.3% (May), 73.3% (June), 84.0% (July). In no full month did even three in ten breakouts follow through.

This isn't a drawdown statistic. It's the terrain breakout traders walk on, every month, in every regime we've tracked. Systems that survive it are built for it — which is exactly why win-rate marketing should make you reach for your wallet and leave.

How to actually trade a 75% failure rate

Accept the base rate — then filter, knowing what filters do. Confirmation rules — candle closes beyond the level, higher-timeframe agreement, session awareness — genuinely improve the odds. What they don't do is make breakouts "safe." Our signals carry those filters and fail three times in four. Filters buy you percentage points; they don't repeal the base rate.

Size every trade for the −1R outcome. Two of three breaks come back through the entry. If a full stop-out costs more than you can shrug at, the failure rate will eventually find you. This is why we quote every result in R — risk first, reward as a multiple of it.

Let winners reach their targets. The math above is unambiguous: at these win rates, +1R profit-taking converts a winning system into a losing one. The 24.5% that follow through must be allowed to pay +3R.

Put the failure where it pays. Same strategy, different clock: H1 beat M15, Asia and London beat the US afternoon, yen crosses beat silver by a mile. You don't need a better breakout — you need your breakouts fired where the winners run.

That last part is the job our engine does all day: it watches the levels, applies the confirmation filters, tags every alert with its session and timeframe context, and then publishes the results — the 24.5% and the 75.5% alike. If you'd rather inherit that discipline than rebuild it, start free.

The honest caveats

This is one firm's tracked data, not a universal constant. The sample is 1,611 completed breakout-family signals (breakout continuation + Asian range breakout) on FX majors and crosses, gold, silver, oil, and US index CFDs, February 24 – August 7, 2026. Every signal had momentum/level filters applied before firing, exits are fixed target/stop with a 72-hour evaluation deadline, and the window covers one broad market regime. A different strategy, exit model, or tape produces a different number — which is why we'll rerun this report and update the figures as the dataset grows. When the number moves, this page moves with it.

Past performance is not indicative of future results. Signals are educational market alerts, not investment advice. Trading involves risk.

Frequently asked questions

What percentage of breakouts fail? In our tracked data — 1,611 completed breakout signals, February 24 to August 7, 2026 — 75.5% failed to reach their target, and 66.7% reversed fully to the stop. Only 24.5% followed through. These were filtered signals; unfiltered level breaks fail more often.

Why do most breakouts fail? Obvious levels are where stops cluster. Price gets pushed through a high, low, or pivot to collect that liquidity, then snaps back — the classic false breakout. Thin sessions, news spikes, and counter-trend breaks make it worse.

Does a high failure rate mean breakout trading doesn't work? No — the same sample netted +337R. Losers averaged −0.97R, winners +3.04R, so the system breaks even near a 24% win rate and ran at 29.4%. Expectancy pays, not accuracy.

Which session has the most false breakouts? New York: 80.2% of its breakout signals failed and it averaged −0.11R. But don't rank sessions by failure rate alone — Asia and the London morning failed ~78% and were the most profitable windows in the data.

Do breakouts fail more on lower timeframes? Barely — M15 failed 74.9%, H1 77.5%, H4 69.9%. The real difference is payout: H1 averaged +0.63R per signal while M15 sat at breakeven.

The takeaway

Three out of four breakouts fail — measured, not folklore, and steady across every month we've tracked. The edge was never in dodging the failures. It's in paying −1R tolls quickly, in the right sessions, on the right charts, so the one break in four that runs pays +3R and the ledger still reads +337R.

The market pays expectancy, not accuracy. We got the receipts — every Saturday, in The Dossier.