How Long Do Market Cycles Last? Typical Durations by Cycle Type (2026)

Market cycles range from 10-day short-term oscillations to 18-year secular cycles, with the most actionable intermediate cycles in the 20-week to 54-month band. In J.M. Hurst's nominal hierarchy, validated by modern spectral analysis, the dominant detectable periods are 10-day, 20-day, 40-day, 80-day, 20-week, 40-week, 18-month (84 weeks), 54-month (4.5 years), 9.2-year, and 18-year. This guide tabulates typical cycle durations by category, explains why they vary, and shows how to measure the dominant cycle length in any market you analyse.

The Short Answer: Cycles Come in Many Durations

There is no single "market cycle length." Every liquid market contains multiple cycles running simultaneously across different timeframes. When someone asks how long a market cycle lasts, the accurate answer depends on which cycle you are observing. A day trader watching 30-minute bars and a pension manager watching monthly returns are both correct when they cite different durations, because both cycles are real.

Spectral analysis via the Goertzel algorithm exposes this multi-scale structure. Rather than fitting a single wave to price data, it decomposes the data into its constituent frequencies and reports which periods carry the most power. The result is a list of coexisting cycles, each with its own length, amplitude, and phase.

Short-Term Cycles (Days to Weeks)

These cycles operate on the trading horizon that swing traders and active investors monitor. They are the shortest periods in J.M. Hurst's nominal hierarchy and the most frequently detected cycles on daily bar data.

The tolerance ranges (for example, 8 to 12 days for the 10-day cycle) reflect Hurst's Principle of Variation: nominal cycles breathe. A 10-day cycle might measure 9 days in one instance, 11 days in the next. The period is alive, not mechanical, which is why static models that lock onto a fixed wavelength drift out of phase with real markets.

Intermediate Cycles (Weeks to Months)

Intermediate cycles are the sweet spot for position traders and quantitative analysts. They are long enough to capture significant price moves (Hurst's Principle of Proportionality states that amplitude scales with period) and short enough to provide multiple opportunities within a typical investment horizon.

FractalCycles analyses of the Dollar Index, S&P 500, Bitcoin, Gold, and Crude Oil have repeatedly detected the 84-week (18-month) nominal across currencies, equities, crypto, and commodities. This confirms Hurst's Principle of Commonality: the same cycle periods recur across asset classes.

Long-Term Cycles (Years to Decades)

Long-term cycles operate at horizons that matter to macro analysts, allocators, and policy makers. They require decades of monthly data to resolve and are the least commonly visualised but the most structurally important.

Kondratiev waves are included for context. They are visible in economic proxies (interest rates, commodity prices, productivity) rather than in equity price series directly, and their length exceeds the reliable detection window of spectral analysis on typical price data.

Why Cycle Length Varies

The tolerance bands in the tables above are not sloppiness. They reflect a core principle of cycle analysis: real cycles are dynamic, not static. Hurst called this the Principle of Variation. A nominal 18-month cycle might measure 81 weeks in one instance and 88 weeks in the next. The cycle compresses during high-volatility regimes and expands during quieter ones.

This is why FractalCycles rejects the static mathematical cycle approach. Rather than locking onto a fixed wavelength and projecting it forward indefinitely, the platform recomputes the spectrum as new data arrives. Each analysis reads the current state of the market fresh, so the reported cycle length reflects what is actually present, not what a rigid historical model says should be there.

Cycle length also varies by instrument. A 10-day cycle in a large-cap US stock may run closer to 9 days, while the same cycle in a volatile commodity might stretch to 12. These variations contain information. A cycle that is compressing often precedes a sharper reversal, and a cycle that is expanding often signals an extended trend.

How to Measure Cycle Length in Any Market

The workflow for detecting the dominant cycle length in any market follows a consistent pipeline:

1. Choose a timeframe appropriate to the cycle you want to detect. Short-term cycles require daily or intraday bars; intermediate cycles require weekly bars; long-term cycles require monthly bars.
2. Apply detrending to remove the overall price drift so oscillations stand out.
3. Run the Goertzel algorithm across candidate periods to produce a power spectrum.
4. Apply the Bartels test to each detected cycle. Only cycles passing 70% or higher warrant attention.
5. Read the dominant cycle from the spectrum table. The period with the highest power that also passes Bartels testing is the dominant cycle length.

This is the same workflow FractalCycles automates for every analysis. You can run it against any symbol (stocks, crypto, forex, commodities, economic series) and get an objective readout of which cycles are active and how long they are, rather than guessing from visual pattern recognition.

Interpreting the Dominant Cycle

Once you have measured a cycle length, the next question is what to do with it. The cycle length tells you the period. To use it, you also need the phase (where you are within the cycle) and the regime (whether cycles are currently producing directional moves or choppy ranges).

Phase is reported on every FractalCycles analysis as Rising, Peaking, Falling, or Bottoming. Regime is captured by the Hurst exponent. Combined, these three readings (period, phase, regime) give you a complete structural picture of where a market is and how its cycles are likely to manifest in price action.

For deeper context on how these pieces fit together, see our guides on what a market cycle is, the four phases of every cycle, and how to read the current phase of a detected cycle.