How to Apply Hurst Cycle Analysis to Any Market

You understand the theory of Hurst cycles. Now you want to apply it to real markets. This guide walks through the practical process of performing Hurst cycle analysis — step by step, from raw price data to a composite cycle projection you can use for trading decisions.

The process works on any liquid market: stocks, ETFs, indices, cryptocurrencies, forex, and commodities. The mathematics does not care what asset you are analyzing — it detects oscillations in any time series.

Step 1: Prepare Your Data

Good analysis starts with good data. For Hurst cycle analysis, you need:

• Sufficient history — A minimum of 10 repetitions of the longest cycle you want to detect. If you are looking for a 40-week cycle, you need at least 400 weeks of data. For daily data analyzing cycles up to 80 days, roughly 800 daily bars (about 3 years) provides adequate coverage.
• Consistent timeframe — Daily bars are the most common starting point. Weekly bars work well for longer-term analysis. Intraday bars are possible but introduce more noise.
• Clean data — Gaps, errors, and missing bars can distort spectral analysis. Use adjusted close prices to account for dividends and splits.

FractalCycles can fetch data automatically from Yahoo Finance for most stocks and cryptocurrencies, or you can upload CSV files for any custom data set.

Step 2: Check the Market Regime

Before searching for cycles, assess the market regime using the Hurst exponent. This tells you what kind of behavior dominates the current data:

• H < 0.45 — Anti-persistent (mean-reverting). The market is actively oscillating, and cycle analysis is likely to be most productive. Mean reversion strategies tend to work well in this regime.
• H between 0.45 and 0.55 — Random walk territory. Cycles may be present but weak. Proceed with cycle analysis but expect fewer validated cycles and lower confidence projections.
• H > 0.55 — Persistent (trending). A strong directional move is underway. Cycles are still present but the trend dominates. Cycle analysis can identify when the trend is likely to pause or reverse, but trend-following strategies may be more appropriate than cycle-based timing.

The Hurst exponent calculator provides this measurement in seconds. It is the single most efficient first step in any market analysis.

Step 3: Run Spectral Analysis

Spectral analysis transforms price data from the time domain into the frequency domain, revealing which periodicities carry the most energy. The result is a power spectrum — a chart showing power (strength) on the Y-axis and cycle period on the X-axis.

Peaks in the power spectrum correspond to dominant cycles. The Goertzel algorithm is used because it efficiently evaluates specific frequency ranges without computing the entire spectrum. This is both faster and more focused than a full Fast Fourier Transform.

When reviewing the power spectrum, look for:

• Clear peaks that rise significantly above the noise floor — these are candidate cycles
• Harmonic relationships — peaks at periods that relate by factors of 2 or 3 (e.g., 20 and 40, or 20 and 60) suggest nested cycles consistent with Hurst's nominal model
• Broad peaks vs. sharp peaks — sharp peaks indicate well-defined cycles; broad peaks suggest the cycle period varies over time

Step 4: Validate with the Bartels Test

This is the most important step and the one most analysts skip. Every time series produces peaks in its power spectrum — even random noise. The Bartels cyclicity test determines which peaks represent genuine cyclical behavior.

The Bartels test works by dividing the data into segments of the proposed cycle length and checking whether the phase (timing of peaks and troughs) is consistent across segments. If a 20-day cycle is real, each 20-day segment should show similar phase alignment. Random noise will not.

Typically, you want a Bartels test p-value below 0.05 (95% confidence) before considering a cycle validated. In practice, the strongest cycles often test at p-values below 0.01.

A raw spectral analysis might identify 10-15 candidate cycles. After Bartels testing, typically 3-5 survive. These survivors are your working set — the cycles with genuine statistical support.

Step 5: Build the Composite Waveform

Take your validated cycles and combine them into a composite waveform. This is simply the sum of the individual cycle sine waves, each with its measured amplitude and phase.

The composite waveform shows:

• Rising sections — Cyclical forces are pushing price upward. Multiple cycles rising simultaneously create the strongest bullish conditions.
• Falling sections — Cyclical forces favor downward pressure. Multiple cycles falling simultaneously create the strongest bearish conditions.
• Composite troughs — Points where the composite waveform reaches a minimum. These indicate where multiple cycle troughs are expected to synchronize, creating high-probability reversal zones.
• Flat or choppy sections — Cycles are conflicting (some rising, some falling). These periods suggest range-bound conditions and higher uncertainty.

Step 6: Project Forward and Trade

The composite waveform can be extended beyond the current date, creating a projection of expected cyclical behavior. This projection is not a price target — it is a timing model that indicates when cyclical conditions favor bullish or bearish positioning.

Practical trading applications include:

• Timing entries — Look for buy opportunities near projected composite troughs, especially when the Hurst exponent confirms mean-reverting conditions
• Timing exits — Consider taking profits or tightening stops near projected composite peaks
• Position sizing — Increase exposure when multiple cycles align in the projected direction; reduce when cycles conflict
• Combining with other analysis — Use cycle projections as a timing overlay on fundamental or technical analysis you already perform

Example: Applying Hurst Cycle Analysis to the S&P 500

Consider a typical analysis of the S&P 500 using daily data over the past 3 years (approximately 750 bars):

1. Hurst exponent measures 0.42 — indicating anti-persistent, mean-reverting behavior. Cycle analysis should be productive.
2. Spectral analysis identifies peaks at approximately 20, 42, and 78 day periods.
3. Bartels testing confirms the 20-day and 42-day cycles at p < 0.05. The 78-day cycle tests at p = 0.08 — suggestive but not conclusive.
4. Composite built from the two validated cycles shows a projected trough in 8 trading days, with the next peak approximately 15 days later.

This type of analysis takes seconds with modern software. The same process that took Hurst weeks of manual computation now runs automatically. Try it yourself — run an analysis on any symbol through FractalCycles and compare the detected cycles to the price chart. The correspondence between cycle projections and actual turning points is often striking.