The number, first
Every discussion of Martin Armstrong’s work eventually arrives at one figure. The Economic Confidence Model counts turning points at intervals of 8.6 years, and 8.6 years is very close to 3,141 days — the first four digits of pi, shifted by three decimal places. That is the whole basis of the nickname.
π × 1,000 = 3,141.59. The two differ by about 0.44 of a day — less than half a day across eight and a half years.
How Armstrong says he found it
Armstrong began publishing commodity forecasts in 1973, and first applied his cycle work publicly in 1977 when he called a strong bullish trend in commodity prices. But the pi observation, on his account, came later and almost sideways.
He was checking whether the crash of 19 October 1987 sat at a precise interval from earlier panics rather than landing there by chance. Converting the 8.6-year business cycle he was working with into days produced 3,141, and he took the match as evidence that the rhythm was structural rather than a curiosity. In a 1999 essay he described the moment plainly: “the total number of days within an 8.6-year business cycle was 3,141… Suddenly, there was clearly more at work than mere coincidence.”
How well does the pi match actually hold?
This is worth being precise about, because both the model’s advocates and its critics tend to round in their own favour. The answer depends entirely on how long you say a year is.
- 365.25 days (the Julian year, and the one Armstrong’s figure implies): 8.6 years = 3,141.15 days. Off pi×1,000 by 0.44 days.
- 365.2422 days (the tropical year, the astronomically correct one): 8.6 years = 3,141.08 days. Off by 0.51 days.
- 365 days (a plain calendar year): 8.6 years = 3,139.0 days. Off by 2.6 days.
So the correspondence is real to four significant figures on two of the three definitions. What it is not is a derivation. Nothing in the model explains why a cycle in human confidence should be pi thousand days long; the relationship is observed after the interval was chosen, not predicted before it. Armstrong treats that as a signature of something deeper. Critics treat it as numerology with good luck. Both readings are available from the same arithmetic, which is exactly why it is worth doing yourself.
Counting a turn on a calendar
The mechanical part of the model is genuinely simple, and it is the reason people can argue about ECM dates without access to Armstrong’s software at all. From any turning point you accept as a starting position:
- Add 3,141 days to reach the next base turn.
- Add 785 days at a time to reach each of the three quarter turns in between.
- Six base turns — 51.6 years — complete one full wave.
That is all the arithmetic there is. Everything else in Armstrong’s work — which asset, which country, which direction — comes from his capital-flow analysis rather than from the cycle count. The cycle supplies dates. It does not, on its own, supply forecasts, and the book is careful about the difference.
Why the shape matters more than the date
A common misreading of the model is that a turn is a crash. In the ECM’s own terms a turn is a change in direction of confidence, and direction depends on where confidence was pointing when it arrived. A turn at the end of a long concentration of capital looks like a panic. A turn at the end of a long unwind looks like a recovery. Same mechanism, opposite mood.
This is also why practitioners weight turns by how many nested levels coincide. A quarter turn on its own is noise; a base turn landing at the same date as a 51.6-year wave boundary is the kind of event the book spends chapters on.
A fixed interval is a strong claim.
Business-cycle research in mainstream economics does not find a constant period. Post-war expansions have run anywhere from about a year to more than a decade, and the consensus explanation involves credit, shocks and policy rather than a fixed wavelength.
The strongest version of the criticism is not that 3,141 days is wrong but that it is unfalsifiable in practice: markets produce enough notable events that a rigid grid will always sit near one, and a turn window measured in months will usually catch something.
Read the record with that in mind. Nothing here is investment advice, and a cycle date is not a reason to move money.
Read further
The Armstrong Economic Code takes the count above and applies it across four decades of Armstrong’s published forecasts, including the post-2020 period, so a reader can see where the grid landed and where it did not. For the framework it sits inside, start with the Economic Confidence Model; for the man who built it, see the profile.