How many breakouts actually continue?

Of the 14,162 times a day pushed past the previous day's high or low, 50.1% closed outside and 49.9% came back inside. Breakout trading isn't about spotting the break: the break, on its own, is a coin flip.

The uncomfortable question about this style

Breakout trading is always explained the same way: price clears a level, the clearing confirms strength, and you ride it. The part left unexplained is how often that confirmation confirms anything. It is an answerable question, and the answer leaves little room for interpretation.

We measured it with our own hourly bars: 5.7 years, January 2021 to September 2026, ten pairs. We defined a breakout in the simplest, least arguable way we could think of: a day whose high clears the previous day's high (or whose low pierces the previous low). Then we asked one thing: did the day close on the far side of the level, or did it come back inside? 14,162 breakouts in total.

The result

The second column counts the pair's breakouts. The third is the share that closed beyond the broken level. The fourth, the share that came back inside.

Pooling all 14,162 events, continuation runs at 50.1%. By pair, the median is 50.5%, and not one pair sits more than 4 points away from the coin, and the whole spread runs from EUR/CHF's 46.2% to USD/JPY's 53.5%.

It's worth saying out loud what this means and what it doesn't. It does not mean breakout trading doesn't work. It means the fact of breaking carries no usable information by itself: if your rule is "enter when it breaks", your edge is zero before the spread and negative after it. Whatever makes this style work has to come from somewhere else.

So where does the edge come from?

From three places, none of which is the break.

From the filter in front of it. Not all breaks are alike, and the difference between a real breakout and a liquidity sweep is exactly what this raw number cannot tell apart. Our 50.1% is the result of filtering nothing: it is the floor a filter has to prove it beats.

From the asymmetry of the outcome. If you are right half the time but the continuations travel further than the reversals cost you, the style wins. That turns the question into one of risk-reward ratio and metrics, not of level detection.

From context. The same level broken on the quiet pair in the dead hour is not the same event as on the active pair during the session overlap. That is precisely what we measured in this series' anchor post: one identical rule flips sign depending on the pair and the hour, and a breakout is no exception to that.

Who does it fit, and who does it not?

It fits someone who can be right half the time without it dismantling their discipline, who has an exit plan written before entering, and who judges results by expectancy rather than by hit count. It fits badly anyone who needs to feel right: in this style you are right half the time by construction, and keeping discipline at that hit rate is the real difficulty — not finding the level.

What this post doesn't give you

• It is one breakout definition among many. "Clearing the previous day's extreme" is simple and checkable, but it is not a channel, a weekly high, or a structural level. A different definition would give a different number, and choosing it is exactly the work of trading this style. The other face of this same measurement — how many days end up inside the previous range — is in how often does price go nowhere?.
• It measures the close, not the trade. A day that comes back inside may have handed you 40 pips before returning. We measured where it ends because that is the only unambiguous point; what you would have collected depends on your exit.
• It excludes cost. With the coin at 50.1/49.9, the cost per trade decides the sign of the result, and we store no bid and ask prices to compute it.
• Daily scale. At hourly or minute resolution the proportion may differ; we have not measured that.

The break isn't the signal: it's the starting line

Almost the entire literature of this style is devoted to identifying the breakout, when that is the part which, once measured, adds nothing. The 14,162 cases say the coin is fair and the work begins right afterwards: which filter you apply, where the stop goes, how far you let it run, and what hour you trade. That turns the 50.1% into the most useful figure in the post: it is your floor. Any filter you add — the hour, the session, the size of the previous candle, the distance to the weekly high — has to lift it from there to be worth anything, and now you have something to measure it against. Almost nobody trading breakouts knows what their floor is; you have just seen yours.