Condensed case — Same chart, two readings: triangle charting rule vs cyclic state table. For Fig. III-9 and full tables → encyclopedia entry.
Source: Hurst (1970), Ch. 3 — How to Tell in Advance if a Chart Pattern Will "Fail". Perkin-Elmer daily high-low, March–September 1961.
Context
Case from a period charting manual: two triangles on the daily. Before judging patterns, Hurst measures four nested periodicities (lows, counts, averages):
| Scale | Average duration | Model nominal |
|---|---|---|
| A, B, C | 13.9 weeks | 13 weeks |
| points 1–7 | 4.6 weeks | 3.25 wk (1961 equiv.) |
| — | 2.3 weeks | 1.625 wk |
| — | 1.15 weeks | half of previous |
The chartist applies the rule: the triangle continues the preceding trend. The cyclic model fills each cycle's state and sees further.
Setup
First triangle (top): three 1.15-week cycles — more properly a flag. At the break, all four cycles point down, decisively. Charting rule: continuation of the prior uptrend → failure forecast.
Second triangle (right): built from 1.5 cycles of 4.6 weeks while the 13.9-week cycle rolls over (double-top mechanics). State: 13.9 and 4.6 down decisively; 2.3 down; 1.15 at a flat/up low. Plus, the prior 13.9-week low was lower → long sum still down. Forecast: downside break — agrees with the chartist, but with cyclic reasoning.
Card — Three-step method
- Measure the stock's cycles (Ch. 2).
- Fill the state table for each cycle.
- Read the pattern: exit follows dominant cycles.
Sequence
- Cycle count — four periodicities on Perkin-Elmer daily; nominal deviations documented (13.9 vs 13; 4.6 vs 3.25).
- First triangle — downside break; charting rule wrong, model right.
- Second triangle — downside break; chartist and model agree, but only the model explains why and would have forecast the first failure.
- Pattern outcome — both cyclic forecasts confirmed on real data.
Outcome / Lesson
Cycle knowledge turns the pattern from bet to information: you can forecast when the triangle rule will fail, not only when it will agree. The case shows pattern verification (Ch. 3) does not replace periodicity measurement — it presupposes it.
Lesson — Same chart, two languages. The chartist logs a «failure» with no explanation; Hurst had anticipated it from the state table.
Full encyclopedia entry
Case Perkin-Elmer (1961) — original Fig. III-9, cycle-by-cycle tables.