Who this entry is for — Duration variation warps real-time envelopes and hides information exactly when it matters. The remedy is radical: take time out of the picture. One extreme per vertical, equal spacing — and the cycle's shape becomes readable again.
Source: J. M. Hurst, The Profit Magic of Stock Transaction Timing, Prentice-Hall, 1970 — Chapter 5, How to Make and Use Non-Real Time Envelopes (pp. 91–93, Fig. V-3).
Prerequisites
The curvilinear envelope and the variation principle — the problem this technique solves.
The construction
Card — How it is built (Ch. 5)
- 1. Tabulate the prices of the successive highs and lows of the component you want to study (from the real chart).
- 2. On a separate sheet, plot the first low on the first vertical, the next high on the second, the following low on the third — one extreme per vertical, until the data is exhausted.
- 3. Form a smooth, constant, minimum-width envelope through the points.
Horizontal distances still represent time, but equal distances no longer mean equal elapsed time: component duration changes from any cause whatsoever are completely eliminated. While durations run regular, the NRT resembles the real envelope; when variation bites, "the non-real time envelope will give you a much clearer picture of what is happening".
The asterisk trick and the projection
In plain words — Every plotted high has its "twin" on the opposite bound, one channel-width away. Marking them all, the channel runs full width up to the last data point — and can be extended two verticals into the future: the estimate of the next high and low.
Beneath each plotted high, mark an asterisk on the lower bound (and above each low, on the upper); the last data point receives its twin at channel-width distance. The channel is now complete and up to date, and extending its bounds two verticals ahead yields the expected values of the next extreme and the one after.
The return to the real chart closes the loop: average durations ± variation → time windows; prices from the NRT chart ± tolerance → two time × price "boxes" where the stock should move next. Continuously updated — "you will be surprised at how often the stock behaves per prediction". As a bonus, the NRT often catches non-cyclic changes that the real-time envelope's uncertainty band masks.
The proof on Gruen
On the day of the sell-triangle's resolution the question was: how much upside is left? Labelling the 4-week cycle's extremes A…H plus the tentative low I, the NRT verdict is blunt: to reach I the envelope must curve over sharply — and even if I were not a cycle low, the curvature would still be forced "unless this low were as high as 14", highly unlikely. Major components topping: prepare to take profits. Continuing the turnover two verticals: next peak ~12¾, next low ~11 — so the actual top at 13 and the sell signal at 12⅜ "come as no surprise".
Warning — The same conclusion had come from the triangle analysis; but the triangle "in many cases will not form". The NRT depends on no pattern: it is the general confirmation. And the book's closing comparison is merciless: the real-time envelope, at the same point, gave no hint of a turnover "for nearly another week — too late".
Summary card
| Element | Rule |
|---|---|
| Construction | One extreme per vertical, constant minimum width |
| What it eliminates | All duration variation |
| Asterisks | Each extreme's twin on the opposite bound → full channel to date |
| Projection | Two verticals ahead → next high and low |
| Back to real time | Time windows (durations ± variation) × NRT prices ± tolerance = "boxes" |
| Added value | Sees turns earlier; depends on no pattern |
Links
- Trailing sell and take-profit — the signal the NRT confirms
- Profit management — Chapter 5's framework
- Curvilinear envelope — the real-time version
- Hurst tradition — chapter index