WhitmanTrading

Bull Flag: The Break Is a 20-Bar High

A bull flag is a sharp advance followed by a shallow, orderly pullback, traded on a break back above the pullback's upper boundary. That break is almost always a new multi-week high, which means the pattern's outcome is the outcome of buying breakouts, measurable directly.

How it works

A candlestick chart of the site's shared price history. The headline on the chart reads: A sharp advance, then a shallow drift against it.
A sharp advance, then a shallow drift against it. Illustrative chart - not real market data.

Two parts. A sharp advance — the pole — and then a shallow, orderly drift downward or sideways — the flag. The drift usually slopes slightly against the advance and is contained between two roughly parallel lines.

A strongly rising stretch of the long price series. The headline on the chart reads: The pole is the move; the flag is the pause.
The pole is the move; the flag is the pause. Illustrative chart - not real market data.

The story is a trend pausing rather than ending. Buyers took profits, the advance stalled without reversing, and when the flag breaks the original move resumes.

A gently rising stretch of the long price series. The headline on the chart reads: A deep pullback is not a flag, it is a reversal forming.
A deep pullback is not a flag, it is a reversal forming. Illustrative chart - not real market data.

Shallowness is the defining condition and it is the one most often ignored. A pullback that gives back most of the pole is not a flag — it is a retracement deep enough to be a reversal, and calling it a flag because you want the trade is where the pattern stops constraining anything.

What the entry actually is

A calmly advancing stretch of the long price series. The headline on the chart reads: The entry is a break of the flag, which is a twenty-bar high.
The entry is a break of the flag, which is a twenty-bar high. Illustrative chart - not real market data.

Strip the picture away and the entry is a purchase at a new multi-week high. The flag’s upper boundary sits just under the pole’s peak, so breaking it means exceeding every price of the last several weeks. The pattern is a breakout with a specific preceding shape.

Which is good news, because breakouts can be measured directly.

A flat but volatile stretch of the long price series. The headline on the chart reads: Eighty-five percent of twenty-bar highs came back within ten bars.
Eighty-five percent of twenty-bar highs came back within ten bars. Illustrative chart - not real market data.

On this site’s shared 576-bar history there are 39 closes above a 20-bar high. Ten bars later, 38% had a higher close — and 85% had at some point closed back below the level they broke.

A declining stretch of the long price series. The headline on the chart reads: Against a fifty-four percent base rate for any bar.
Against a fifty-four percent base rate for any bar. Illustrative chart - not real market data.

The base rate for any bar at all, over the same ten-bar horizon, is 54%. So breaking out did substantially worse than doing nothing in particular on this data.

Two things to say about that immediately. The series is synthetic — it has no trends driven by anything, so a breakout cannot work on it. And that is exactly why the number is useful: it is the null model. 85% failure and 38% follow-through is what a breakout looks like when there is no mechanism behind it, and it is the figure a real instrument’s breakouts have to beat before a flag method is doing anything at all.

Most published flag material offers no comparison of any kind, which makes it impossible to tell whether the shape adds anything to the underlying breakout.

In practice: the parts worth keeping

A candlestick chart with a volume histogram beneath it, with the volume histogram emphasised. The headline on the chart reads: Volume should fall through the flag and return on the break.
Volume should fall through the flag and return on the break. Illustrative chart - not real market data.

The volume signature is the only genuinely independent input. Participation should thin through the flag — fewer people trading a pause — and expand on the break. A break on falling volume is the one advance warning the pattern offers, and it is available before the entry rather than after.

A declining stretch of the long price series, with the entry price and the level at which a stop would trigger drawn as horizontal lines. The headline on the chart reads: The stop goes under the flag, not under the pole.
The stop goes under the flag, not under the pole. Illustrative chart - not real market data.

The stop goes under the flag, which is close to the entry. This is the pattern’s real advantage over most formations: the invalidation level is nearby, so the position can be sized meaningfully without risking much. That is a structural benefit and it does not depend on the pattern predicting anything.

A flat, quiet stretch of the long price series. The headline on the chart reads: Measured from the break, the target is the pole length.
Measured from the break, the target is the pole length. Illustrative chart - not real market data.

The measured move projects the pole’s length up from the break. Arithmetic, not a forecast — useful for deciding whether the trade is worth taking given the stop distance, and nothing more.

