Classic reversal chart patterns are price formations that, on completion, are read as signaling the end of the prevailing trend and the start of a move in the opposite direction. The two most-cited families are the Head & Shoulders (a three-peak top, or its mirror the inverse H&S bottom) and the Double Top / Double Bottom (two equal peaks or troughs, the "M" and "W"). Each is only considered complete — i.e. a tradable signal — when price breaks a defined support/resistance line (the neckline for H&S, the intervening trough/peak for doubles). Their core tension: the geometry is intuitive and widely watched, but pattern recognition is subjective and prone to hindsight bias, and the measured real-world performance of these formations is materially below the textbook ideal. This node covers structure, the break trigger, the measured-move target, and the evidence base; it defers exact swing entry/stop/sizing mechanics to the Swing Trading branch.

How it's formed

Head & Shoulders Top. After an uptrend, price makes a peak (left shoulder), a higher peak (head), then a lower peak (right shoulder, ideally near the left shoulder's height). The two intervening reaction lows are connected to form the neckline, which may slope up, down, or be horizontal. The pattern is not complete and the trend is not reversed until price closes below the neckline (StockCharts ChartSchool). The inverse Head & Shoulders (H&S Bottom) is the same shape flipped after a downtrend — left shoulder, lower head, right shoulder — completing on a break above the neckline.

Double Top / Bottom. Two prominent peaks (top) or troughs (bottom) at roughly the same price, separated by an intervening trough (top) or peak (bottom). Critically, a double top is not confirmed until price closes below the intervening trough's low, and a double bottom not until price closes above the intervening peak's high. Two peaks at similar levels are not a double top until that confirmation break occurs — a common labeling error.

Volume. Classic theory wants volume to confirm: in an H&S top, volume ideally peaks on the left shoulder/head and is lighter on the right shoulder, then expands on the neckline break (ChartSchool). For bottoms (inverse H&S, double bottom), an expansion of volume on the upside break is the desired confirmation. Volume confirmation is treated as supportive, not mandatory.

The break trigger and the measured-move target

The actionable signal is the neckline/confirmation break — not the formation of the peaks themselves. The conventional measured-move target projects the pattern's height from the break point:

• H&S Top: target = neckline breakout price − (head high − neckline). H&S Bottom: target = breakout price + (neckline − head low).
• Double Top: target = breakout price − (peak height − intervening trough). Double Bottom: target = breakout price + (intervening peak − trough low).

StockCharts explicitly calls these targets "rough guides," not precise forecasts. Bulkowski refines the projection by the percentage of patterns that actually reach the full target (see below) rather than assuming 100% do.

Standing & evidence

These are among the most recognized patterns in technical analysis, taught in the CMT curriculum and traced to Edwards & Magee. But their reputation outruns their measured performance, and any precise hit-rate must be attributed to a named study — never asserted as fact. The most rigorous public base rates come from Thomas Bulkowski's Encyclopedia of Chart Patterns and thepatternsite.com, who manually measured thousands of "perfect" examples. His key figures:

• H&S Top: ~19% break-even failure rate, ~16% average decline, ~68% pullback rate, ~51% reaching the measured target (Bulkowski).
• H&S Bottom (inverse): 11% break-even failure rate, 45% average rise, 65% throwback rate, 71% reaching target, ranked #13 of 39 patterns; based on 3,197 perfect trades (thepatternsite.com).
• Adam & Adam Double Top: 25% break-even failure rate, 15% average decline, 64% pullback rate, 64% reaching target, ranked #19 of 36; from 1,114 perfect trades (thepatternsite.com).
• Double Bottom (commonly cited Eve & Eve / aggregate): roughly 39% average rise with ~16% failure rate per Bulkowski; figures vary by Adam/Eve sub-variant — cite the specific variant when precision matters.

Two honest caveats from Bulkowski's own work matter most for a RAG layer:

1. "Break-even failure rate" is his term for patterns that break out but move less than 5% before reversing — i.e. the pattern "worked" directionally yet didn't pay. It is not the chance the pattern fails outright; real disappointment rates are higher once normal stop-outs are counted. 2. The throwback/pullback after the break materially affects outcomes. Bulkowski finds patterns that throw back to the neckline tend to underperform those that run cleanly — yet throwbacks occur in roughly two-thirds of cases (64–68% above), so the most common path is the weaker one. The "buy/sell the retest" heuristic is double-edged.

Crucially, these are measured on hand-picked textbook examples; live, ambiguous charts perform worse. Subjectivity in drawing the neckline and declaring "equal" peaks, plus hindsight bias (patterns are obvious after the fact), inflate the perceived reliability. There is no broad peer-reviewed consensus that these patterns carry a standalone tradable edge after costs; their value is best treated as conditional and confirmation-dependent.

Strengths & limitations

• Strengths: clear, objective trigger (the neckline break is a definable line, not a feeling); a built-in target and a natural invalidation level; bottoms (inverse H&S, double bottom) measure notably better than tops in Bulkowski's data.
• Limitations / failure modes: the #1 misuse is anticipating the pattern before the confirmation break — acting on three bumps or two peaks that never confirm. Patterns "fail" via false breakouts (busted patterns that reverse back through the neckline), partial moves under 5%, and chop in range-bound regimes. Up-sloping/ambiguous necklines and unequal peaks degrade reliability. Performance is regime- and timeframe-dependent, and the throwback caveat above means the textbook clean target is the minority outcome.

System relevance

This node defines the pattern and its evidence base; the exact swing entry, stop placement, and target/scaling mechanics belong to the Swing Trading branch (cross-link there — do not duplicate). For the Delvantic system, the operative discipline is that the Augustus trade-setup agent should treat a reversal pattern as a signal only after confirmation (neckline/trough break), carry the attributed base rate (Bulkowski's failure/throwback figures, not an invented hit-rate), and downgrade conviction when a throwback is in progress — and ultimately weight the pattern by Cairn's measured live track record over the textbook number.

Sources

• Bulkowski, Encyclopedia of Chart Patterns (2nd ed.) / thepatternsite.com — Head-and-Shoulders Bottom: https://thepatternsite.com/hsb.html (11% BE failure, 45% avg rise, 65% throwback, 71% target, 3,197 trades)
• Bulkowski / thepatternsite.com — Head-and-Shoulders Top discussion: https://www.thepatternsite.com/HSTExplained.html
• Bulkowski / thepatternsite.com — Adam & Adam Double Top: https://thepatternsite.com/aadt.html (25% BE failure, 15% avg decline, 64% pullback, 64% target, 1,114 trades)
• Bulkowski / thepatternsite.com — Failure Rate Study (definition of break-even failure): https://thepatternsite.com/FailureRates.html
• StockCharts ChartSchool — Head and Shoulders Top: https://chartschool.stockcharts.com/table-of-contents/chart-analysis/chart-patterns/head-and-shoulders-top
• StockCharts ChartSchool — Head and Shoulders Bottom & Double Top/Bottom structure and measured-move rule

Flag: H&S-Top headline figures (19% / 16% / 68% / 51%) come from secondary citations of Bulkowski rather than a directly fetched top-level top page; the H&S-Bottom and Adam & Adam Double Top figures are directly from thepatternsite.com. Double-bottom 39%/16% is an aggregate — sub-variant (Adam/Eve) numbers differ. Treat all hit-rates as Bulkowski's hand-picked-sample base rates, not live edge.