Module features 

ML Feature Engineering Toolkit (ta.features.*)

This module provides rich, multi-dimensional feature extractors built on top of QuantWave’s existing high-quality indicators (especially Ehlers DSP and Regimes). The goal is to make it trivial to build stable, no-lookahead feature matrices for ML pipelines and strategy research.

All extractors follow the Universal Indicator pattern where possible:

• Implement Next<Input> for streaming use
• Provide equivalent batch (Polars) paths in quantwave-polars
• Must eventually prove batch == streaming via proptests (see task quantwave-tha)

Design principles (from quantwave-4ub research notes):

• Rich outputs (structs or tuples) over single scalars when useful
• Strong metadata (reuse/extend IndicatorMetadata)
• Easy composition with regimes and (future) PA events
• Zero lookahead by construction

Sources recorded (per AGENTS.md):

• cyber_cycle.rs:35 (returns (cycle, trigger))
• hurst.rs (persistence value, excellent regime feature)
• regimes/mod.rs + 12 submodules (HMM probs, GMM, PELT, etc. as meta-features)
• Ehlers papers in references/Ehlers Papers/implemented/
• Prado “Advances in Financial Machine Learning” (for future fractional differencing / entropy)

Re-exports

pub use cyber_cycle::CyberCycleFeatureExtractor;

pub use cyber_cycle::CyberCycleFeatures;

pub use ehlers_autocorrelation::EhlersAutocorrelationFeatureExtractor;

pub use ehlers_autocorrelation::EhlersAutocorrelationFeatures;

pub use griffiths_dominant_cycle::GriffithsDominantCycleFeatureExtractor;

pub use griffiths_dominant_cycle::GriffithsDominantCycleFeatures;

pub use hurst::HurstFeatureExtractor;

pub use hurst::HurstFeatures;

pub use instantaneous_trendline::InstantaneousTrendlineFeatureExtractor;

pub use instantaneous_trendline::InstantaneousTrendlineFeatures;

pub use regime::regime_to_features;

pub use regime::RegimeFeatures;

pub use regime_probs::regime_to_prob_features;

pub use regime_probs::RegimeProbFeatures;

pub use trendflex::TrendflexFeatureExtractor;

pub use trendflex::TrendflexFeatures;

Modules

cyber_cycle

Cyber Cycle feature extractor wrapper.

ehlers_autocorrelation

Ehlers Autocorrelation feature extractor wrapper.

griffiths_dominant_cycle

Griffiths Dominant Cycle feature extractor wrapper.

hurst

Hurst Exponent feature extractor wrapper.

instantaneous_trendline

Instantaneous Trendline feature extractor wrapper.

regime

Basic regime feature helpers for ML pipelines.

regime_probs

Regime probability features (soft labels) for ML pipelines.

trendflex

Trendflex feature extractor wrapper.

Traits

AsFeatures

Common trait for feature extractors that want to expose a stable “feature vector” view. (Future expansion point for a unified FeatureVector trait.)