A long-horizon candlestick view of the same price series. The headline on the chart reads: On a higher timeframe the flag is inside one candle.
On a higher timeframe the flag is inside one candle. Illustrative chart - not real market data.

Aggregate the chart and the flag vanishes into the pole. The formation exists at the resolution you chose to look at, which is worth knowing before treating it as a property of the market.

A candlestick series containing several opening gaps, with the largest opening gap marked. The headline on the chart reads: A gap out of the flag is the entry you never get.
A gap out of the flag is the entry you never get. Illustrative chart - not real market data.

A gap out of the flag is common on stocks and removes the entry. The pattern resolves in your direction at a price you could not buy, and chasing it means entering with the stop much further away than the plan assumed.

A candlestick chart of the site's shared price history, annotated with the round-trip cost. The headline on the chart reads: And each break traded costs a share of a bar.
And each break traded costs a share of a bar. Illustrative chart - not real market data.

Each break traded costs 2% of a typical bar’s range in round-trip costs on this history — and with a failure rate anywhere near the measured one, most of those payments buy a stop-out.

A 72-bar candlestick section of the shared price history. The headline on the chart reads: The shape is drawn; the orders are not.
The shape is drawn; the orders are not. Illustrative chart - not real market data.

And the parallel lines around the flag are yours. The order book has no boundaries in it. If the break produces movement, it is because a lot of people were watching the same obvious high.

What a bull flag is not

It is not a guarantee of continuation. It is a shape that sometimes precedes continuation and often precedes a failed breakout.

It is not a pennant. A flag drifts within parallel lines; a pennant converges. The distinction matters mainly because the pennant’s boundaries move, so its break price does too.

It is not defined by any number. How sharp must the pole be, how shallow the flag, how many bars either part may take — none of it is specified, which is what makes the pattern findable almost anywhere.

And it is not independent of the breakout it contains. Any evaluation of flags that does not compare against plain breakouts is measuring breakouts and calling it flags.

When it fails

A sideways, range-bound candlestick series. The headline on the chart reads: In a range every pause looks like a flag.
In a range every pause looks like a flag. Illustrative chart - not real market data.

The dominant failure is the flag that keeps drifting. What was a shallow pause becomes a deep retracement and then a reversal, and there is no bar at which the pattern officially fails — it simply stops being called a flag.

The second failure is the false break. Price clears the flag, triggers the entry, and closes back inside within a bar or two. At 85% of 20-bar breakouts returning below the level on this data, this is the ordinary case rather than the exception.

A third is drawing the flag after the break. With hindsight, the pullback that preceded any successful breakout can be bounded by two lines and called a flag. That is pattern-fitting, and it is why flag examples always look immaculate.

A fourth is the missing volume check. A break on thinning participation is the single most useful disqualifier available, and it is the one skipped most often because it rules out trades that look good.

And a fifth is treating a fast chart flag as the same object. A flag on a five-minute chart is a twenty-minute pause. It shares a name and none of the context with a flag on a daily chart.

The original data

39 closes above a 20-bar high on this site’s shared 576-bar history: 38% were followed by a higher close ten bars later, against a 54% base rate for any bar, and 85% closed back below the broken level within ten bars. The 10-bar and 55-bar lookbacks are recorded alongside them in research/series-measurements.json, produced by site/measure_series.py.

A 72-bar window of the shared price history, cut short at the decision bar. The headline on the chart reads: The flag is three bars old and tight. Buy the break?
The flag is three bars old and tight. Buy the break? Illustrative chart - not real market data.

Run those three numbers on your own instrument and you will know something most flag traders do not. Count closes above the 20-bar high, count how many were higher ten bars later, count how many closed back below the level, and compare all of it against the base rate for an ordinary bar. If your instrument’s breakouts do not beat the base rate, no amount of flag-shaped context in front of them will change that — and if they do beat it, you have a real number to compare the flag version against, which is the only way to find out whether the shape is adding anything.

Breakout is what the flag entry actually is, measured directly. Bear flag is the mirror formation. And pullback covers the pause the flag is a tidy special case of.

What I actually do

Flags were the first pattern that made money for me and the first one I stopped trusting for the reason I originally trusted it - it looks like a trend taking a breath. What I eventually had to accept is that I was buying a breakout with a nice picture attached, and the picture was not changing the arithmetic of the breakout.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.