Symmetrical Triangle: Volatility With Lines On
A symmetrical triangle is a converging pair of trend lines with no directional expectation attached, traded on a break of either side. What the shape records is contracting volatility, which volatility indicators measure numerically without requiring anyone to choose which highs and lows to connect.
How it works
Highs coming down, lows coming up, both at roughly the same rate. Draw a line through each and they converge toward an apex.
Unlike almost every other named formation, it makes no claim about direction. The ascending triangle is labelled bullish, the bear flag bearish; this one is explicitly neutral, and the trade is a break of either side.
That honesty is genuinely to its credit — and it also means the pattern is providing less than the others appear to, rather than more.
What it is really measuring
Highs falling and lows rising is a description of a shrinking range. Bars are getting smaller and price is covering less ground. That is volatility contracting, and it is the entire content of the formation.
Average true range measures the same thing as a number. So do Bollinger Band width and the squeeze indicator. None of them requires choosing which highs and lows to connect, none of them changes when you change your mind, and all of them are comparable across time and instruments.
That is the practical argument against the drawing. If the information is “volatility has contracted,” a number that says so is strictly better than a picture that implies it — because the number can be compared with last month’s, and the picture cannot.
And both boundaries here are drawn, which doubles the judgement. In an ascending triangle at least the horizontal side is objective. Here, two different observers can produce two different triangles from the same bars, with different break prices and different apexes.
In practice: trading a break with no direction
With no directional claim, the method is to trade whichever side breaks. That is a legitimate approach — it is the logic behind breakout systems generally — and it has a specific cost: you will sometimes be taken in one direction, stopped, and then taken in the other.
On this site’s shared 576-bar history, of 39 closes above a 20-bar high, 85% closed back below that level within ten bars, and 38% had a higher close ten bars later against a 54% base rate for any bar.
The series is synthetic, so nothing can work on it — which makes those numbers the null model. They are what breakout trading looks like with no mechanism present, and they are the figures a real instrument has to beat before a triangle-break method is doing anything.
Volume contracting through the formation and expanding on the break is the classical requirement, and it is the only element that is not price. It is also the check that separates a genuine volatility contraction from a stretch of chart that happens to fit two lines.
The stop is the opposite boundary, and it moves every bar. Near the apex the two sides can be closer together than a single ordinary bar’s range, which means a stop there is removed by noise almost immediately. That is not a reason to avoid the pattern; it is a reason to decide in advance which part of the formation you are willing to trade.
Aggregate the chart and the triangle becomes a few small bars. The convergence needs enough bars to be visible, so the pattern is a property of the resolution you chose.
A gap out of a tight apex is extremely common on stocks, because contraction often precedes a scheduled event. The pattern resolves decisively and there was no price at which to enter.
Each side traded costs 2% of a typical bar’s range in round-trip costs on this history. A formation that produces a false break one way and a real one the other charges you twice for one move.
And the convergence is in your drawing. The order book does not narrow. What narrows is the range of prices being traded, which is volatility, which is the thing a number already measures.
What a symmetrical triangle is not
It is not a coiling spring. The energy metaphor is borrowed from physics and there is no stored quantity anywhere in a price series.
It is not a pennant. A pennant is brief and follows a sharp move, and it carries a directional expectation. This is longer and neutral.
It is not more reliable near the apex. Compression makes the stop tighter and the noise relatively larger, which are not obviously improvements.
And it is not a substitute for a volatility measure. It is a slower, less comparable way of reading the same thing.
When it fails
The characteristic failure is the double break. Price clears one side, stops out, and then breaks the other. A neutral pattern traded on both sides is uniquely exposed to this, and it converts one market move into two losing trades and a late entry.
The second failure is the apex overrun. Price reaches the meeting point and drifts sideways. There are no lines left, no pattern, and no signal — just a quiet market that a volatility indicator would have described more usefully.
A third is the arbitrary line choice. Two observers, two triangles, two break prices. Any result that depends on which four points you connected is not a result about the market.
A fourth is finding them in ranges. Inside a trading range, a stretch of narrowing bars occurs regularly, and a triangle can be drawn on most of them.
And a fifth is treating the break as directional information. Whichever side goes first is the side that went first. The pattern explicitly does not claim to know, and the break does not tell you either — it just commits you.
The original data
39 closes above a 20-bar high on this site’s shared 576-bar history: 38% higher ten bars later against a
54% base rate, and 85% back below the level within ten bars. At a 10-bar lookback, 53 events, 43% higher
and 70% back below. All of it is in research/series-measurements.json, produced by
site/measure_series.py.
Notice how the failure rate rises with the lookback: 70% at ten bars, 85% at twenty, and 100% of the eleven 55-bar breakouts. The longer the high you break, the more likely price came back to it on this data. That is a direct argument against the intuition that a bigger breakout is a better one, and it is the specific thing to test on your own instrument — because if the relationship runs the same way there, then the most impressive-looking breaks are the ones to be most careful with, whatever shape preceded them.
Related
Average true range is the numeric version of what this pattern is looking at. Ascending triangle is the variant with one objective boundary. And breakout is what trading either side of the shape actually amounts to.
Symmetrical triangles were the pattern I drew most and used least, and it took me a while to work out why: the shape was telling me volatility had fallen, which I could have read off a single indicator line without deciding which four points to connect. The drawing was doing work the number already did.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.