Pennant Pattern: The Converging-Triangle Cousin of the Bull Flag

A stock runs 18.40 to 22.00 in twenty minutes on volume nobody has seen all week. Then it stops. For the next half hour the highs step down a little and the lows step up a little, the candles get smaller, the volume bars shrink to nothing, and the whole thing coils into a shape about the size of your thumbnail. That coil is a pennant. It is the market taking a breath after a sprint, and the reason traders care about it is that the breath is usually short and the sprint usually resumes. The pattern is simple to draw and easy to draw badly, which is most of the problem with it.

Quick Answer

Everything else is detail about how to tell a real one from the dozens of near-misses that show up on any volatile session. The rest of this covers the structure, the bullish and bearish construction side by side, the volume signature that does most of the confirming, the short version of how a pennant differs from a flag, the entry, stop and measured-move target, the ways it fails, and what an automated read can and cannot see in a chart image.

What Is a Pennant Pattern?

A pennant is a continuation pattern, which is the single most important thing to fix before anything else. It does not call a top or a bottom and it is not a reversal signal. It marks a pause inside a move that is already running, and the base case is that the move resumes in the same direction. StockCharts files flags and pennants together as short-term continuation patterns that mark a small consolidation before the previous move resumes, and that sentence is the whole thesis of the setup.

The construction has three parts, and the Corporate Finance Institute describes them in its entry on the pennant pattern as a first flagpole, the pennant itself, and a second flagpole. The first flagpole is the sharp initial move, and it should have broken something on the way: a trendline, a prior high, a level people were watching. A move that just drifts up on nothing is not a pole. The pennant is the consolidation between two converging trendlines, one drawn across the descending highs and one across the ascending lows, forming a small symmetrical triangle that narrows toward a point. The second flagpole is the breakout that resumes the original direction.

The mechanics underneath are ordinary. A fast move leaves a lot of traders holding a quick profit and a lot of others watching from the sidelines. The profit-takers sell into the pause, which caps the highs. The latecomers buy the dip, which lifts the lows. Neither side is committed enough to break the other, so the range narrows from both sides at once until somebody gives. If the original imbalance is still there, the side that created the pole wins the argument and the move continues. Drawing those two boundaries cleanly is the skill that decides everything downstream, and the same rules from drawing a trendline that actually holds apply here: connect the swing points, not the wicks you wish were there.

Bullish vs Bearish Pennant

The two versions are mirror images and the mirror is exact, which is rare enough among chart patterns to be worth saying out loud. A bullish pennant needs a sharp up-move inside an existing uptrend. A bearish pennant needs a sharp down-move inside an existing downtrend. Same coil, same volume behaviour, same measured move, opposite direction on all of it.

The context requirement does real work and people skip it. A sharp up-move that happens during a downtrend, followed by a converging pause, is a counter-trend bounce coiling against the dominant direction. You can still trade it, but you are trading it against the trend and the base case no longer favours you. This is the same trend-alignment filter that shows up in a working momentum trading strategy, where the pattern is the trigger and the trend is the permission slip.

Read the last row and you can decode most of the naming confusion around this pattern. Bull pennant, bullish pennant and rising pennant all point at the same chart. Bear pennant, bearish pennant and falling pennant point at the inverse. The direction of the pole is the only thing that settles it.

What Does Volume Do Inside a Pennant?

Volume is where a pennant earns the benefit of the doubt, and it moves in three stages. Heavy on the pole, drying up through the coil, expanding again on the break. The Corporate Finance Institute puts it plainly: the first flagpole must be met with large volume, followed by weakening volume in the formation of the pennant, ending with large volume during the second flagpole. StockCharts makes the same point about the break specifically, noting that an expansion of volume on the resistance or support break lends credence to the validity of the formation.

