Triangle Patterns: Ascending, Descending and Symmetrical

The flat edge names the triangle, the sloping edge supplies the bias

Price has made four or five swings that keep shrinking, and the two lines you drew across the highs and lows are converging. Before you name it, look at which line is flat. A flat top at 200 with lows stepping up from 188 to 197 is an ascending triangle. A flat floor at 100 under highs falling from 112 to 103 is a descending triangle. Two sloping lines make a symmetrical triangle.

The bias comes from the sloping edge. Rising lows mean buyers step in earlier on each dip while sellers keep offering at 200, so the sellers are the side being tested. Falling highs say the reverse about the buyers at 100. When both edges slope, neither side has given ground, and StockCharts says the direction can only be determined after a valid breakout.

StockCharts calls the ascending triangle a bullish formation that usually continues an uptrend, and the Fidelity Learning Center’s chart-pattern webinar deck says prices can break either way, more commonly upward, and gives the reverse reading for the descending triangle. Count the touches. Two on each edge is the minimum; a third touch on the flat edge makes the level worth trading against.

A descending triangle at the end of an uptrend has a close cousin: a head and shoulders top with a flat neckline, which also puts a horizontal floor at 100 under lower highs. If the middle high stands well above the two beside it, the head and shoulders guide is the better fit, and its target uses the head-to-neckline distance.

Measure the height at the widest point and project it from the break

Three daily charts, one measuring rule. The ascending triangle has its flat top at 200 and lows at 188, 192, 195 and 197. Its height is the first swing: 200 minus 188, or 12. A daily close above 200 projects 200 plus 12, so the target is 212.

Same rule, other direction. The descending triangle has its flat floor at 100 and highs at 112, 108, 105 and 103. Height: 112 minus 100, or 12. A close below 100 projects 100 minus 12, so the target is 88. The symmetrical triangle has highs at 60, 58 and 56.5 and lows at 50, 52 and 53.5, a height of 10, so the target is 10 points beyond whichever edge breaks.

That is the whole triangle pattern target method. StockCharts measures the widest distance of the pattern and applies it to the breakout, and the Fidelity deck gives the same instruction for all three shapes: highest peak to lowest trough, then add for an upward break or subtract for a downward one. Measure the first swing, since the later ones are smaller by definition.

Log scale changes the arithmetic, because there the height is a ratio rather than a point count. In the descending example, 100 divided by 112 is 0.893, and 0.893 times 100 is 89.3, not 88. On a stock at 100 the gap is small; on a coin that doubled inside the pattern it is not. Pick one convention and check the log toggle at the bottom right of a TradingView chart.

Now put the plan on the ascending case. Entry 200.5 on the break. Stop 196.5, just under the last higher low at 197, which makes the risk 4 points. Target 212, which makes the reward 11.5 points. The ratio is 11.5 divided by 4, about 2.9R before commissions and slippage.

The stop sits under 197 rather than just under 200 because the flat line is where a throwback lands, and a stop inside that zone gets hit while the break is still working. Run your own prices through the risk-reward calculator with costs included; the ratio only goes down from there. This page shows how to measure a triangle. It does not tell you whether to take the trade.

Plan for the return trip. StockCharts notes that a broken 200 turns into support and that price sometimes revisits it before the move begins; Bulkowski’s ascending-triangle page puts that throwback at 64% of upward breaks. The breakout and retest guide covers the entry on that return. StockCharts calls volume expansion on the break preferred and not always necessary, so a screenshot without a volume pane leaves that check open.

A flag hangs from a pole, a wedge slopes one way, a triangle does neither

The look-alikes cause the most trouble on a 15-minute chart, where a 20-bar triangle and a 20-bar flag can sit side by side. Three checks settle it: whether the two edges converge or run parallel, how long the pause took, and what happened just before it. Flags need a pole. If you cannot find the near-vertical move the flag hangs from, it is not a flag.

StockCharts files a symmetrical triangle shorter than three weeks under pennants, and the flag or pennant target uses the pole rather than the pause: a 10-point pole projects 10 points from the flag break, whatever the flag’s height. The bull flag and bear flag guide works through that measurement. A triangle’s target uses only its own height, which is why the 12-point ascending example ignored the move before it.

