A head and shoulders is a five-point formation: a peak, a higher peak, then a lower peak, with the two troughs joined by a neckline. It fails often, and the reason is simple — the shape carries no information until price closes through that neckline. Traders who enter on the shape are trading a drawing. Traders who wait are trading a level.
That distinction is the whole page. Everything below is either evidence for it or a consequence of it.
The five points that define the pattern
The formation has a precise anatomy, and it is worth stating exactly, because most disagreement about whether a head and shoulders "counts" comes from two people marking different things.
- Left shoulder — a peak in an existing uptrend, followed by a pullback.
- Head — a higher peak, followed by a pullback to roughly the level of the first one.
- Right shoulder — a peak that fails to exceed the head, usually closer in height to the left shoulder.
- Neckline — the line drawn through the two troughs either side of the head. It is frequently sloped, not horizontal.
- The break — price closing through the neckline. This, and only this, is the trigger.
An inverse head and shoulders is the same formation upside down at the end of a downtrend. Everything on this page applies to both.
What happened when the Federal Reserve tested it
This pattern has been studied more rigorously than almost any other, which makes it one of the few places a trader can replace opinion with evidence.
In 1995 the Federal Reserve Bank of New York published Staff Report No. 4, "Head and Shoulders: Not Just a Flaky Pattern", by P.H. Kevin Chang and Carol L. Osler. They built a computer-implemented algorithm using criteria taken from published technical analysis manuals, applied it to daily exchange rates of major currencies against the dollar across the whole floating-rate period from March 1973 to June 1994, and compared the resulting profits against 10,000 simulated price series generated by bootstrap under a random-walk null hypothesis.
Two findings are worth carrying around.
First, the result was split. The rule showed forecasting power for the German mark and the Japanese yen, and did not for the Canadian dollar, the Swiss franc or the French franc. A pattern that works on two instruments out of five is not a law of markets; it is a behaviour some markets exhibited and others did not.
Second, and more uncomfortable for anyone selling the pattern: the authors concluded the rule was profitable but not efficient, because it was dominated by simpler trading rules. The elaborate shape earned less than plainer methods applied to the same data. That is arguably the most useful sentence ever written about the head and shoulders.
How common is it, really?
Common enough that finding one is never the constraint. In the Journal of Finance study by Andrew Lo, Harry Mamaysky and Jiang Wang — the same automated-detection work discussed in double tops and double bottoms — head and shoulders and its inverse were the second most frequently detected of ten classical patterns across US stocks from 1962 to 1996, with over 1,600 occurrences of each.
That abundance is exactly why the pattern feels reliable in hindsight. There are enough of them in any price history that the ones which worked are easy to find and screenshot, and the ones which did not are easy to overlook. This is the survivorship problem described in how to tell if trading results are real: a gallery of successes is not evidence of a success rate.
Why it fails — the three common ways
| Failure mode | What you see | What it usually means |
|---|---|---|
| No break | The right shoulder forms, price drifts sideways, the neckline never gives way | There was never a trade. You drew a pattern that did not trigger. |
| Break and reclaim | Price closes through the neckline, then closes back above it within a few bars | The break had no participation behind it. Your invalidation has printed. |
| Right shoulder overrun | The right shoulder exceeds the head | The formation is void. The trend made a new high, which is the opposite of a reversal. |
Notice that two of the three are not really the pattern failing. They are a trader having committed before the pattern gave a signal. A head and shoulders in progress is a hypothesis, and the neckline close is the test.
The measured move, and what it is not
The conventional target is the vertical distance from the top of the head down to the neckline, projected from the break point. It is useful arithmetic and a poor forecast.
Nothing in market structure connects the height of a past swing to the length of a future move. What the measurement genuinely tells you is how much room the formation has created — a tall head over a distant neckline describes a volatile instrument, a shallow one describes a quiet one. Use the projection to sanity-check whether the trade has enough room to be worth the risk, which is the calculation in the risk-reward ratio, then let the real levels on your chart decide where you take profit.
Where the stop goes
Above the right shoulder for a top, below it for a bottom. That is the last price at which the pattern's own logic still holds: exceed it and the sequence of lower highs is broken, so the reason for the trade has gone.
Placing the stop just above the neckline is the common mistake and a costly one, because the neckline is routinely retested from the other side after a break — that retest is standard behaviour, covered in what is a retest in trading. A stop there will be taken out by the most ordinary thing the pattern does.
Because the shoulder-to-neckline distance can be wide, the stop distance has to set the position size rather than the other way round. That direction of travel is the entire point of position sizing from risk. If the correct stop makes the position uncomfortably small, the honest conclusion is that this particular head and shoulders is not worth trading.
How to use it without pretending
Three uses hold up under the evidence above.
- As a structure marker, not a signal. The formation tells you a trend has stopped making new highs. Trade the neckline as a level, using the process in support and resistance.
- As confluence on a level you had already marked. If the neckline coincides with a level from your pre-session markup, the level is the reason and the pattern is the detail.
- As a reason to stop adding. A right shoulder failing to exceed the head is a clean, early argument against pressing an existing long.
What does not hold up is the anticipated entry — selling the right shoulder because the picture looks right. The Fed study measured a rule that waited for the break. Anything done before the break is untested and unlike what was measured. The wider framework sits in technical analysis basics, and the neighbouring formations are covered in chart patterns: flags, triangles and wedges.
Frequently Asked Questions
What is a head and shoulders pattern?
A head and shoulders is a five-point formation: a peak, a higher peak, then a lower peak, with the two troughs between them joined by a line called the neckline. It is read as a reversal signal, but the shape alone signals nothing. The pattern is only considered triggered when price closes through the neckline.
How reliable is the head and shoulders pattern?
Less reliable than its reputation suggests, and it depends on the market. When the Federal Reserve Bank of New York tested an algorithmic version on daily exchange rates from March 1973 to June 1994, it found forecasting power for the German mark and the Japanese yen but not for the Canadian dollar, Swiss franc or French franc. The researchers also concluded the rule was dominated by simpler trading rules.
Where do you put the stop on a head and shoulders?
Above the right shoulder for a topping pattern, below it for a bottoming one. That point is the last price the pattern's logic says should not be exceeded, so it is the level that genuinely invalidates the idea. A stop placed just above the neckline is too tight, because the neckline is routinely retested after the break.
What is the measured move target on a head and shoulders?
The conventional target is the vertical distance from the top of the head down to the neckline, projected from the break point in the direction of the break. It is a geometric projection, not a forecast, and nothing obliges price to reach it. Treat it as one possible target among the levels already on your chart rather than as an expected outcome.
Bottom line
The head and shoulders describes a real event — a trend that stopped making highs and then lost its support. It has been tested more honestly than most chart patterns, and the honest answer that came back was mixed: it worked on some currencies, not others, and simpler rules did better. That is not a reason to discard it. It is a reason to demote it from signal to context: wait for the neckline close, put the stop above the shoulder, and size the trade from that distance. The shape is the story. The close is the trade.