RI:
Martin Armstrong presents his Economic Confidence Model (ECM) as a reliable tool for understanding and forecasting the economy through fixed-period Pi-based cycles anchored to specific dates in history. These cycles are treated like rigid clockwork waves that dictate booms, busts, and major events.
This is bad economics because real economic activity is the result of complex, interdependent, and constantly evolving factors — monetary policy, technological change, geopolitics, psychology, and random shocks. It cannot be accurately captured or predicted by fixed sinusoidal cycles with predetermined turning points. What appears as recurring cycles in hindsight is often the product of overfitting historical data and numerological pattern-seeking rather than a genuine causal model. A real economic model should be falsifiable and adaptable to new data. Armstrong’s ECM functions more like a predetermined timetable than a scientific model of economic behavior. Relying on it for investment or -policy decisions is therefore poor economic reasoning.
-
Martin Armstrong’s ECM begins with a simple historical average: He took 26 major panics over 224 years and derived an ~8.6-year cycle, then anchored it and projected it forward. To improve the fit, he expanded it into a fractal system of multiple overlapping cycles:
- 8.6-year main cycle
- 2.15-year sub-cycles
- 51.6-year generational waves
- 309.6-year super-cycles
- plus various panic and volatility layers
This explosion of sub-cycles creates massive overfitting. With so many potential turning points (roughly one per year) and vague definitions of what counts as a “hit,” virtually any event can be retrofitted by cherry-picking the right layer. The model becomes unfalsifiable because it relies heavily on hindsight selection rather than making clear, testable predictions in advance. In practice, the ECM functions like a computer program whose rules are written after seeing the results. Armstrong openly describes “back-testing” deeper into history and adding complexity whenever the original cycle didn’t fit — classic data-snooping that destroys genuine predictive power while creating the illusion of accuracy. A real economic model should be based on causal mechanisms, remain falsifiable, and demonstrate reliable out-of-sample performance.
Armstrong’s ECM does none of these. It is a flexible pattern-seeking framework, not a scientific model of economic reality.
For a much more detailed breakdown with specific historical examples and charts, see my full critique here:
Martin Armstrong's Economic Confidence Model (ECM) – A Critical Examination