A wedge is the usual mislabel. Both edges slope the same way, so a rising wedge climbs while it narrows, and the Fidelity deck asks for at least five touches, three on one line and two on the other, before it counts. Its resolution usually runs against the slope: the deck draws the rising wedge breaking down from a climax peak, the opposite of what rising lows imply in an ascending triangle.

The bias is a tendency, and a break that closes back inside has failed

Ascending triangles usually break up. That sentence hides a number. In Bulkowski’s bull-market sample of more than 1,400 ascending triangles, his “perfect trades” (page updated July 2025), the break went up 63% of the time, which leaves 37% that went down through the rising line, and the measured target was reached after 70% of the upward breaks, so three in ten stopped short of it.

The descending triangle is less settled. Bulkowski’s descending-triangle sample of more than 1,300 is closer to a coin flip, with 53% breaking up, while the Fidelity deck still calls downward the more common outcome. Two references, two answers. The break decides. The 100 floor in the example is a level to watch rather than a verdict to lean on.

Where the break happens matters too. StockCharts puts the ideal break on a symmetrical triangle between half and three-quarters of the way from the first swing to the apex, and Bulkowski’s ascending breaks came 64% of the way there on average. A break at the apex comes when the swings have shrunk to nothing, and a full 12-point projection is a lot to ask after 60 bars of going nowhere.

Then there is the break that does not hold. The Fidelity deck separates a false breakout, where price crosses the line and returns back through it almost at once, from a trap, where price then breaks the other way. A working rule: if a bar closes outside the boundary and a bar closes back inside within the next two, the break has failed. It is not a new signal in the other direction.

On the ascending example, a close at 201.5 on Monday followed by a close at 198 on Wednesday is an exit at 198, well before the stop at 196.5. A fresh trade needs a fresh close, this time under the rising line, with the same 12-point height projected downward from that close.

Check the edges on your own screenshot before you trust the label

Before you upload anything, check three things: two touches on each edge, the height at the first swing, and whether the last closed bar sits inside or outside the boundary. An unclosed daily bar poking above 200 at 2 pm is not a break.

Keep the whole triangle and the move before it in the frame, about 40 bars or more. Keep the price axis. Crop the order panel and the balance, because the whole screenshot is what gets uploaded.

TradeGPT reads that screenshot as an image, with no API and no broker connection. When it finds a triangle, the Patterns & Signals section names it, the outline is drawn on your chart, and Key Levels lists each level with a strength rating and a distance from price. If the read supports a trade, the Execution section carries an entry, stop loss, take profits and an invalidation condition; otherwise the plan is flat.

Keep the axis text crisp in the crop and its first quoted price lines up with your own. The sample report shows the full layout on a finished example, and you can analyze your own chart once the triangle is in a PNG or JPEG under 5 MB.

Questions traders ask about triangle patterns

Do ascending triangles always break upward?

No. In Bulkowski’s count the break went up 63% of the time, so roughly one in three broke down through its rising line instead. The flat top at 200 shows where sellers have been; it does not show that they will run out first. Wait for the close, and size the position for the 37%.

What is the difference between a triangle and a pennant?

A pennant is a small symmetrical triangle after a pole, and StockCharts files anything under three weeks under that name. The target changes with the label: a pennant projects the pole, a triangle projects its own height. On a 15-minute chart, a pole and a 12-bar coil is a pennant; 60 narrowing bars with no pole is a triangle.

How long should a triangle take to form?

StockCharts gives one to three months on a daily chart as typical, within a range from a few weeks to many months. Count bars on intraday charts: one to three months is roughly 20 to 65 daily bars, so on a 15-minute chart with 26 regular-session bars a day the same shape takes one to two and a half sessions.

What if the triangle breaks and then immediately reverses?

Treat it as a failed break and exit on the close back inside, without waiting for the original stop. Do not flip the trade on that bar. A reversal only becomes a signal when a bar closes beyond the opposite edge; on the symmetrical example that is a close under the rising line, with the same 10-point height projected downward.