Waves, Not Cycles
Interview with Elliott Prechter
Notes from my conversation with Elliott Prechter, recorded for the Foundation for the Study of Cycles interview series.
Elliott waves and time cycles both claim that markets repeat, and they claim it about different things. That distinction carried this episode of the FSC interview series, recorded with Elliott Prechter, President of Qualitative Analytics and developer of EWAVES, the software that automates Elliott wave analysis. The lineage behind him is three generations deep: R.N. Elliott catalogued the pattern in the 1930s and published “The Wave Principle” in 1938, A.J. Frost and Robert Prechter codified it in their 1978 book, and Robert’s son Elliott has spent the past decade turning it into executable code. His own path runs through MIT (2002), Microsoft (2006), a stint as CTO of an algorithmic hedge fund, and a guest piece in The Elliott Wave Theorist that recommended Bitcoin in 2010 at 6 cents.
Form, not period
A cycle repeats in time. It has a length, the length drifts, and the analyst’s job is to measure it and project the next trough. A wave repeats in form. Elliott’s observation was that markets advance in five waves and correct in three, and that this five-three structure is self-similar at every degree, from the intraday wiggle to the multi-decade advance. From an analytical perspective the two schools operate in different domains: cycle analysis lives in the frequency domain and asks when, wave analysis lives in the pattern domain and asks where in the structure you stand. Both are recurrence claims about the same tape, made along different dimensions, which is precisely why the comparison is worth an hour of conversation rather than a dismissal from either side.
Teaching the machine to count
The standard objection to wave analysis has always been the analyst: ask five Ellioticians for a count and you receive five counts. Prechter’s answer was not to train a neural network on past charts but to build symbolic AI, encoding the wave model’s own rules and guidelines into software. EWAVES generates candidate counts and grades every wave with an elliotticity score, an absolute measure from 0 to 100 percent of how well the structure conforms to the model, and its pattern recognition is scale-invariant by construction, mirroring the fractal claim at the heart of the theory.
Two results from this work deserve attention beyond the wave community. First, Prechter and the mathematician Sacha Sardo-Infirri published research showing the model can distinguish real market returns from randomized versions of the same data across the Dow, the Nasdaq, gold and Bitcoin. That is a falsifiable test, and wave analysis spent decades being accused of not offering any. Second, his stability argument: a form-based model is stationary across all of market history, while parameter-fitted quantitative models morph with each regime and burn out when the regime changes. Whatever one thinks of wave counts, that is a serious methodological position.
Waves out of cycles
The bridge between the schools is older than most practitioners on either side know. J.M. Hurst made the connection in 1970, arguing that summed cycles of different lengths generate exactly the wave-like structures Elliott catalogued: a long rising component with a shorter cycle riding on it produces five waves up and three down without any pattern being imposed. My own research extends the point beyond Elliott counts. Overlay the right composite of cycles and almost any classical formation falls out, head-and-shoulders included. The patterns are not the cause of anything; they are what cycle summation looks like on a chart.
As a result I treat the two models as complementary projections of the same object rather than rivals. The cycle model explains where the form comes from and supplies the clock, since a dominant cycle can be measured and its next trough projected. The wave model reads the form directly in the here and now, without measuring a single period, and tells you where inside the structure the market stands. In practice the combination is stronger than either input: a completed five-wave decline arriving in the time window of a projected cycle trough is the kind of agreement that separates a setup from a statistic, the same conditions-first logic Larry Williams argued two episodes ago from an entirely different toolbox.
The full conversation is on the FSC channel:
Elliott Prechter: Teaching a Machine to Count Elliott Waves
EWAVES: https://ewaves.com
Elliott Wave International: https://www.elliottwave.com
Foundation for the Study of Cycles: https://cycles.org