T
Traderoid

Wavelet Transform

Time-frequency multi-scale price decomposition

IndicatorSignal ProcessingCyclicalAdvanced

Overview

Wavelet Transform decomposes price series into time-frequency components simultaneously, revealing cyclical patterns across multiple scales (timeframes) at once. Unlike FFT which loses timing info, wavelets preserve when frequencies occur. Morlet or Mexican Hat wavelets stretch/compress across scales = continuum frequency analysis. Reveals dominant market cycles (4-day, 10-day, 25-day, etc.) with timing precision = superior to fixed-period oscillators.

Wavelet power spectrum shows intensity of each frequency at each time point. High power = dominant cycle active. Watch for power concentration at 2-3 specific periods = mechanical trading signal. When dominant cycle shifts abruptly (cycle death), regime change = trend/consolidation transition. Requires statistical signal processing knowledge but delivers unique edge unavailable from standard indicators.

Professional quantitative traders use wavelets for cycle timing, regime detection, and optimal entry/exit windows. More sophisticated than moving averages; far superior to fixed-period oscillators for adaptive trading.

Formula

Wavelet = ψ(t) = Morlet or Mexican Hat basis function
Scaled Wavelet = ψ((t-b)/a) with scale 'a', position 'b'
WT(a,b) = Σ price(t) × ψ((t-b)/a) = continuous wavelet transform
Power(a,t) = |WT(a,t)|² = energy intensity at scale & time
Scale-to-Period: Period = 2π×a/frequency = convert scales to cycles
Wavelet power maps frequency strength over time; concentration = dominant cycle active.

Parameters

ParameterTypeDefaultDescription
Wavelet TypeSelectionMorletMorlet for sharp peaks, Mexican Hat for broad cycles
Frequency RangePeriod Bars4-128 barsMin/max cycle lengths to analyze
Power ThresholdPercent75th percentileHighlight only significant power concentrations

Common Use Cases

1. Dominant Cycle Detection

Identify current market cycle (4/10/25-day periods). High concentration = mechanical signal for period-matching trades.

2. Regime Change Detection

Power shifts between scales = cycle death/birth. Sudden change = regime switch from trending to choppy.

3. Entry/Exit Timing

Peak power = optimal entry window within dominant cycle. Trough = optimal exit. Mechanical timing.

4. Confluence Across Cycles

When 2+ dominant cycles align in phase = strongest signals. Multi-scale power concentration.

Advantages & Limitations

Advantages

  • Multi-Scale Analysis: Simultaneous view of all frequencies; reveals relationships FFT misses.
  • Time-Frequency Precision: Know WHEN frequencies are active; superior to Fourier.
  • Adaptive Cycles: Markets naturally cyclical; wavelets reveal true dominant period.
  • Quantitative Edge: Rare edge; most traders unaware; automation possible.

! Limitations

  • Computationally Intensive: CWT requires O(n²) calculations; slow on long histories.
  • Interpretation Difficulty: Power maps complex to read; requires DSP knowledge.
  • Edge Artifacts: Boundary effects at chart edges reduce reliability on recent bars.
  • Lookback Dependency: Cycle detection lags; must wait ~2 periods for confirmation.

Tips & Best Practices

📊 Focus on Concentration

Watch power map for tight clustering at 1-2 periods. Broad/scattered = no dominant cycle = avoid trading.

⚡ Ignore Edge Artifacts

Recent bars (last ~10% of lookback) unreliable due to boundary effects. Trade only stable cycles from older bars.

🎯 Use on Daily+ Charts

Wavelets work daily/weekly; intraday cycles too noisy. 4-128 bar range = 4-day to 25-week cycles on daily.

⚠️ Confirm with Price

Wavelet signal not standalone. Confirm cycle prediction with price structure (FVG, Order Block).

Example Strategy

1. Scan Power Map

Daily chart: plot wavelet power (4-128 bars). Identify tight concentration = dominant cycle present.

2. Measure Cycle Period

When power concentrates at specific period (e.g., 10 bars), note it. Market currently cycling at that frequency.

3. Time Wave Projection

Assume price oscillates at detected period. Project next ~(period/2) bars = expected trend reversal. Plan entry.

4. Exit at Cycle Peak/Trough

When price reaches projected peak/trough (cycle peak), exit. Stop loss beyond prior cycle extreme point.

Related Nodes