Inside Bar Pattern: Why Your Chart Decides If One Exists

An inside bar is usually introduced as something the market prints: a bar whose high is lower than the previous high and whose low is higher than the previous low, read as a pause before a move. Every guide on the first page of results describes it that way.

What that description omits is the thing that determines whether the bar is on your screen in the first place. Containment is a comparison between two adjacent bars, and what counts as a bar is a setting on your chart rather than an event in the market. Change the setting and the pattern appears, moves or vanishes, with no tick of price having changed.

What the Platform Compares When a Bar Is Called Inside

MetaTrader 5 stores each bar in a structure that holds the start time of the period, an open, a high, a low, a close, and volume. The reference describes the high as the highest price of the period and the low as the lowest price of the period. Those two words carry the whole argument of this page: the extremes belong to the period, not to the day, the session or the instrument.

Nothing in that structure names an inside bar. There is no constant for it, no flag, no field. The pattern exists only when someone compares two of those records and finds that the later high sits below the earlier high while the later low sits above the earlier low. The comparison is done by a reader, an indicator or a scanner, after the platform has finished its work.

The two bars have common names. The earlier and larger one is usually called the mother bar, the contained one the child bar. Those labels are conventions of chart reading rather than platform terminology, which is worth knowing before assuming that two tools mean the same thing by them. The mechanics of how a single candle is built from the four stored prices are the same whichever label is used.

The Pattern Belongs to the Period, Not to the Instrument

MetaTrader 5 enumerates 21 predefined chart periods, from one minute through to one month, alongside an identifier that simply means whichever period the chart is currently set to. Each of those periods slices the identical stream of quotes into a different set of bars.

The consequence is arithmetic rather than opinion. A four-hour bar is built from the same ticks as the six one-hour bars beneath it, but its high is the highest price across the whole four hours. A high that ends one hourly bar and is exceeded in the next disappears entirely from the four-hour record.

So a containment relationship that holds between two hourly bars can be absent between the four-hour bars covering the same hours, and a containment that holds on the four-hour chart can have no counterpart below it.

This is why a question like how reliable the pattern is cannot be answered as asked. The pattern is not one object with one behaviour. It is a comparison performed on whichever of the 21 partitions the chart happens to be showing, and the answer changes with the partition. A claim about the pattern that does not name a period has left out the variable that generated it.

The same reasoning applies to reading several periods at once, where a signal on one chart is checked against the chart above it. That practice has its own logic and its own failure modes, set out under why your daily chart differs from another broker’s, and the point here is narrower: the disagreement between periods is not noise to be filtered out. It is what the definition produces.

Why Two Brokers Can Disagree About Whether Today Printed One

Above the four-hour period the boundary question sharpens, because a daily bar has to start somewhere and the platform does not use your clock to decide where. In the MetaTrader reference, the timestamp a bar carries originates with the broker trading server, and whatever the local machine happens to be set to has no bearing on it. Price history reaches the terminal as blocks of minute bars sent by that same server.

A daily bar is therefore the aggregate of the server’s day. Two brokers whose servers sit on different offsets cut the week at different hours, so their daily bars are built from different spans of the same market. One broker’s Tuesday can contain a high that falls into another broker’s Wednesday.

Once the boundary moves, containment can move with it. A daily bar that sits inside its predecessor on one platform may extend past it on another, because the extremes were gathered over a differently placed 24 hours. Neither chart is wrong and neither is showing a data error.

The full mechanics of the clock your broker builds its bars on, including how to read your own offset off the platform, belong to that page. What matters here is the consequence: a daily pattern is a property of a specific server’s calendar, and two screenshots of the same instrument can honestly disagree.

Where the Equal-High Convention Changes What a Scanner Returns

The containment test has to be written down before a tool can run it, and there is a choice in the writing that changes the output. If the test requires the later high to be strictly below the earlier high and the later low strictly above the earlier low, a bar that matches the previous high exactly is excluded. If the test allows equality, that same bar is included.

Both versions are defensible readings of the word inside. They are not the same filter. On instruments and periods where a high or a low repeats exactly, the two produce different lists from identical data, and the difference is largest wherever prices cluster on round numbers or where a quiet session revisits the same level.

None of this is exotic. It is the ordinary consequence of turning a visual idea into a comparison, and it explains why two scanners pointed at one chart return different counts without either being broken.

The Chart Types Where an Inside Bar Cannot Form

Everything above assumes bars delimited by time. Some chart types are not, and the pattern behaves differently on each.

On a range chart every completed bar spans the same configured amount, because reaching that amount is what closes the bar. Two completed bars therefore have equal ranges, and a bar of equal range cannot be strictly contained by the one before it. The same holds for Renko, where each brick is a fixed size by construction. On these charts the pattern is not rare. It is unavailable.

Charts built on a fixed number of ticks or a fixed traded volume behave differently again. There the count is fixed but the price range covered is not, so one bar can span less than the bar before it and containment can occur. What has gone is the period: the bar no longer represents an interval of time, so a statement about a daily or four-hour inside bar has no equivalent.

How each of these constructions decides to print a bar is set out under Renko and range construction.

The practical point is that a pattern presented as universal is defined on one family of charts. Moving it to another either removes it by construction or changes what it refers to.

What Containment Does Not Tell You About Direction

A frequent question is whether the pattern is bullish or bearish. The comparison itself answers neither, and it is worth being precise about why.

Containment states one thing: the later period traded within the extremes of the earlier one. It says the range narrowed. It carries no information about which boundary is approached next, because the same relationship holds whether the smaller bar closed near its high, near its low or in the middle. Direction, where it is inferred at all, comes from something outside the comparison, such as where the bars sit relative to a level or a trend.

The colour of the contained bar is sometimes offered as the tie-breaker. That is a fifth price added to a test defined on two, and nothing in the containment relationship makes the close of the smaller bar predictive of what follows.

Some readers combine the pattern with a long-wicked bar instead, a pairing covered under the pin bar and inside bar combination. That is a different setup with its own conditions, not a direction supplied by the inside bar.

The Performance Figures Nobody Sources

Pages on this pattern carry numbers. Percentages for how often the breakout continues, target win rates, thresholds for how far into the mother bar the child should sit, entry distances measured in pips.

Every such figure read while preparing this page was stated without a source. Not one cited a study, a published backtest, a sample size, a period, an instrument or a date range. In several cases the figures appear alongside an entry method described as having the highest win rate, with nothing given to compare against.

A number of that kind cannot be repeated responsibly, because there is nothing behind it to check. The pattern also has no single definition to measure, as the sections above set out: any honest measurement would have to state its period, its containment convention and its data source before its result meant anything. No performance figure appears above, and that is a decision rather than an oversight.

Which Reading Applies to Your Chart

Which of the sections above governs your screen depends on three settings, and each has a different answer.

If the chart is time-based and intraday, the pattern is defined and the period is the variable that matters most: the same instrument will show different inside bars at each of the periods available, and a claim carried over from another period does not transfer.

If the chart is daily or higher, the server calendar joins it, and a daily pattern belongs to one broker’s day rather than to the market.

If the chart is Renko or range-based, the pattern cannot form, and a strategy described in terms of it has no object on that screen.

In all three cases the containment test says the range narrowed over one period relative to the one before. Whatever is added to that, about direction, probability or where a stop belongs, comes from somewhere else and needs its own evidence.