Triangle pattern breakout: what ascending, descending and symmetrical triangles do and don't tell you

Trade a descending triangle as a guaranteed break lower and, on Thomas Bulkowski's numbers, you would be wrong a little more often than right. Of the 1,300-plus descending triangles he logged in US stocks, 53% broke out upward.

The chart does not tell you which way a triangle will go. In his bull-market samples the upward share was 63% for ascending triangles, 60% for symmetrical ones and 53% for descending ones. That is a lean, and not a strong one. The samples are stocks; crypto has no public dataset of comparable size, so we do not give a crypto success rate here. What you can do is write down, before the break, what would confirm a move up, what would confirm a move down and what would cancel each, then wait for a candle to close outside a line.

Two people looking at the same stretch of chart often draw two different triangles, and most of that comes down to the drawing. So the drawing comes first.

Types of triangle pattern: two lines, and how many touches

All three are squeezed between two trendlines. What separates them is which line, if any, runs flat.

• Ascending triangle: flat top, rising bottom. Price keeps stalling at the same level while each dip stops higher than the last.
• Descending triangle: flat bottom, falling top. Price keeps finding support at one level while each bounce tops out lower.
• Symmetrical triangle: both lines slope, the top one down and the bottom one up, narrowing to a point on the right.

Bulkowski's pages on the ascending triangle and the descending triangle (checked 2026-10-08) put it plainly: in the first, "the top trendline is horizontal and the bottom one slopes upward"; in the second, "the bottom one horizontal and the top sloping downward". For how to read swing highs and lows, see higher highs and higher lows.

Two lines on a chart are not yet a triangle. His identification rules on the symmetrical triangle page include two that people skip: "Price must touch one trendline at least three times and the other trendline at least twice", and price must cross from side to side, "filling the triangle with price movement, not white space." Lines joined after two touches each may be framing nothing more than ordinary chop.

The parts of the drawing you decide yourself

Hand the same chart to two people and you can get a wide triangle and a tight one. The difference usually sits in four places.

1. Wicks or bodies. Draw through the wicks and the lines sit further apart, so price has to travel further before it counts as outside. Draw through the bodies and the lines are tighter, and a break comes sooner.
2. Which swing you start from. Move the starting point by one candle and the slope changes, and so does the point where the lines meet, the apex. That point matters later.
3. Which timeframe. A tidy symmetrical triangle on the 1-hour chart might be two or three candles on the daily.
4. What counts as a touch. A near miss, or a wick that pokes through and closes back inside: touch or break? Bulkowski's rules leave that to you.

Our sketches draw through the wicks. The step we are stricter about is the start: the first swing high and low that stands out on the chart, with no reaching one candle further left to make up a third touch. Shift the start left to get the count and the triangle has been drawn, not formed. The flat top of an ascending triangle is a resistance line by another name; how to find support and resistance covers how to place one.

Ascending, descending and symmetrical triangle breakout odds

The usual line is that ascending triangles are bullish and descending ones bearish. Capital.com's guide to trading triangles words it more carefully: many traders read the ascending version as "a pattern with a bullish bias, although the direction is not confirmed until price breaks out", and the descending one the same way with a bearish bias.

Here is what Bulkowski counted, all in US stocks during bull markets:

• Ascending, more than 1,400 patterns: 63% broke upward, 37% downward.
• Symmetrical, over 3,000: 60% upward, 40% downward.
• Descending, more than 1,300: 53% upward, 47% downward.

The ascending triangle's upward lean is real in that data, and still about a third of them went the other way.

The descending row is the one that cuts against the textbook. A shape usually labelled bearish broke upward slightly more often than not, and his page says so directly: "Breakout Can be in any direction but is upward 53% of the time."

Three conditions travel with those numbers. He says they are based on "perfect trades", patterns identified by his own rules, not every triangle you might spot. The pages carry the heading "Bull Market Results", so bear markets would need their own count. And they are stocks. Crypto trades around the clock and moves harder, so how much of the lean carries over is a separate question.

How close to the apex does the breakout usually come?

Extend both lines to the right and they meet at the apex. A lot of people picture price squeezing all the way into that point before it moves. The samples say otherwise.

Bulkowski measured how far along the distance from the start of the triangle to the apex the breakout happened. For symmetrical triangles the average was 74% of the way; for ascending triangles, 64%. Both figures cover breakouts in either direction. Most breaks, in other words, arrive with some room still left before the point. Those percentages also depend on where you started drawing: shift the first touch and the distance you are measuring changes.

