The case in thirty seconds
In plain words — In Chapter 6, Hurst takes one stock, calculates four price zones in sequence, and shows that the next four extremes fall inside them. The case teaches the procedure well, but cannot tell us how often it would work elsewhere.
Alloys Unlimited first appeared in Chapter 1. It returns here to demonstrate
the computational methods. Hurst observes a
trading cycle between 17 and 22 weeks and adopts T = 20: a 10-week
half-span, a 20-week full-span, and smoothing of a roughly 12-week residual.
The following numbers and outcomes are those reported in the 1970 book.
Four consecutive estimates
The first crossing centre is 41. From the previous low near 32, the completed leg measures nine points; repeating it above 41 yields a target of 50 and a zone of 48–52. The observed top is 48¾, eight weeks later.
The half-span then turns down. From 48¾ to the crossing at 42 is about seven points, producing a target of 35 and a zone of 33½–36½. The observed low is 35⅛, four weeks later.
The third reading starts at 35⅛ and uses a crossing at 43¾. The 8⅝-point distance gives a centre of 52⅜ and a zone of 50⅝–54⅛. Price enters the zone quickly, remains there for three weeks, and reaches 52⅞.
Finally, the downward crossing at 46¾, 6⅛ points below the top, yields a centre of 40⅝ and a zone of 39⅜–41⅞. The observed low is 41, one week after price enters the zone.
Why the full-span mattered
After the 32⅛ low, the 48¾ high was below the previous 49¾ high. By eye, the channel might still have appeared to slope down. The 20-week full-span average was rising instead. In Hurst's interpretation, the slower components had already turned upward and the trading-cycle channel had probably bottomed at 32⅛. The envelope would only confirm that conclusion at the next top, three weeks later.
The inverse provides another check in the final phase. Hurst reads a 12.7-week component with roughly seven points of movement. Combining it with the longer cycle gives an estimate near 40½, close to the half-span's 40⅝ and the observed low of 41. This is convergence inside one analytical framework, not independent evidence.
What the case shows—and what it does not
The case shows that the procedure can be reconstructed step by step and that the four published outcomes are consistent with their calculated zones. It is not a backtest: it covers one stock, a selected interval, and parameters read from the chart. It does not reveal how many cases were discarded, the false signal rate, or net performance across a predefined universe.
Timing is also less precise than price. Hurst discusses one theoretical five-week estimate against nine observed weeks, while other zones were reached after roughly three to six weeks. The book itself cautions that not every stock will provide as much information as Alloys.
A modern evaluation would require rules fixed before testing, point-in-time data, every signal, costs, and an out-of-sample period. Multiple-testing research explains why a successful example is insufficient when stock, window, and configuration can be selected after observing the data. That research does not examine Alloys or Hurst; it states the evidence standard missing from this historical case.
Sources
- J. M. Hurst, The Profit Magic of Stock Transaction Timing, Prentice-Hall, 1970, ch. 6, pp. 99–113, figs. VI-1–VI-8 — data, calculations, and interpretation of the Alloys Unlimited case.
- Campbell R. Harvey, Yan Liu, and Heqing Zhu, … and the Cross-Section of Expected Returns, NBER Working Paper 20592, 2014 — multiple testing and evidence thresholds; not an evaluation of Hurst's method.
Related entries
- Hurst computational methods — the general procedure.
- Half-span and full-span — calculating the zone.
- Inverse moving average — the internal check on the last estimate.
- Operating philosophy — Alloys' first appearance.