What Are Chart Patterns? Why the Shapes Are Just Structure With a Name

Human beings are shape-finding machines. We see faces in plug sockets and animals in clouds, and we have been doing it for long enough that it is clearly load-bearing rather than a glitch. Point that machinery at a price chart and it will find shapes there too — which is either the reason chart patterns work or the reason they fool people, depending on who you ask, and the honest answer is that it's a bit of both.

So this article does two things. It teaches you the shapes, because they are genuinely useful shorthand and because everyone else in the market is looking at them. And it is straight with you about what they're worth, which is more than nothing and considerably less than the textbooks imply.

A pattern is just structure with a name

Start with what a pattern is made of, because it's less than you'd think. A swing high is a peak — a candle that poked up higher than the ones either side of it. A swing low is a dip. That's it; that's the raw material. Our guide to swing highs and lows goes deeper, but a peak and a dip is the whole vocabulary you need here.

Now put two peaks at roughly the same price with a dip between them. Draw a line across the peaks and another under the dip. Congratulations: you have drawn a double top, one of the most talked-about patterns in trading. You didn't change anything about the price. You didn't discover anything the chart wasn't already showing. You drew two lines and used a name.

This is worth sitting with, because it dissolves most of the mystique in one go. A pattern is not a signal the market emits. It is a description a person applies afterwards — a filing system for arrangements of peaks and dips that have shown up often enough to be worth having a word for. Useful, in the way that "cul-de-sac" is more useful than "road that stops". But a word, not a spell.

The three parts every pattern shares

Here's the part that saves you memorising a dozen definitions. Whatever it's called, every chart pattern is doing the same three jobs.

1. 1The structure — the swing highs and lows, and the lines you draw through them. This is the shape itself: the ceiling price keeps failing at, the floor it keeps bouncing off, the two rails closing in on each other. It's a picture of a market that hasn't made its mind up yet.
2. 2The trigger — the break. The moment price closes outside the shape. Until that happens there is no signal at all, only a drawing: a triangle that hasn't broken is just a chart getting quieter. The trigger is what turns a shape into an event, and "closes outside" is doing real work in that sentence, because price poking a toe through and pulling it back is not the same thing.
3. 3The target — how far the move might run. The convention is the measured move: take the height of the pattern at its widest, and project that same distance from where the break happened. It isn't a prophecy. It's an order-of-magnitude estimate, on the logic that a shape which took a month and eight points to build is unlikely to resolve in a one-point move.

Learn those three and the dozen names stop being a dozen things to learn. They become one thing with a dozen labels, which is a much easier afternoon.

The two families

Patterns get sorted by what they suggest happens next, and there are only two answers plus a shrug.

The shrug is the symmetrical triangle, which is squeezing from both sides at once and genuinely doesn't tell you which way it will resolve. Some people call these bilateral patterns, which is a grand word for "could go either way". It is still useful, mind — knowing that a decision is coming is worth something even when you don't know which decision.

One honest caveat before the catalogue: these are tendencies, not rules. A continuation pattern breaking the other way is a completely normal Tuesday, and we'll get to the numbers on that shortly.

Ascending triangle

A flat ceiling with a rising floor underneath it. Sellers keep offering at one particular price, and buyers keep coming back willing to pay more than they did last time. The gap narrows, the sellers eventually run out of stock to sell, and price goes through. Usually upward — but see the numbers section, because "usually" is doing more work in that sentence than most articles admit.

Descending triangle

The mirror image: a flat floor with a falling ceiling. Buyers keep defending one price while every rally tops out lower than the last, which is the sort of thing that ends the way you'd expect. Bearish by reputation.

Symmetrical triangle

Both rails converging — lower highs and higher lows squeezing into a point. Nobody is winning; the range is simply getting tighter until something gives. The direction of the eventual break is the thing you're waiting to be told, not the thing the pattern tells you.

Double top and double bottom

Two attempts at the same ceiling, both rejected, and then the floor between them gives way. That floor has a name — the neckline — and it's the trigger; the two peaks on their own are just a chart that failed twice, which happens constantly. A double bottom is the same shape upside down: two tests of a floor, then a break up through the ceiling between them.

Head and shoulders

Three pushes up. The middle one (the head) goes highest, the two either side (the shoulders) are lower and roughly level with each other, and a line drawn under the dips between them is again the neckline. The story it tells is a trend running out of people: each wave of buying is weaker than the one before, and the last one can't even get near the high. Break the neckline and the reversal is confirmed. Turn the whole thing upside down and it's an inverse head and shoulders, which says the same thing about a downtrend.

