Head and Shoulders

1. 1Left shoulderAn advance to a high, then a pullback. Ordinary trend behaviour at this stage.
2. 2HeadA higher high, then a pullback to roughly the previous trough.
3. 3Right shoulderAn advance that fails below the head, then a decline.
4. 4Neckline breakPrice falls through the line joining the troughs. Only here is the pattern complete.

The neckline is frequently not horizontal. A downward-sloping neckline describes a structure that was already deteriorating; an upward-sloping one describes a shape that needed more to break it. Both are valid and the slope is worth noting.

A trend making higher highs stops making them. The right shoulder is the first advance that fails to exceed the previous one, which is a change in the structure of the trend rather than a mysterious shape.

Framing it that way is useful because it makes the pattern's real requirements visible. It needs a prior uptrend to reverse, a genuine failure to make a new high, and a support level that actually broke. A shape satisfying the outline but not those three conditions is a coincidence.

Any noisy price series produces sequences of three peaks with a higher middle one regularly and by chance. The pattern is identified far more often than it completes, and it is identified most confidently in hindsight, after the neckline has already broken.

The discipline that helps is refusing to name it before confirmation, and requiring the shape to appear after a genuine advance rather than in the middle of a range where three peaks mean nothing at all.

Convention projects the distance from the head to the neckline downward from the break point as a target. It is a rule of thumb with no mechanism behind it.

It is best read as a rough scale for the pattern rather than as a forecast: a shape whose head sits forty percent above the neckline is describing a much larger structural change than one where the gap is four percent, and the projection is a way of expressing that difference.

The conventional target projects the distance from the head to the neckline downward from the break point. There is no mechanism behind it: it is a rule of thumb that assigns a target proportional to the size of the structure, and its main virtue is that it scales.

That scaling is the defensible part. A pattern whose head sits forty percent above its neckline is describing a much larger change in the security's situation than one where the gap is four percent, and any target that ignored the difference would be worse than one that did not.

The same criticism applies to every measured move in the pattern vocabulary, including flag poles and triangle heights. All of them are the same idea: size the expected move to the size of the structure, because nothing better is available from the chart alone.

Three peaks with a higher middle one is one of the most common arrangements a noisy series produces. In a random walk it occurs constantly, which means finding one is close to no information at all.

What narrows it is the set of conditions the description implicitly requires and that a visual identification skips. There has to have been a prior advance to reverse; the middle peak has to be a genuine failure to sustain a new high rather than an ordinary fluctuation; and the neckline has to be a level that actually held more than once.

• Without a prior uptrend there is nothing to reverse, and three peaks in a range mean nothing.
• Without a defined neckline that held at least twice, the break is a break of a line drawn to fit.
• Without volume declining into the right shoulder, the description of exhausted demand is not supported.
• Identified after the neckline has already broken, the pattern has forecast nothing.