Who this entry is for — The last chapter is "for those individuals of curious mind" who want to investigate market cyclicality on their own, with no mathematical background. It is the bridge between the operational method (Ch. 1–8) and its scientific foundation (the Appendices) — and the toolbox that produced Fig. IX-4.
Source: J. M. Hurst, The Profit Magic of Stock Transaction Timing, Prentice-Hall, 1970 — Chapter 11, Spectral Analysis — How to Do It and What It Means (pp. 168–186).
Prerequisites
The DJIA's six components (what these tools produce) and the X motivation (why it is worth it).
Why numerical analysis
In plain words — The temperature in Los Angeles is a continuous line; a stock's price is not: it exists only when a transaction occurs. A stock history is a sequence of distinct numbers — and numerical analysis was formulated precisely to squeeze information out of such sequences.
The chapter's reasoning in three steps: price histories are sequences of numbers; on those numbers we need spectral analysis of price motion; and of the possible routes, the one that applies directly to the numbers is numerical analysis.
What a "frequency spectrum" means
The fluctuations, regularities and periodicities of the earlier chapters are, precisely, sine waves. A sine wave is completely described by three quantities: period (the duration), amplitude (peak to valley) and phase (its position in time against a reference). A frequency spectrum is the map of the existence and nature of those sine waves inside a series.
From the period comes the frequency — its reciprocal: a 6-month cycle makes 2 cycles per year — and the angular frequency ω = 2π/T, in radians per unit time (converting everything to radians/year avoids confusion: from weekly, ×52).
The chapter's vital point — Any time history can be reproduced to any accuracy by summing sine waves: having a spectrum does not prove the series was born by summing cycles. Draw a ruled line and analyse it, and Fourier will hand you a spectrum for that too. But methods exist to tell the two cases apart — and for stock prices the answer is sharp: the random and fundamental parts of price motion are not generated that way, while the "X" part is. "If this were not true, knowledge of spectral components would not necessarily imply predictability." This is where the ≈23% stops being an artifact and becomes an intrinsic process.
The toolbox
| Tool | What it does | Entry |
|---|---|---|
| Fourier analysis (Lanczos method) | Measures amplitude and frequency of every "slice" of the spectrum — the starting point when you suspect hidden periodicities | Fourier analysis, step by step |
| Numerical filters (Ormsby) | The sieve: isolates a band of periods, suppressing the rest — the tool that produced Fig. IX-4 | The Ormsby numerical filters |
| Curve fitting | Drawing conclusions from the outputs: least-squares line, parabolic interpolation, Prony's method | below |
Curve fitting, briefly
Filter outputs must be interpreted. Three techniques cover "99% of all your market research work":
- The least-squares straight line — to give the trend of an output an equation (say, the drift of a cycle's period over time). Six numbers to jot down (N pairs, C=Σω, D=Σt, E=Σt², F=Σωt, G=(Σt)²) and two constants: A = (CE−DF)/(NE−G), B = (NF−CD)/(NE−G) → ω = A + B·t.
- Parabolic interpolation (Appendix V) — a parabola through each consecutive triple of outputs: brings filters designed on different spacings to a common interval ("for the charts in this book, all filter results interpolated down to one week").
- Prony's method (Appendix VI) — the directly sinusoidal fit: frequency, amplitude and phase measured objectively from the output.
Summary card
Card — The chapter's final warnings
- "Sampled" data hides conceptual traps: before applying a new method, always search the references for the known stumbling blocks.
- Fourier is powerful, but results should be verified and expanded by other spectral methods before conclusive decisions.
- Numerical filters are the natural next step; statistics and curve fitting complete the kit.
- These tools are at home in universities — "they do not appear to have been used with equal vigor in the analysis of stock price motions. Why don't you try them for yourself — perhaps it will make the competitive edge difference".
Links
- Fourier analysis · The Ormsby filters — the tools, one by one
- The DJIA's six components — the masterpiece these tools produced
- Cyclic X motivation — what the spectrum proves
- Appendix I — the DJIA's high-resolution spectrum
- Hurst tradition — chapter index