ML Portfolio

Production-grade machine learning systems for cryptocurrency trading and research. 27 modules covering reinforcement learning, quantitative trading, NLP, and core ML infrastructure.

---

Reinforcement Learning

Module	Description
dqn	Rainbow DQN with prioritized replay, dueling networks, noisy nets
ppo	Proximal Policy Optimization for continuous trading actions
hierarchical-rl	Options framework with temporal abstraction for multi-timeframe strategies
curiosity-exploration	ICM-based curiosity-driven exploration for novel market regimes
meta-learning	MAML/Reptile for fast adaptation to new trading pairs

Quantitative Trading

Module	Description
order-flow-detection	Real-time order flow imbalance detection and whale tracking
harmonic-patterns	Gartley, Butterfly, Bat, Crab pattern detection with ML validation
elliott-wave	CNN-based Elliott Wave pattern recognition
volatility-forecasting	GARCH family models + ML-based realized volatility prediction
fear-greed-index	Multi-signal Fear & Greed Index for market sentiment
automl-pipeline	Automated feature engineering, model selection, and hyperparameter tuning

NLP & Sentiment

Module	Description
sentiment-engine	Multi-source sentiment aggregation (news, social, on-chain)
social-sentiment-analyzer	Twitter/Reddit/Discord sentiment with influence weighting
nlp-sentiment	Transformer-based crypto-specific sentiment classification

Core ML

Module	Description
anomaly-detection	Isolation Forest + Autoencoder ensemble for market anomalies
generative-models	GAN/VAE for synthetic trading data augmentation
graph-networks	GNN for crypto correlation and dependency modeling
attention-mechanisms	Multi-head attention for temporal pattern recognition
bayesian-networks	Uncertainty-aware predictions with calibrated confidence
continual-learning	EWC/PackNet for non-stationary market adaptation
explainable-ai	SHAP/LIME integration for trade decision explanations
model-compression	Quantization, pruning, distillation for low-latency inference

Infrastructure

Module	Description
gym-environments	OpenAI Gym environments for crypto trading simulation
testing	ML-specific testing: data drift, model regression, A/B validation
common	Shared utilities, data loaders, feature store connectors
xgboost	Gradient boosting baselines and feature importance analysis

Other

Module	Description
prophet-forecasting	Facebook Prophet for price and volume time series forecasting

---

Quick Start

git clone https://github.com/KeepALifeUS/ml-showcase.git
cd ml-showcase

# Install a specific module
cd reinforcement-learning/dqn
pip install -r requirements.txt
python train.py

Architecture

ml-showcase/
├── reinforcement-learning/   # RL agents for trading
├── trading/                  # Quantitative analysis modules
├── nlp/                      # Sentiment analysis pipeline
├── core/                     # Foundational ML components
├── infra/                    # Testing, utilities, environments
└── prophet-forecasting/      # Time series forecasting

Each module is self-contained with its own dependencies and tests.

License

MIT