Flags and pennants

A sharp run — the pole — followed by a small, tidy drift in the other direction, then the run continues. If the drift is a narrow sloping channel it's a flag; if it converges to a point it's a pennant. They are the most intuitive patterns on the list, because they look exactly like what they are: a market that moved hard, paused to catch its breath, and got going again. The measured move here uses the pole's height, not the flag's, which is why flags carry the biggest targets of anything in this article.

Cup and handle, wedges, rectangles

A cup and handle is a long rounded bottom followed by a small dip near the top — a U with a notch on its right-hand side — and is read as bullish. A wedge is a triangle that tilts: both rails sloping the same way while converging, which usually breaks against the tilt (a rising wedge tends to break down, a falling wedge to break up, which is the most counter-intuitive thing on this page). A rectangle is simply a range — a flat ceiling and a flat floor, price ping-ponging between them until it doesn't.

Why they look more reliable in books than on your screen

Every pattern in every textbook worked. This is not because patterns always work. It is because the author picked the examples afterwards, and a chart where the pattern failed makes a poor illustration of the pattern. Nobody is being dishonest; it's just that hindsight is an extraordinarily flattering lens, and a chart with the answer already printed on the right-hand side is a very different object from the same chart with that part covered up.

The failure mode nobody frames and hangs on the wall is this one: a textbook structure, a genuine close beyond the rail, and then nothing.

There is a second problem, and it's the one academics keep pointing at: patterns are subjective. The paper that put chart patterns through a proper statistical wringer opens by naming this directly — the shapes in a price chart, it notes, are often in the eye of the beholder. Two competent people given the same chart will draw slightly different lines, disagree about whether the right shoulder is a shoulder, and reach different conclusions. That doesn't make the exercise worthless, but it does mean "the chart is showing a head and shoulders" is a statement about the analyst as well as the chart.

So do chart patterns actually work?

Partly, sometimes, less well than advertised, and the honest numbers are more interesting than either the evangelists or the sceptics let on.

Take the ascending triangle, which has a reputation as one of the dependable ones. In the most widely cited public catalogue of pattern statistics — Thomas Bulkowski's, drawn from more than 1,400 samples — it breaks upward 63% of the time. Not 90%. Sixty-three. Of those upward breaks, about 70% reach the measured-move target, and 64% throw back to retest the breakout level first, which is a polite way of saying the trade goes against you before it goes your way. And 17% fail outright almost immediately.

Read that as an edge rather than an answer and it's a perfectly reasonable thing to have. Read it as "the triangle means it goes up" and you will be wrong roughly two times in five, which is often enough to hurt.

The academic picture is similarly middling-but-not-nothing. The landmark study applied automated pattern recognition to US stocks from 1962 to 1996 — sidestepping the subjectivity problem by having a computer do the seeing — and concluded that several technical indicators do carry some information and may have practical value. That is a careful sentence, and it was chosen carefully. A later survey of 95 studies found 56 with positive results, 20 negative and 19 mixed, while cautioning that a lot of the positive findings suffer from data-snooping and understated costs. Something is there. It is not a machine that prints money.

And the standing regulatory point applies to all of it: past performance cannot predict how a strategy will perform in future. That is true of pattern statistics exactly as it is true of fund returns.

How to use one without fooling yourself

The practical shift is small and it changes everything: treat a pattern as a hypothesis with an expiry date, not a forecast.

1. 1Find the structure before you name it. Look for the swing highs and lows first and see what shape they make. If you go looking for a head and shoulders you will find one, because you can always find one.
2. 2Wait for the trigger. A pattern that hasn't broken has told you nothing yet. Most of the pain in pattern trading comes from acting on the drawing rather than the break.
3. 3Write down what would prove you wrong — before you act. It's usually the opposite rail: if an ascending triangle breaks up and then closes back under its old ceiling, the idea has failed. That level is your invalidation point, and knowing it in advance is the entire difference between a plan and a hope.
4. 4Size it so being wrong is survivable. Two times in five, remember. A pattern with a 63% hit rate is a good tool and a terrible thing to bet the account on.
5. 5Check what the wider chart is doing. A continuation pattern pointing the same way as the trend it sits inside is a much better proposition than one arguing with it.

None of that requires believing the market is drawing you messages. It only requires noticing that a lot of people are watching the same lines, that orders pile up around them, and that this is enough on its own to make the lines matter a little.