2013issue C0426-29
Constructing head-and-shoulders milestone points
A head-and-shoulders can be constructed from a small set of named closing-price landmarks. Two scale-free parameters locate those milestone-points, which still require a later condition set before they outline a pattern.
- A simple percentage-swing connector is a poor construction tool because real charts produce many zigzag variants of the same textbook shape.
- Classic formations are scale-free, so the construction uses an n-window and a factor rather than a fixed swing threshold.
- The labeled milestone-points can be drawn even when no head-and-shoulders is present, and they need not be the most prominent local highs and lows.
- A later condition set, including neckline-penetration, is required before the points outline a completed pattern.
A poor tool for a scale-free shape
A simple percentage-swing connector is a poor construction tool for head-and-shoulders because real charts produce many zigzag variants of the same textbook shape. Classic chart formations are treated as scale-free. They may span a few bars or hundreds, and their swings may be tiny or large, so a fixed swing threshold is inefficient.
What the construction is designed to do
The construction is designed to stay scale-free, use as few parameters as possible, cover many head-and-shoulders variants without becoming too loose, and ignore price action outside the formation. The procedure is meant to mark completed closing-price head-and-shoulders structures after the close has already pierced the neckline, and to do so in uptrends, sideways markets, and downtrends.
This article only prepares the milestone-points that a later neckline-penetration test will use.
Start with two parameters
Two parameters start the high search. The n-window, with default 5, is the recent-bar lookback used to find the lowest close X that starts the head search. The factor, with default 1, is the left-to-right duration multiplier that sets how far left of the head the search for the left shoulder may travel. It scales the left-half duration from the right-half bar count between the head and X.
That pairing is a scale-free-construction. It locates the points from relative bar distances and local extremes so the same steps can apply from a few bars to a long formation.
Mark the head and the left-half anchor
The head H is the highest close between the first leftward close B that is at or below X and the recent low X. D is then the close Lhalf bars before H.
Locate the shoulder highs
Temporary lows in the rightmost two-thirds of D-to-H and the leftmost two-thirds of H-to-X locate shoulder highs. H1 is the highest close from D to tL1, and H2 is the highest close from tL2 to X.
Close the skeleton with inner and outer lows
Inner lows are built from local troughs and fractional averages. L1 is the first leftward close at or below (tL1 + 2E)/3, and L2 is the first leftward close at or below (tL2 + F)/2. Outer lows close the skeleton. L0 is the first leftward close at or below L1 plus one-fifth of the H1-to-L1 range, and L3 is the first leftward close from the last bar that is at or below L2.
These labeled closing-price highs and lows, H, H1, H2, L0, L1, L2, and L3, are the milestone-points. A recognition procedure must first define them before any pattern test can run. In this usage, a head-and-shoulders is a five-peak closing-price structure whose left shoulder, head, and right shoulder are located from those named milestone highs and lows rather than from a visual sketch.
Points first, pattern later
These milestone-points can be drawn even when no head-and-shoulders is present. They also need not be the most prominent local highs and lows, so a later condition set is required before the points outline a pattern.
All readings on this track · 37 readings
- 1982Head and shoulders as a three-path completion test
- 1984Stock low clusters as a cycle baseline
- 1985Four-phase construction of the head-and-shoulders reversal
- 1989Volume-confirmed reversal patterns, stops, and measured objectives
- 1991The journal as one checklist for taken and skipped trades
- 1991Head and shoulders as a direction hypothesis
- 1991Candlestick body and shadow construction with three-Buddha peaks
- 1992A three-count drill that binds candlesticks, head and shoulders, and entry rules
- 1997Constructing bump and run reversal channels
- 1998Testing reversal formations in bond futures
- 1999Construction first: extra shoulders, the neckline, and the diamond test
- 1999Dead-cat bounce, rollover, and failed reversals
- 1999Evaluating time gaps in bond reversal patterns
- 2000Constructing head and shoulders and double reversal patterns
- 2001Constructing broadening and complex bottoms
- 2001Constructing a slanted head-and-shoulders when the chart is tilted
- 2002Head and shoulders with dominant-cycle timing
- 2002Trendline breaks, right shoulders, and trailing stops
- 2003Confirmation tests for bearish top patterns
- 2003Commodity top hypotheses on a dollar rebound
- 2003A head-and-shoulders test during a bear rally
- 2004Pattern breakouts need a primary-trend filter
- 2004Reading candlestick closes on trendline and neckline tests
- 2004Candle diagnosis needs Western targets and stops
- 2004Head-and-shoulders neckline construction
- 2005A familiar chart condition is a hypothesis, not a completed decision
- 2005A 50-day ceiling and a rising-floor stalemate
- 2006A complete trading plan from philosophy to checklist
- 2006Thin-market head and shoulders with two averages and MACD confirmation
- 2010Head and shoulders as a playback-tested setup
- 2011Turning a head-and-shoulders outline into a breakout hypothesis
- 2011Volume-confirmed head and shoulders on AIG and Citigroup in 2007
- 2013Constructing head-and-shoulders milestone points
- 2013Head-and-shoulders geometry versus the filter stack
- 2013Algorithmic head-and-shoulders construction
- 2018International relative strength as a double-top case study
- 2019Structure invalidation before comfort-stops