Spectral Cycles vs Bollinger Bands

Bollinger Bands, developed by John Bollinger in the 1980s, consist of a moving average with bands set at standard deviation intervals above and below. The bands expand during volatile periods and contract during quiet periods, providing a dynamic envelope around price. Spectral cycle analysis identifies recurring oscillations through frequency decomposition. While both methods analyze price structure, they measure fundamentally different properties and serve different analytical purposes.

How Bollinger Bands Work

The standard Bollinger Band construction uses:

• Middle Band: 20-period simple moving average
• Upper Band: Middle band plus 2 standard deviations
• Lower Band: Middle band minus 2 standard deviations

The bands capture approximately 95% of price action (assuming normal distribution). When price touches or exceeds a band, it is considered extended relative to recent volatility.

Volatility Versus Structure

This is the fundamental distinction between the methods:

Bollinger Bands measure volatility—how far price has strayed from its recent average. Band width indicates current volatility level; band touches indicate extension relative to that volatility.

Spectral analysis measures structure—what periodic oscillations exist in price behavior. Cycle detection identifies the rhythms underlying price movement, regardless of current volatility.

A market can be highly volatile (wide bands) while exhibiting strong cyclical structure, or it can be low volatility (narrow bands) with weak cyclical structure. The measurements are orthogonal.

Mean Reversion Assumptions

Bollinger Bands implicitly assume mean reversion. When price touches the upper band, it is "extended" and expected to return toward the middle. This assumption works during ranging markets but fails during trends.

Spectral analysis makes no mean reversion assumption. It identifies cycles and their current phases. A cycle approaching its peak suggests a coming decline, but not because price is "extended"—rather because that is the structural rhythm.

The Hurst exponent from spectral analysis tells you whether mean reversion assumptions are valid. Low Hurst (below 0.5) supports mean reversion logic; high Hurst undermines it.

The Band Squeeze

One of Bollinger Bands' most valuable signals is the squeeze—a contraction of the bands indicating low volatility that often precedes significant price movement. This aligns with the principle that compression precedes expansion.

Spectral analysis approaches compression differently. Rather than measuring band width, it examines whether multiple cycles are converging toward the same phase—a condition that can produce amplified moves when cycles align.

The squeeze is a volatility phenomenon; cycle convergence is a structural phenomenon. Both can precede significant moves, but for different underlying reasons.

Period Selection

Bollinger Bands use a fixed 20-period lookback by default. This period was chosen somewhat arbitrarily—it represents approximately one trading month. Different instruments and timeframes may have different optimal periods.

Spectral analysis discovers the significant periods rather than assuming them. If a market exhibits a dominant 35-bar cycle, spectral analysis detects it. Bollinger Bands with a 20-period setting would not align with this cycle structure.

This suggests an adaptation: set Bollinger Band periods to match spectrally detected cycles. A 35-bar dominant cycle might call for 35-period bands rather than 20.

Statistical Properties

Bollinger Bands assume price returns are approximately normally distributed. The 2-standard-deviation bands should contain 95% of observations under normality.

Market returns are not normally distributed—they exhibit fat tails and kurtosis. This means band touches occur more frequently than the normal assumption predicts. Upper and lower band touches may not be as "extreme" as the 95% confidence interval suggests.

Spectral analysis makes no distributional assumptions. The Bartels test measures phase consistency; Monte Carlo simulations make no parametric assumptions about return distributions.

Strengths of Bollinger Bands

• Visual clarity: Clear price envelope showing recent range
• Volatility measurement: Band width directly shows volatility
• Squeeze detection: Identifies low-volatility compression
• Universal availability: Built into every charting platform
• Simplicity: Easy to understand and apply

Limitations of Bollinger Bands

• Mean reversion assumption: Fails during trends
• Fixed period: May not match market structure
• No cycle awareness: Does not identify underlying rhythms
• Distribution assumptions: Normal distribution rarely holds
• Lagging indicator: Based on moving average

Complementary Usage

These methods can be combined effectively:

1. Use spectral analysis to identify dominant cycle periods
2. Set Bollinger Band period to match the dominant cycle
3. Use Hurst exponent to determine if mean reversion is valid
4. When Hurst is low, trust band touches as reversal signals
5. When Hurst is high, treat band touches as breakout continuation signals
6. Monitor cycle phase to contextualize band touches

This framework uses spectral analysis to calibrate and contextualize Bollinger Band signals.

Bandwidth Versus Cycle Power

An interesting parallel exists between Bollinger bandwidth and spectral cycle power:

• Narrow bands (low bandwidth) often coincide with low cycle amplitude
• Wide bands (high bandwidth) often coincide with high cycle amplitude
• Expanding bandwidth may indicate cycle entering its high-power phase
• Contracting bandwidth may indicate cycle weakening

Monitoring both provides a more complete picture of market structure evolution.

Conclusion

Bollinger Bands and spectral analysis answer different questions. Bollinger Bands measure volatility and extension relative to recent behavior. Spectral analysis identifies the underlying oscillation structure generating that behavior.

For volatility assessment and squeeze detection, Bollinger Bands remain valuable. For understanding timing structure and cycle phase, spectral analysis is superior. The most complete analysis uses both: spectral methods to understand structure, Bollinger Bands to measure current volatility within that structure.