Head and Shoulders: You Draw the Neckline
A head and shoulders is a three-peak formation with a higher middle peak, completed when price breaks the neckline drawn under the two intervening lows. Which peaks qualify and where the neckline sits are both decisions made by the person drawing it, which is where most of the pattern's apparent reliability comes from.
Covered on this page: TradingView.
How it works
Three peaks in sequence. The first and third are roughly level — the shoulders — and the middle one is higher, the head. The two lows between them define a line called the neckline.
The pattern is complete when price breaks below that neckline, and it is read as the end of an advance: buyers made a higher high, failed to make another, and then failed to hold the level that had been supporting the whole structure.
The neckline is a line you draw. Two lows define it, and which two lows are “the” lows is a judgment. Move one choice by a few bars and the line tilts, the break happens at a different price, and the measured target changes.
Every step is a judgment, and that is the subject
Which peaks count as shoulders? There is no threshold. A minor high three bars wide and a major high thirty bars wide both look like peaks on the right chart, and the pattern does not specify a minimum size.
How level must the shoulders be? Also unspecified. “Roughly equal” covers a range of a few percent on most charts, which on a volatile instrument is a great deal of latitude.
How much retracement between them? Unspecified again.
The textbook picture is symmetrical and real charts are not. Actual formations have uneven shoulders, sloping necklines and different durations either side of the head. Every judgment call gets made in the direction of the pattern you are already looking for, which is not dishonesty — it is how pattern recognition works, and it is why the rules have to be written down first.
On the S&P 500 from 1950 to 2026, 53.1% of days closed up — barely better than a coin. With direction that close to even, runs of closes in one direction tend to be short, so on a fast chart most “shoulders” are two- or three-bar wiggles that would not have been named at all if the surrounding shape had been different.
In practice: the parts that are actually specified
The measured move is the one piece of arithmetic in the pattern. Take the vertical distance from the head to the neckline, and project it downward from the break point. That is the target.
It is arithmetic, not a forecast. Nothing about the geometry of a completed shape creates an obligation for price to travel a matching distance. What the target does provide is a way to decide in advance whether the trade is worth taking, given where the stop has to go — and that is genuinely useful even if the target is never reached.
The retest is the only entry with a defined risk. Price breaks the neckline, returns to it, and fails there. That gives an entry near the line with a stop just above it, which is a much smaller distance than entering on the break with a stop above the right shoulder.
It also frequently does not happen, and waiting for it means missing the moves that run straight down. That trade-off is real and unresolvable; what matters is deciding which version you are trading before the break, not after.
The conventional stop is above the right shoulder, because that is the price which invalidates the structure. On a formation of any size, that is a long way from a neckline entry — which caps position size and is the main reason the pattern’s apparent hit rate matters less than people think.
The classical volume reading is that participation declines through the formation — heaviest on the left shoulder, lighter on the head, lightest on the right. It is the one part of the pattern that uses an input other than price, and it is the part most often skipped.
Aggregate the chart and the formation vanishes. Thirty daily bars become six weekly bars, and six bars cannot show three peaks and two troughs with any clarity. The pattern exists at the timeframe you chose to look at.
A gap through the neckline completes the pattern at a price you could not transact at, and removes the retest entry entirely. The pattern “worked” and the trade was unavailable — a distinction that backtests on closing prices tend to lose.
Each attempt pays the spread and commission on a round trip, which is small relative to the pattern’s own scale — one of the few advantages of a formation that takes weeks to build.
And no order exists in the order book because a shape acquired a name. If the neckline produces reaction, it is because it is a visible horizontal level that many people drew in roughly the same place — the same self-fulfilling mechanism that operates on any obvious level, and no more than that.
What a head and shoulders is not
It is not a rule. It has no numeric thresholds anywhere in its definition, which means two competent people can disagree about whether one exists on the same chart.
It is not complete before the break. “A head and shoulders is forming” is a prediction about the future stated as an observation about the present.
It is not a double top. That is two peaks at similar heights. This requires a higher middle peak, and the difference is the whole structure.
And it is not more reliable for being famous. Its familiarity is why it gets found everywhere, which is a reason for more caution rather than less.
How this connects
Three peaks with the middle one highest, which makes it a chart pattern with all the identification looseness that implies.
The neckline is the part that matters. It is a support level drawn under the pattern, and the trade is a break of it - so this is a breakout with a shape attached.
The inverse head and shoulders is the mirror, built identically at a low.
Reversals are the claim, and the pattern is one of the few with a plausible story: successive peaks failing to exceed the prior one describes buying that is weakening.
And the measured target is the weakest part - projecting the head’s depth from the neckline rests on symmetry rather than evidence.
When it fails
In a range the pattern is everywhere. Price oscillates between boundaries, so sequences of three peaks with a higher middle one occur constantly by chance. A shape you can find anywhere is not constraining your decisions.
The second failure is redrawing the neckline after the break. Price breaks, then recovers, and the line gets adjusted to the next low so the pattern is “still valid.” At that point the pattern has become unfalsifiable.
A third is the right shoulder that keeps forming. Until price breaks the neckline, the formation is incomplete, and a right shoulder can extend indefinitely — often into a new high that destroys the structure entirely.
A fourth is entering on the break without a plan for the retest. Break entries have wide stops; retest entries frequently never trigger. Choosing between them after seeing which happened is deciding with hindsight.
And a fifth is treating the measured target as a probability statement. It is a distance derived from the shape. Price reaching it, exceeding it or ignoring it are all ordinary outcomes, and the target’s value is in sizing the trade beforehand, not in expecting the number.
The original data
The relevant measurement for this pattern is not a hit rate — it is how long directional moves actually last on the data underneath it. Across the S&P 500’s daily closes from 1950 to 2026, 53.1% of days finished up and 46.9% down, a split close enough to even that one close says little about the next.
That figure is the reason the peak-selection problem is not academic. With direction that close to a coin flip, a chart covering a couple of years of daily bars contains a great many local peaks, and any three of them in the right arrangement satisfy an unquantified pattern definition. Write down what counts as a peak — a minimum number of bars, a minimum retracement — before you go looking, and the number of head and shoulders formations on your chart falls by an order of magnitude. That single constraint does more for this pattern than any refinement of what to do after the break, and counting the up and down runs on your own chart is the way to calibrate it.
Related
Chart patterns is the parent page on named shapes and how much of them the observer supplies. Inverse head and shoulders is the mirror formation. And reversals covers the event this pattern is trying to identify.
I write the peak-selection rule down before I open the chart. If you can draw a head and shoulders on almost any chart handed to you, the pattern is not constraining your decisions, and that is the opposite of skill. Decide in advance what counts as a shoulder and the shape starts doing some work.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.