What do triangles say about a narrowing argument?

A triangle is a range whose width shrinks. Highs come down, lows come up, or one side stays flat while the other advances. Whatever the variant, the underlying fact is the same: the gap between what buyers will pay and what sellers will accept is closing.

Three variants, three different stories

• Ascending triangle — flat highs, rising lows. Buyers pay progressively more on each dip, while sellers keep offering at one specific level. That flat ceiling often means a single large holder with a lot to sell at a chosen price, or a round number that many participants anchored to. Supply is concentrated; demand is climbing toward it.
• Descending triangle — the mirror. A flat floor absorbs everything, while each rally stops lower than the last. Demand is parked at one level; sellers accept less each time.
• Symmetrical triangle — both sides converging. Neither side is anchored to a level; the disagreement is simply narrowing. This is the least informative of the three, because it's also the shape that noise produces most easily.

A worked example

Rounded and illustrative: over eight weeks, a stock's rally highs come in at $100.30, $100.10 and $99.80 — effectively one flat ceiling near $100. Its pullback lows over the same span rise from $92 to $95 to $97. The range has compressed from about 8 points to about 3.

Read it as behaviour, not prophecy: someone keeps offering stock near $100, and buyers keep raising their bids into it. Each round trip is smaller. Traders watch for what happens when the compression ends — whether the flat level's supply is finally exhausted, or whether the rising bids give up. Which of those happens is not encoded in the triangle. The shape describes the pressure, not the outcome.

Where triangles mislead

Two specific traps:

Volatility compresses on its own. Volatility clusters and mean-reverts — quiet periods follow quiet periods, and every quiet period narrows the range. That produces triangle-shaped price action with no buyer-versus-seller drama behind it at all. A converging chart is weak evidence for a story about resting orders. Here is a stretch of exactly that, with nobody anchored anywhere:

The apex invites over-fitting. As the lines converge you can nudge either one slightly and change the apex date by weeks. People then treat the apex as a scheduled event. It isn't a date the market knows about — it's an artifact of two lines you drew.

A related honesty check: the shape of every trendline depends on whether you're on a linear or logarithmic price scale. The same eight weeks can be a clean triangle on one and a shapeless drift on the other. Pick your scale before you draw, not after.

In the data

The narrowing itself is measurable. On the year of QQQ below, the 14-day ATR in the lower pane falls as daily ranges compress, and the 20-day Bollinger bands squeeze towards their middle line at the same time.

Neither measure describes a triangle. Both are symmetric windows over the last few weeks of bars, with no idea of the two converging boundaries the pattern is drawn from, so a triangle and a directionless quiet stretch produce the same contraction in both.

Try it now

1. Classify the first schematic — which side is anchored? — and describe in one sentence who that suggests is doing what. Then measure the range width at the start and at the end and express the compression as a ratio.
2. Now do the same to the second schematic, which has no anchored side and no story at all. If your compression ratio comes out similar, you have just discovered why a narrowing chart is weak evidence — and the indicators agree with you. On the five years of QQQ below, the 14-day ATR in the lower pane falls whenever ranges compress, and the 20-day Bollinger band narrows alongside it; neither knows anything about a converging boundary, so neither can tell the two schematics apart.

3. Failure hunt on real prices: in the same five years, find a triangle that resolved against the direction its sloping side implied — an ascending one that broke down, or a descending one that broke up.