Who this entry is for — Chapter 2 showed what the envelope is; here is the how: the step-by-step manual procedure, how long it stays valid, and the state table that turns channels into predictions.
Source: J. M. Hurst, The Profit Magic of Stock Transaction Timing, Prentice-Hall, 1970 — Chapter 4, How to Construct Curvilinear Envelopes and Prediction of Price Turns Using Envelopes (pp. 69–71).
Prerequisites
Curvilinear envelope and Hurst nominal cycles. Here theory meets pencil.
How much chart you need
In plain words — First rule: the chart must show at least one and a half cycles of the component longer than the one you want to trade — because that one, with its larger magnitude, decides where price goes.
Start from a weekly high-low chart covering at least 1.5 cycles of the periodicity next longer than the trading cycle — and, if practical, the second longer one too. The reminder is proportionality: long regularities have much larger magnitudes than your trading cycle, "and can therefore be very influential in determining where the price of the stock goes, regardless of what your trading cycle is doing".
The procedure, step by step
Card — Envelope construction (Ch. 4)
- 1. Spot the dominant component by eye (usually some variation of the 13- or 26-week nominal).
- 2. Lightly sketch the two bounds of a band following that movement; measure the vertical thickness at several points (scratch paper), pick the average and correct to constant width.
- 3. If data points remain outside, widen until everything is enclosed.
- 4. Repeat while narrowing: there is normally a minimum width that contains everything except 2–3 stray points, obvious because they sit alone between your two bands. Erase the wider one: it has served its purpose.
- 5. From the final band's contacts: clearly demarked lows → count weeks low-to-low, discard variants, average. Tabulate.
- 6. The dominant component's peaks and valleys are points of a second envelope: sketch it with the same technique. With two channels you have identified three components (inside the narrow channel a still shorter one usually shows: mark its lows with a check, without forcing a third channel).
The compromise between the two criteria — enclose everything, minimum width — is explicit in the book: on the guiding example three brief overruns stay outside, and for each Hurst shows the channel "could not have been drawn in any other manner without considerable widening".
The state table
In plain words — For every identified component: its average duration, when it made its last low, hence where it stands now. Filled in for the past, it explains every move on the chart; filled in for today, it is the prediction.
With the average durations in hand, prepare a table showing the condition of every component just prior to each major move on your chart: weeks counted from the last low, remembering that any periodicity peaks and turns back down at one half its average duration. If the channels are drawn properly, "the resulting tabulations will completely explain every fluctuation you see".
Then re-run the tabulation for the last data point, hunting the golden configuration: two or more components due to bottom out nearly together, with the sum of all longer ones hard up. Once found, you track the issue closely — on to Chapter 4's signals.
Validity over time (and the free alert)
Warning — The envelope is valid history from the last well-defined high or low backward: never redraw it there. Forward it is a light estimate, updated bar by bar; the maximum lag is half a cycle, and it resets to zero at every new extreme — the "particularly low-risk decision point". Bonus: a single day outside the estimated channel is an immediate alert — either fundamentals are moving, or the long sum was misestimated.
Two practical notes from the chapter: check whether a whole number (2–3) of shorter components fits inside your chosen trading cycle — each says something about the state of the next longer one; and keep a daily chart alongside (≥1.5 cycles of the trading component) as a fine vernier on timing. And if a stock makes no "cyclic sense": drop it — "there are literally thousands to choose from".
Summary card
| Step | Rule |
|---|---|
| Data | ≥1.5 cycles of the component longer than the trading one |
| Width | Constant; the minimum enclosing all but 2–3 stray points |
| Second envelope | On the dominant's peaks/valleys → three components known |
| State table | Weeks from last low; peak at half duration |
| Setup | ≥2 lows due together + long sum hard up |
| Validity | History to the last extreme; forward only an estimate, lag ≤ ½ cycle |
Links
- Graphic buy timing — where the table leads
- Case Gruen Industries — the procedure applied to blind data
- Curvilinear envelope · Nesting envelope
- Hurst tradition — chapter index