Inside bar: continuation, not reversal

What an inside bar is, exactly

Inside bar or harami: the extremes decide On the left, an inside bar: its high and its low sit entirely within those of the mother bar, which carries the two levels of the trade. On the right, the body of the second candle sits within the body of the first, but its high goes beyond: that is a harami, not an inside bar. Two distinct definitions, not to be mixed in the same count.

An inside bar is a candle whose high is lower than the previous candle's high and whose low is higher than the previous candle's low. It sits entirely inside the one before it, extremes included.

Two consequences of that definition, often misunderstood. The previous candle is called the mother bar, and it carries the levels that matter: its high and its low become the two boundaries of the trade. And the colour of either candle is not part of the definition, contrary to what many presentations imply.

The meaning is simple: for a whole period, the market went looking for neither the high nor the low of the previous period. Volatility contracted, range narrowed, and nobody took control. It is a pause, not a decision.

You will meet the pattern under several names depending on the school, including inside day on daily charts, inside candle, and harami in candlestick vocabulary. The harami is slightly stricter, it compares bodies rather than extremes. Those are two distinct definitions and they must not be mixed inside one study.

What the numbers say, and they contradict the usual teaching

The reference study on this pattern covers 29,641 cases, from December 1989 to January 2013, and is published by Thomas Bulkowski. Here is what it gives, quoted as it stands.

The pattern acts as a continuation 62% of the time. In other words, in close to two cases out of three, price resumes in the direction it came from. It reverses the trend in only 38% of cases. Every course that presents the inside bar as a reversal signal is therefore working against the statistic.

Failure rate on an upward breakout: 32%. Average rise: 10%. Price target met in 80% of cases. Performance rank: 10 out of 23 patterns of this type, so above the median, which makes it an honest pattern, not a miracle.

And one figure nobody quotes, because it is inconvenient for everyone: performance decays decade after decade. Eleven per cent in the 1990s, ten in the 2000s, nine in the 2010s. The author draws a sentence from it worth reading in full: it is today roughly 18% harder to make money with this pattern than it was in the 1990s. A pattern everybody knows ends up arbitraged by everybody.

Source: Thomas Bulkowski's research pages on thepatternsite.com, consulted in September 2026.

Continuation 62 times in 100, and a pattern wearing out At the top, Thomas Bulkowski's study of 29,641 cases: the pattern carries on in the direction price came from in 62% of cases, and reverses the trend in only 38%. At the bottom, its performance decade after decade: 11%, then 10%, then 9%. It is now about 18% harder to make money with it than in the early 1990s.

The two markets where you are better off leaving it alone

The same study separates markets, and the separation is clear. On stocks, the pattern beats its benchmark. On ETFs and on cryptocurrencies, it does markedly worse.

That is rare information and it deserves to be taken seriously, because it explains a very common frustration: many traders learn the pattern on US stock examples, then apply it to bitcoin or to an index ETF, and cannot understand why it returns nothing.

The likely reason lies in the nature of the contraction. On a stock, an inside day often signals a genuine pause in order flow between two sessions. On a market open twenty-four hours a day, the notion of a period is an arbitrary slicing, and an inside bar may be no more than the activity trough of an Asian session.

How to trade it, and where the boundaries go

The standard mechanism comes down to two levels, and they belong to the mother bar, never to the inside bar. You enter on a break of its high for a buy, of its low for a sell.

The stop goes on the other side of the mother bar, and that is what makes the pattern attractive: both boundaries are known in advance, so risk is quantifiable before you even enter. It is also what makes it expensive when the mother bar is tall, since the stop moves away accordingly.

Given the 62% continuation figure, the most defensible use is not to play the break in either direction, but to take the break that goes with the preceding move. It is a simple correction to make, and it puts the statistic back on your side instead of against you.

A word on inside bars in series. Two or three consecutive inside bars signal a stronger contraction, hence often a cleaner break. In exchange, the mother bar moves further back in time and its level loses its freshness.

The boundaries belong to the mother bar In a rise, a mother bar then its inside bar. You enter on the break of the mother bar's high, in the direction of the previous move, and the stop goes on the other side of the mother bar, never of the inside bar. Both levels are known before entry, so the risk is too, and it grows with the size of the mother bar.

What it gives in your own journal

The figures above cover US stocks on daily data between 1989 and 2013. If you trade an index on fifteen minutes in 2026, they concern you only indirectly, and the decade-by-decade decay is warning enough not to take them at face value.

The measurement that concerns you happens in two steps. First, tag your entries taken on an inside bar. Then, and this is the point specific to this pattern, record for each one whether you took the break with the preceding move or against it. Separate those two populations. If the general statistic holds for you as well, the gap between them should be visible.

Then look at the maximum adverse excursion of those trades. On this pattern in particular, it answers a precise question: is your stop on the far side of the mother bar earning its place, or did most of your winning trades never come close to that level? In the second case, you are paying for a wide stop for nothing and you could size up at equal risk.

Tradoshi computes that excursion on every imported trade and lets you filter the journal by label, so the two populations sit side by side without a spreadsheet.

Frequently asked questions

Is an inside bar bullish or bearish? Neither. It is a volatility contraction, and it resumes in the direction of the preceding move in 62% of recorded cases. Direction comes from what happened before, not from the pattern.

Are an inside bar and a harami the same thing? No. The inside bar compares extremes, the harami compares bodies. A candle can be a harami without being an inside bar. Pick one definition and hold it.

Which timeframe? The published figures cover daily data. The lower you go, the more arbitrary the slicing of periods becomes, and the more the pattern loses its meaning on a market that never closes.

Should volume be low on the inside bar? It is the most commonly recommended filter, and it is consistent with the idea of a pause. It is not quantified in the study cited here, so treat it as a hypothesis to verify in your own journal.

And on cryptocurrencies? The study explicitly concludes to underperformance on that market as on ETFs. If you trade crypto, the prudent stance is to treat the pattern as unproven until your own record says otherwise.

Key takeaways

High lower and low higher than the previous candle's. Extremes are what count, not bodies, otherwise it is a harami.
Continuation in 62% of cases across almost 30,000 examples: the pattern works with the preceding move, not against it.
Both boundaries of the trade belong to the mother bar, so risk is known before entry, and it grows with the size of that bar.
Performance decaying decade after decade, roughly 18% harder than in the early 1990s. A pattern everybody knows ends up arbitraged by everybody.
Underperformance recorded on ETFs and cryptocurrencies. On those markets, treat it as unproven until your own journal says otherwise.

Going further

These blog articles dig into this lesson's ideas, one subject per article.

Patterns in trading: spotting what makes you win or lose The gap and go strategy explained