Skip to content
CyclepediaThe trading encyclopedia
SESSION
ITENESPT-BRFRDE
James Marsden Hurst 1924—2005 Biography and works
Chapter 3.1 — Verify Your Chart Patterns

Chart patterns in Hurst's cyclic model: what Chapter 3 verifies

In Chapter 3 Hurst reconstructs channels, reversals, triangles, and moving averages with his cyclic model; Cyclepedia separates explanation from proof.

Cite this entry

This reference cites the Cyclepedia entry. The documents it draws on are listed in the entry’s sources.

On this page

In one sentence — Hurst tries to explain channels, head and shoulders, double tops, triangles, and moving averages as sums of cyclic components; a possible explanation, however, is not yet empirical validation.

Why does Hurst devote a chapter to patterns?

In Chapter 3 of The Profit Magic of Stock Transaction Timing, Hurst poses a question internal to his model: if the cyclic component proposed in Chapter 2 exists, can it generate shapes that chartists had already observed for decades?

The book answers with simulations and historical charts. A short component is added to one or more slow components: amplitude describes how much each component oscillates, and phase its position within the cycle. Changing these relationships produces familiar shapes. The result shows that the model can produce those patterns; it does not necessarily identify the cause of every real formation. Hurst treats the agreement as support for his model: that is the author's thesis, distinct from testing on new data.

Four families, one verification question Interactive map of Hurst’s proposed explanations for channels, reversal patterns, triangles, and moving averages. From shape to hypothesis A pattern flags a structure to test, not an already-proven forecast 01 Trend and channel short wave + slow background i 02 Reversal patterns same waves, different phase i 03 Triangles and flags amplitude that contracts i 04 Centered average filter, not forecast i Hurst 1970: historical explanation · modern test: rules, samples, costs Cyclepedia diagram · Emiciclo
The Chapter 3 path: from observed shape to the hypothesis intended to explain it, while keeping the boundary between historical reconstruction and modern testing visible.
Select the highlighted points to explore the detail

Channels: a sum that tilts an oscillation

In Figure III-1 (pp. 52–53), A is an idealized cyclic component; B is the sum of longer components, assumed straight and rising over the displayed interval; C = A + B is simulated price. The resulting lows rise and two parallel boundaries delimit a channel. These are neither three separate cycles nor data from a stock. The entry on trend lines and channels explains the transition from a straight line to an envelope, the curved band containing the oscillations.

Reversal patterns: relative phase changes

In Figure III-2 (p. 54), A represents roughly 18 weeks and C roughly 18 months; B remains the background of still longer components. Here C is therefore a component, whereas in III-1 it was the resultant. Figure III-3 sums the three contributions and III-4 draws a head-and-shoulders pattern on the result.

The short cycle keeps oscillating while the long one reaches its top and turns. By shifting the short cycle in time, Hurst produces a double top (Figs. III-6/7); he extends the reasoning to double bottoms and V turns. The shapes depend on the relationships chosen in the construction; they are not independent signals demonstrated in the market.

Triangles: shape does not decide the breakout

Hurst links triangles, flags, pennants, and wedges to a short component with varying amplitude and duration while the sum of longer components pauses (Fig. III-8, pp. 58–59). In other cases the same shape may accompany the turn of a longer cycle. The triangle shape therefore calls attention to a structure but does not determine the next direction by itself.

Moving averages: filters, not natural patterns

The final section interprets a moving average as a filter: the window length, or span, controls the attenuation of fluctuations. Hurst proposes placing it at the centre of the observed interval (pp. 63–66).

For regularly spaced data, a symmetric arithmetic average of N = 2k + 1 observations requires k bars after its centre. With an even N, the centering convention and time labels must be stated. Shifting the drawing does not make data available sooner. The entry on cyclic moving averages examines the historical formulation and the discrete calculation.

What is actually verified?

The chapter shows constructions compatible with the model: simple sums can recreate familiar shapes. This supports a possible explanation without proving its uniqueness or predictive effectiveness. The 1961 Perkin-Elmer case adds an example selected by the author, but it is not an independent test and does not provide a general estimate of accuracy, false signals, or profitability after costs.

Lo, Mamaysky, and Wang compare returns conditioned on patterns with unconditional returns (2000, Sec. III). The incremental information found does not amount to net profits and does not test Hurst's components.

Sources

  • J. M. Hurst, The Profit Magic of Stock Transaction Timing, Prentice-Hall, 1970, Ch. 3, pp. 51–67 — constructions, the Perkin-Elmer case, and the chapter's eleven conclusions.
  • Andrew W. Lo, Harry Mamaysky, and Jiang Wang, Foundations of Technical Analysis, The Journal of Finance 55(4), 2000, pp. 1705–1765, Secs. II–III — definitions, detection, and statistical information.
Suggest an edit My proposals