8 min read The Chartix Editorial Team
A monitor on a dark desk at night, two converging lines of cyan light narrowing a row of abstract candle shapes, a city skyline beyond the window

Most chart patterns arrive with a direction attached. A triple top is a bearish reading, a rounding bottom a bullish one, and the argument is about whether the shape is really there. The symmetrical triangle pattern is different. Its shape says that the range is narrowing, and nothing else. The direction is the part you have to wait for, and the most useful thing anyone has measured about it is how often the wait ends in the wrong direction.

A symmetrical triangle is a chart pattern in which price makes lower highs and higher lows, so a falling upper trendline and a rising lower trendline converge toward an apex. It describes a contracting range, not a direction: a close outside one line confirms a read, and a close back through the far side of the pattern invalidates it.

That second sentence is the whole discipline. The rest of this post is the checkable version of it: what qualifies a triangle, what the “continuation pattern” label is and is not based on, and why this shape produces more busted breakouts than most.

What makes a symmetrical triangle pattern, in checkable terms?

Bulkowski, whose public pattern statistics are unusually detailed, gives identification guidelines that are stricter than the usual “draw two converging lines”:

  1. Two trendlines that converge. The upper line slopes down, connecting lower highs. The lower line slopes up, connecting higher lows. If either line is flat, it is a different pattern (an ascending or descending triangle) with a different claim.
  2. Enough touches. Price must touch one trendline at least three times and the other at least twice, forming distinct peaks and valleys. Two touches on each side is a drawing, not a pattern.
  3. Price fills the shape. Price has to cross the triangle from side to side, filling it with movement rather than leaving white space inside the lines. A triangle drawn across a few scattered highs and lows, with empty space in the middle, is a line exercise.
  4. Any prior trend. Unlike reversal patterns, Bulkowski lists the price trend leading into a symmetrical triangle as “can be any direction”. There is no context requirement about the trend before it, which is unusual and is the first hint that the pattern carries no direction of its own.

If the touches and lines are new territory, the support and resistance post covers the same problem for horizontal levels: a line is real when price respects it more than once, and imagined when it was drawn to fit two points after the fact.

An illustrative example, with prices made up for the purpose: a stock rallies to 50, pulls back to 44, rallies to 48.50, pulls back to 45.20, rallies to 47.40 and pulls back to 46. The highs (50, 48.50, 47.40) are falling and the lows (44, 45.20, 46) are rising. Three touches on the upper line, three on the lower, and price has crossed the shape five times. That is a symmetrical triangle. The same six numbers in a different order would not be.

What does the narrowing actually record?

Read structurally, the triangle is two trends at once. The falling highs are the lower highs of a short-term downtrend; the rising lows are the higher lows of a short-term uptrend. Each rally is sold a little earlier than the last, and each dip is bought a little earlier than the last. Neither side is winning; both are getting more willing to transact at a price closer to the middle.

Volume usually reflects that. Bulkowski reports that volume trends downward during the formation 84% to 86% of the time, which fits the reading: as the range narrows, fewer participants find a reason to act. The receding volume describes the pattern. It does not tell you which way it resolves.

Is a symmetrical triangle a continuation pattern?

This is the claim most ranking pages make, and it is where the measured data and the tradition part ways. The textbook description treats the triangle as a pause in a trend that resumes in the same direction once the coil resolves.

Bulkowski’s bull-market figures, based on what he describes as over 3,000 perfect trades (the page does not state the period covered), put the breakout direction at upward 60% of the time, with no requirement on the trend leading in. A 60/40 split, in figures he reports only for bull markets, is not much of a directional property. It is closer to what you would expect from a shape that records indecision and then a decision.

He also ranks the pattern’s performance after the breakout poorly: 36th of 39 chart patterns for upward breakouts and 34th of 36 for downward breakouts, with a break-even failure rate of 25% for upward breakouts and 37% for downward ones. His glossary defines break-even failure as the share of patterns that fail to move at least 5% in the breakout direction. These are one analyst’s figures on his own sample and they do not transfer to another market or timeframe, but they are specific, published, and they point the same way: treat the triangle as a question, not an answer.

What confirms it, and what invalidates it?

