CYCLES · method · Updated 2026-11-05 · Derek William Frazier

What Is the Principle of Variation in Cycle Analysis?

Variation is the rule that no real cycle ever runs exactly its nominal length — the 80-day cycle averages about 68 calendar days trough to trough, but any single real instance of it can run anywhere from roughly 60 to 80 days. That spread isn't an error in the model. Hurst named it as its own principle because the drift carries information, not just noise.

PLATE · THE RULER HAS TOLERANCE BUILT INEvery real cycle runs short or long of its nominal average — never exactly on it.two real 80-day cycles, same model, different lengthsshort cycle · 62 daysran bullishlong cycle · 78 daysran bearishnominal · 686080days, trough to troughThe drift isn't noise to ignore. Short cycles here ran bullish, long ones bearish.CYCLICAL MARKETS · THE CYCLITECNICAL METHODEducational. Not advice. No performance promise.

Why does a "nominal" length even have a range built in?

Because the nominal lengths on the cycle ladder — 80-day, 40-day, 20-day, 10-day — are measured averages across many years of data, not fixed periods like a calendar month. Real markets don't run on a metronome. Every actual trough-to-trough span wobbles around its average: sometimes a little short, sometimes a little long, occasionally well outside the usual band. Hurst's answer to that wobble wasn't to tighten the model until it stopped happening — it was to name the wobble itself as a principle, because pretending cycles run on rails would have made the method brittle the first time real price didn't cooperate.

The practical range for the 80-day rung, in this method's own use, runs about 60 to 80 calendar days. That's not "anything goes" — a count that implies 45 days or 110 days isn't a long 80-day cycle, it's the wrong trough. But inside that 60–80 band, the model expects drift as the normal case, not the exception.

Does the drift itself mean anything, or is it just slop to tolerate?

It means something, and this is the part that turns variation from an excuse into a read. Within that 60-to-80-day range, where a given cycle actually lands has tended to carry a directional lean: 80-day cycles that finish short of the roughly 68-day average have tended to run bullish, and the long ones — out toward 75 to 80 days — have tended to run bearish. The eight FLD interactions that structure a single 80-day span can compress or stretch with the same tendency; a sharp decline that completes a major cycle projection has tended to be followed by contracted, fast-running shorter cycles underneath it.

So the day count isn't just bookkeeping you do after the fact to confirm a trough. It's part of the read itself — a cycle that's running short is telling you something different than one that's running long, on top of whatever direction the FLD crosses and the translation are already telling you.

Variation shows up at every level of the ladder, not just the 80-day rung. The same logic that makes a short 80-day cycle lean bullish applies to a short 20-day cycle inside it — and because the rungs sum, a cluster of rungs all running short or all running long at the same time compounds the lean rather than canceling it out.

How is this different from just saying "the model is approximate"?

Every real-world model is approximate somewhere — that alone isn't a principle worth naming. What makes variation specific is that it tells you where to expect the slop (around the nominal average, inside a bounded range) and what the slop means when it shows up (short leans one way, long leans the other). A vague "approximate" wouldn't give you either of those. Variation gives you a tolerance band with real information inside it, which is a sharper claim than "cycles aren't exact."

It also sets a boundary the other direction. Variation covers the normal 60-to-80-day wobble. It does not cover a trough count that's off by weeks — that's not a long cycle, it's a miscounted one, and the fix is to recount from a different candidate low, not to stretch the definition of "80-day" until it fits.

What does this actually look like when you're counting a live cycle?

You're partway through what you believe is the current 80-day cycle, and day 68 — the average — comes and goes with no trough in sight. Without the principle of variation, that's alarming: the model said 68, and 68 passed. With it, day 68 is just the midpoint of a normal range, and the real question is whether price is still inside the 60-to-80 window, not whether it missed an appointment.

If the trough lands at day 61, you've met a short cycle — read it with the bullish lean the pattern has carried. If it stretches to day 79, you've met a long one — read it with the bearish lean. If day 85 comes and goes with still no trough, that's the signal something is wrong with the count itself, not that this particular cycle is an unusually extreme but valid 80-day instance.

Day the trough actually formsWhere that sits in the modelWhat it suggests
~60–65Short of the ~68-day averageHas tended to lean bullish
~66–70Close to the averageNo strong lean either way
~71–80Long of the averageHas tended to lean bearish
85+ with no trough yetOutside the normal rangeLikely a miscounted cycle — recount from a different low

What variation does not tell you

Questions traders ask

If cycles vary this much, how is "80-day" a useful name at all?

Because the range is bounded and the average is real. A cycle that's "around 68 days, usually somewhere between 60 and 80" is a far more specific, testable claim than no claim at all — it tells you when to start watching for a trough, roughly how far the window extends, and what a trough's position inside that window has tended to mean. The name survives the variation; it just isn't a metronome.

Is a cycle that lands exactly at 68 days the "best" or most reliable one?

Not especially — it's simply the average case, with no particular bullish or bearish lean attached either way. The informative cases are the ones that land toward either edge of the range, because those are the ones where the drift itself is saying something. A dead-average cycle is the least informative outcome, not the most trustworthy one.

How does variation interact with a confluent low?

Confluence asks whether several rungs are bottoming together; variation asks how far any one of those rungs has drifted from its own average on its way to that shared trough. The two ideas combine rather than compete: a cluster of rungs that are all running short of their averages and bottoming together carries a stronger bullish lean than the same confluence with rungs running at or past their averages — the alignment and the drift are both part of the read, not substitutes for each other.

This is a concept page, not a signal — variation describes the shape of the slop built into every count, and nothing here replaces logging an actual call with an invalidation level. How every read on this site gets scored, hit or miss, is covered on the methodology page.

Keep reading

What Is the 80-Day Cycle in the Stock Market?What Is a Future Line of Demarcation (FLD)?The Nominal Cycle Ladder: 80, 40, 20, 10 Days, ExplainedWhy Do Market Bottoms Form Faster Than Tops?
Get our AI tools and everything we send out New AI trading tools and our Claude Code builds, plus session invites. All free.
Get them free