The reason this sequence matters more than the shape is that it tells you what the pause actually is. Contracting volume through the coil says the pullback is people taking profit rather than a new wave of sellers arriving. Heavy volume through the coil says the opposite, and a coil that drifts sideways on volume as big as the pole is distribution wearing a pennant costume. That distinction, and the way it reads off the volume bars under the price, is the same one covered in reading volume properly on an intraday chart, and it applies identically to flags, pennants and any other pause inside a trend.

This is also the part of the pattern that an automated read handles well, because it is entirely visual. The pole is a run of large-bodied candles over tall volume bars. The coil is smaller bodies over short volume bars. The break is one decisive candle over a tall bar again. SnapPChart already describes exactly this relationship when it grades continuation setups, in language like volume increasing on the pole and decreasing through the consolidation, which is the textbook signature. Nothing about that requires live data or order flow. It is a shape and a histogram, both of which are in the picture.

Pennant vs Flag vs Wedge

Short version, because this ground is already covered elsewhere on the site. A flag consolidates inside two roughly parallel trendlines, so the pause is a channel of more or less constant width. A pennant consolidates inside two converging trendlines, so the pause narrows as it goes. Both sit on a pole and both are continuation patterns. The longer treatment of the difference between a bull flag and a pennant lives in the bull flag guide and there is no point restating it here.

The comparison that trips people up more often is the pennant against the wedge, because both narrow. The test is the slope of the two boundaries. A pennant has them tilting toward each other from opposite sides, so the shape sits roughly level. A wedge has both boundaries pointing the same way, which is why a rising wedge leans bearish and a falling wedge leans bullish regardless of what came before. Against a symmetrical triangle the difference is scale rather than shape: a pennant is a small, fast version that hangs off a pole, while a full symmetrical triangle takes many more bars and can appear without one.

The practical order is boundaries first, label second. Draw the two lines, read their slopes, then decide what you are looking at. Naming the pattern before you draw it is how a rectangle ends up in somebody's journal as a pennant. The wider reference of chart patterns and what each one implies sorts the rest of the family the same way.

Where Do the Entry, Stop and Target Go?

Illustrative numbers, built so the arithmetic can be checked by hand rather than lifted from a real session. A stock runs from 18.40 to 22.00 on the heaviest volume of the day, which makes the pole 3.60 tall. It then coils for about half an hour between a line descending from 22.00 and a line ascending from 21.10, and by the end of the coil those two boundaries are roughly 25 cents apart around 21.75. Volume through the coil is a fraction of what built the pole.

• Entry on a close through the boundary
  A candle closing above the upper converging line, around 21.85 in this example, with a volume bar visibly taller than the ones inside the coil. Closing through matters more than touching through, because a wick past the line and back inside is the most common way this pattern fakes you out.
• Or entry on the retest, with a rejection candle
  Price breaks, pulls back to the broken upper line, and puts in a candle that refuses to go back inside. The fill is worse than the breakout entry but the stop is tighter and the invalidation is cleaner. The trade-off between taking the break and waiting for the retest is the same one covered in the break-and-retest comparison.
• Stop just beyond the opposite boundary
  For a bullish pennant, below the lower converging trendline or the lowest low of the coil, around 21.35 here. For a bearish pennant, above the upper trendline or the coil high. The logic is that price inside the coil again means the breakout failed, so there is nothing left to be right about.
• Target by measured move
  Project the height of the pole from the breakout price. StockCharts states the method directly: the flagpole length can be applied to the resistance or support break of the flag or pennant to estimate the advance or decline. Here that is 21.85 plus 3.60, or 25.45.

Run the numbers on that and something looks too good. Risk from 21.85 down to 21.35 is 50 cents. Reward to the measured target is 3.60. That is better than 7 to 1, which is not a ratio real trading hands out often. The honest reading is that a tight coil produces a flattering number on paper precisely because the stop is close, and a close stop is also the easiest stop in the world to get clipped by noise. Treat the measured move as an estimate rather than a price you are owed, size the position off the stop distance the way the reward-to-risk calculation assumes you will, and plan for the realistic version where taking partial profit before the full projection is how most of these actually get closed. Placing the stop at the level that genuinely invalidates the structure, rather than at a round number or a comfortable dollar amount, is covered properly in the guide to stop placement.