Because the shape carries no direction, there are two possible reads and each has its own confirmation and its own invalidation:

  • Needs: two converging lines with at least three touches on one and two on the other, and price filling the space between them.
  • Upward read, confirmed by: a close above the upper trendline. A wick through the line is not a close, and the confirmation candle post explains why a close is the standard rather than an intraday poke.
  • Upward read, invalidated by: a close back below the bottom of the triangle (the lowest low inside it).
  • Downward read, confirmed by: a close below the lower trendline.
  • Downward read, invalidated by: a close back above the top of the triangle (the highest high inside it).

Bulkowski uses the top and bottom of the whole triangle, rather than the sloping trendline, as the price that busts the pattern. The distinction matters. A close back inside the lines puts the read in doubt; a close through the opposite extreme ends it.

Busted triangles, and the double bust

A busted pattern is one that breaks out in one direction, fails, and closes beyond the opposite side. Bulkowski’s definition for symmetrical triangles is precise: price breaks out, moves no more than 10%, reverses, and closes beyond the side opposite the breakout. In his bull-market data, symmetrical triangles with downward breakouts bust 48% of the time (his busted-pattern page does not state the sample size or period for that figure).

His main page adds a second wrinkle that is worth knowing because it looks like a third pattern. Symmetrical triangles have, in his words, a tendency to double bust: the first breakout fails, the reverse breakout also fails, and the final move goes in the original direction after all. A hypothetical sequence: price breaks below the triangle, rallies through the top, then falls back through the bottom and keeps going. Keep the scale in proportion, though. On his busted-pattern page, 67% of busted symmetrical triangles single bust and 6% double bust, so the double bust is a recognisable shape rather than the usual outcome.

For a reader, the practical consequence is not a rule about what to do. It is a rule about what to conclude: a first close outside a symmetrical triangle is weaker evidence than a first close outside most other patterns, and the invalidation levels above exist precisely because the first break is so often not the last one. The false breakout problem is general, but this shape has more of it than most.

Where does the breakout usually happen?

Bulkowski reports that breakouts occur on average about 74% of the way from the start of the triangle to the apex, for both directions. That fits the mechanism: as the lines converge, the range eventually becomes too narrow to contain normal movement, and something gives. It also gives you a checkable warning. A triangle that has crept all the way to its apex without breaking has stopped being a coil and become a flat range, because the lines no longer bound anything.

He adds that the market tends to turn when price reaches the apex area, which is a separate observation from the breakout itself and one more reason not to treat the apex as a target.

The measured-move convention, described rather than forecast

The traditional projection takes the height of the triangle at its widest point, from the highest peak to the lowest valley, and projects it from the breakout price. Bulkowski multiplies that height by the percentage of his patterns that actually reached the full target, 58% for upward breakouts and 36% for downward breakouts in his bull-market sample, before adding or subtracting it. The adjustment is the honest part: in his own data the full textbook distance was reached a minority of the time on downward breaks. The projection is a convention for describing a pattern’s scale. It is not a price forecast.

Where it misleads

Triangles drawn on too few points. Two highs and two lows can always be joined into converging lines. The touches rule exists because without it every pause becomes a triangle.

The lines drawn to the wicks on one side and the bodies on the other. Pick one method and apply it to both lines. Mixing them lets you draw whatever triangle you were hoping for. The short version of anchoring a line honestly is that the method has to be chosen before you see where it lands.

Assuming the prior trend tells you the direction. It is the most repeated claim about this pattern and the least supported by measured data.

Ignoring the timeframe. A triangle on a 15-minute chart is a few hours of narrowing inside what may be a single daily candle. The read only applies to the timeframe it was drawn on, and how to read a stock chart covers why the same move can look like two different things on two timeframes.

What this post is not

It is not a way to trade triangles. It is a way to describe one accurately: two converging lines with enough touches to be real, a contracting range that says nothing about direction, two possible breakouts with two different invalidation levels, and a measured tendency for the first break to fail. If you can state all four about the triangle on your screen, you have read it. If you can only state the shape, you have drawn it.

For the candlestick patterns that do carry a direction, each with its context requirement and invalidation, the candlestick pattern library has all eighteen.

Sources

Every figure above was checked against these before publishing.