If price does grind right into the tip, the lines are almost on top of each other and a tiny move counts as "outside". Calling a break there is a lot less clear than calling one while the triangle is still wide.

His descending triangle page adds a research note: "Research shows that after the breakout, the market turns when it reaches the triangle apex." We cannot test that on crypto, so we treat it as a reminder rather than a rule: keep the apex marked on the chart even after the break.

Which close confirms the break in a market that never closes?

The same descending triangle page states the trigger: "The pattern confirms as a valid one when price closes outside one of the trendlines." A close, not an intraday poke. A wick that spikes through and closes back inside has not broken anything yet.

Stock exchanges close once a day. Crypto does not, so "close" here means the moment a candle on your chosen timeframe finishes: every 15 minutes on the 15-minute chart, every four hours on the 4-hour.

The daily candle needs one more question: which time zone was it cut in? The Binance developer documentation (checked 2026-10-08) runs its standard kline streams on UTC+0 and also describes a variant that cuts the day at UTC+8: "the 1d klines will open at the beginning of the UTC+8 day, and close at the end of the UTC+8 day." Cut the same stretch of trading at UTC+8 midnight and at UTC midnight and the daily closes land eight hours apart, which is enough for one version to finish outside the line and the other inside it.

When we log a break, three things go next to it: the timeframe, the time that candle closed, and which line it closed beyond. If one is missing, we don't call it a break yet. Choosing the timeframe has its own page: which candlestick timeframe to use.

Does the breakout candle need volume?

Inside the triangle first. The symmetrical triangle page reports that volume trends downward 84% to 86% of the time while the pattern forms. Volume drying up during the squeeze is the normal case in his data, not a special tell.

Then the breakout itself: "Patterns with heavy breakout volume (above the 30-day average) do better." Note the yardstick, the 30-day average, and the wording, "do better". Nothing there says a quiet breakout does not count.

You will find rules elsewhere that name an exact volume multiple. We could not find a sample behind them, so we leave them out. Capital.com's caution fits here: "Volume can help with confirmation, but false breakouts are still common." Volume helps; it does not rule out a fake. For reading the two together, see reading volume alongside price, and for breaks that come with volume and still fail, how to spot a false breakout.

Triangle pattern breakout and retest, and what a bust looks like

Bulkowski's glossary has a word for each direction. A throwback is when "Price breaks out upward, climbs, and then returns to the breakout price within 30 days." A pullback is the downward mirror. Most traders just call either one a retest.

His throwback and pullback rates:

• Ascending: 64% throwbacks after upward breaks, 63% pullbacks after downward ones.
• Symmetrical: 62% and 65%.
• Descending: 60% and 58%.

All six sit around 60%. Coming back to the breakout level is what usually happens; it is not a bonus setup. What price does after it gets back there, these numbers do not cover.

A break that runs out of steam and turns back is what he calls a failure, and he defines two kinds:

• Break-even failure: price moves less than 5% after the breakout before turning. For symmetrical triangles the rate is 25% for upward breaks and 37% for downward ones; for ascending triangles, 17% and 38%.
• Busted pattern: "Price that moves no more than 10% after the breakout before reversing and zipping to the other side of the chart pattern". Symmetrical triangles also "have a tendency to double bust -- the final breakout direction is the same as the original one."

So we write two sets of conditions before the break. The upside set: which close above the top line confirms it, and where a drop back would cancel it. The downside set mirrors that. Whichever way price moves first, open that set, and keep the other one to hand.

Triangle breakout target: measuring the height

Capital.com gives the usual method: "A common target projects the triangle's height from the breakout point." Take the height of the triangle and lay it off from where price broke out, in the direction of the break.

That page measures the height at the triangle's widest point and projects it from the breakout. Bulkowski's descending triangle page measures differently: from the highest peak down to the flat line, then scales it by the share of patterns that reached the target. Our sketches use the widest point, near the start, between the two lines. Measure another way and the number moves.

An example with made-up numbers, purely to show the arithmetic: an ascending triangle with its flat top at 100 and the low at the widest point at 90, so a height of 10. A close above 100 is the upward break; add 10 and you get 110. Had it broken down through the rising line, you would subtract 10 from the level where it broke.

That 110 is arithmetic. It moves the pattern's height outside the line and says nothing about whether price will go there or stop when it does. We only use it to see how much of that height a move has already covered, never as a price to wait for.