Wavelets

[thestephenromano]

I've had great success in the past using Wavelet analysis to inform trading decisions using both automated and non-automated indicators and strategies. Anyone here have a similar experience, and if so do you know of any MotiveWave specific studies or add ons (happy to pay)? I would love to incorporate Wavelets back into my toolbox in a solution that is all under one roof.

 

[thestephenromano]

For anyone wondering, this is a high-level from Claude AI:

What are Wavelets?

Wavelets are mathematical functions that decompose time series data into different frequency components, each with a resolution matched to its
scale. Unlike Fourier transforms, wavelets are localized in both time and frequency, making them ideal for analyzing non-stationary financial
data.

Key Trading Applications:

1. Multi-Resolution Analysis
- Decomposes price into different time scales (D1-D7 in your code)
- D1: High-frequency noise (shortest cycles)
- D7: Low-frequency trends (longest cycles)
- Each level represents roughly double the period of the previous
2. Noise Filtering
- Separates market noise from underlying trends
- Higher decomposition levels (D5-D7) reveal cleaner trend signals
- Lower levels (D1-D3) often contain market microstructure noise
3. Cycle Detection
- Identifies dominant market cycles at different timeframes
- Helps traders align with the prevailing rhythm of the market
- Your auto-scaling targets ~390 minute sessions (typical trading day)
4. Support/Resistance Levels
- Wavelet coefficients can identify price levels where trends change
- Zero-crossings in detail coefficients often coincide with turning points
5. Volatility Analysis
- Energy at different scales indicates volatility regimes
- Sudden changes in wavelet coefficients signal regime shifts

 

[T_Tech]

I could probably build something but I'm not very familiar with wavelets, do you have an example indicator from other platforms?

 

[thestephenromano]

I could probably build something but I'm not very familiar with wavelets, do you have an example indicator from other platforms?

Appreciate that but I've actually started doing this myself. Here's the first pass/ proof of concept. Price is decomposed into 7 levels here, notice the "noise" in April as expressed in the High Frequency levels (D0, D1) and then the more dominant trend signals in the longer cycles.

 

[thestephenromano]

Trick is to intelligently remove the "noise" from the underlying signal in order to construct a cleaner path.

 

[T_Tech]

Trick is to intelligently remove the "noise" from the underlying signal in order to construct a cleaner path.

Super interesting, thanks for sharing!