In plain words — Hurst redraws extremes as if every oscillation lasted the same time: the channel becomes tidier, but the latest point cannot be confirmed without seeing what happens next.
What problem is it trying to solve?
An ordinary envelope surrounds prices in calendar time. Oscillations assigned to the same component, however, do not have identical duration and amplitude. When a low arrives earlier or later than expected, the channel can look irregular and its center line can be harder to read.
In Chapter 5 Hurst introduces a non-real-time, or NRT, representation. It does not change the prices of the extremes; it changes their horizontal position. Each selected high or low is placed on the next pseudo-time vertical so that half-waves occupy regular graphic intervals.
How is the transformation constructed?
The process begins with a component already selected on the ordinary chart. The analyst identifies a sequence of high, low, high, and so forth. Those extremes are copied onto equally spaced verticals while each extreme keeps its observed price. Between two extremes, a line can be drawn or enough shape retained to read the channel.
The result removes part of duration variation from the display. Hurst can then draw a more uniform envelope and observe whether recent extremes bend its boundary or median. In the Gruen case he uses this representation to estimate an upper area around 12¾ and a later lower area around 11 while preparing the exit discussed in the chapter.
Those figures are historical estimates tied to that drawing. They are not universal levels and do not come from an algorithm that can be applied without choosing a component and its extremes.
Why is it called non-real-time?
An extreme is not recognizable at the exact instant it forms. To know that a point is truly a low, later rising prices must be observed; a high requires the reverse. The historical chart can therefore position an extreme perfectly only after confirmation.
The right edge also lacks the next turning point. Hurst proposes estimating the new extreme as data arrive, but the complete retrospective figure contains information the operator did not possess earlier. This difference is the central issue, not a cosmetic flaw.
How can false anticipation be avoided?
A modern evaluation must record, at each update, which extremes were already confirmed and which estimate was visible then. It is not valid to reconstruct all pseudo-time using final extremes and treat that curve as if it had been available in real time.
Rules for confirmation, revision, and deletion of extremes must be fixed; vintage versions of the chart should be compared; every signal included; and development separated from testing. Research on backtest overfitting more generally shows why selecting a configuration after many trials can produce results that degrade out of sample. It does not test the NRT envelope, but it supplies the relevant methodological boundary.
Evidence boundary — Pseudo-time may make a historical structure easier to read, but it creates no new information. If extremes are fixed with later data, a retrospective signal does not demonstrate real-time anticipation.
Sources
- J. M. Hurst, The Profit Magic of Stock Transaction Timing, Prentice-Hall, 1970, Ch. 5, pp. 91–94, Figs. V-2 and V-3 — construction and use of the non-real-time envelope in the Gruen case.
- David H. Bailey, Jonathan M. Borwein, Marcos López de Prado, and Qiji Jim Zhu, The Probability of Backtest Overfitting, Journal of Computational Finance, 2015 — selection risk and out-of-sample degradation; it is not a study of Hurst’s method.
Related entries
- Profit and exit management — chapter context
- Trailing and profit-taking — operating use in the case
- Curvilinear envelope — general concept