The bearish version runs the same arithmetic upside down. A stock drops from 62.00 to 56.80, making the pole 5.20. It coils for thirty-five minutes, then closes below the lower converging line at 57.60. The stop goes above the upper boundary at 58.25, so the risk is 65 cents. The measured target is 57.60 minus 5.20, or 52.40. Same structure, same three-stage volume, same caveat about the ratio flattering itself.

Why Do Pennants Fail?

Pennants form easily, which is their main weakness as a screen. Any sharp move followed by a brief pause can be drawn as converging lines if you are willing to be generous with the trendlines, so the raw population of things that look like pennants on a volatile day is large and most of them are noise. The filters below are what separates the population from the setups.

• The coil never actually contracted
  Two lines you can draw are not the same as a range that narrowed. If the last few candles inside the coil are as tall as the first few, the equilibrium the pattern is supposed to represent never formed and the break is a coin flip.
• The pause outlasted the move
  A pennant is short by definition. When the consolidation runs many times longer than the pole, the original imbalance has had time to be absorbed and what you are left with is a range that happens to sit above a big candle.
• Volume stayed heavy through the coil
  Heavy volume inside the consolidation says real sellers are present in a bullish pennant, or real buyers in a bearish one. That is the opposite of the profit-taking pause the pattern assumes.
• News landed mid-pattern
  A headline, an earnings release or a halt resets the reason the pole existed. The shape on screen survives the news, but the logic behind it does not, and a pattern that is right for the wrong reason is just luck.
• The pole was already extended
  A pennant that forms after the fourth leg of a run is a pause inside a move that has already paid out. Entering there is the chasing problem, not a pattern problem.

False breakouts get more common as volatility rises, because a wider average candle makes it trivial for price to poke through a boundary and reverse without anything having changed. That is the argument for waiting on the close rather than the touch, and for considering the retest entry instead of the breakout entry when the tape is jumpy. It is also the argument against entering the fourth pennant of the day on the same ticker, which is a real habit and is closer to chasing an extended chart than to trading a pattern.

How AI Reads a Pennant Off a Screenshot

The useful thing about a pennant, from a tooling point of view, is that every input is in the image. The pole is a visible run of candles. The boundaries are lines through visible swing points. The contraction is a measurable change in candle size. The volume sequence is a histogram sitting under the price. None of it needs a data feed, which is why a static chart screenshot is enough to grade the setup.

What an AI read is doing on a chart like this is checking the same list a trader would, in a fixed order, without getting talked into anything by the fact that it has been staring at the ticker for an hour. Did the pole break a level. Did the coil genuinely narrow. Did volume fall through the coil and return on the break. Is the stop close enough that the reward justifies it. The output is a grade plus an entry, an alternative entry, the stop with the reasoning for that level, and targets. A C-grade pennant is a useful answer, and the trades you skip because of it compound quietly in a way that no single winner does. The broader picture of what a single-image read can and cannot carry sits in the guide to how AI reads a chart.

If you want to see it on this specific pattern rather than in the abstract, the two strategy pages for it are the direct route: grade a bull pennant setup for the long side and grade a bear pennant setup for the short. Both walk the same pole, coil, volume and breakout checks against your own chart rather than a textbook one.

The limits are worth stating as plainly as the capabilities. A screenshot contains no order book, no time and sales, no news and no live quote, so nothing in the read predicts which way the coil breaks before it breaks. It grades the structure that exists. That is a smaller claim than pattern recognition marketing usually makes, and it is the one that survives contact with a real session. Stacking the pennant with other evidence on the chart, the way confluence between independent signals works, does more for the odds than any single pattern name ever will.