NEXUS — 7-Layer ML Crypto Signal Engine

NEXUS is a multi-layered machine-learning framework for cryptocurrency market analysis and signal generation. It combines regime detection, deep learning, and multi-timeframe analysis to produce high-confidence trading signals.

Architecture — 7 Signal Layers

#	Layer	Description
1	Regime Detection	Market regime classification (trending / ranging / volatile)
2	Momentum Analysis	Multi-timeframe momentum scoring with EMA/RSI confluence
3	Volume Profile	Order flow and volume distribution analysis
4	ML Classifier	PyTorch neural network for pattern classification
5	AltFins Integration	External signal enrichment from AltFins API
6	Sentiment Layer	On-chain and social sentiment aggregation
7	Scoring Engine	Weighted ensemble of all layers → final signal score

Features

• PyTorch deep learning — custom neural network for price-pattern recognition
• Binance API integration — real-time and historical OHLCV data ingestion
• AltFins signal enrichment — third-party signal validation layer
• Backtesting suite — historical signal validation across multiple market regimes
• Regime awareness — signal weights adapt to current market conditions
• Multi-asset scanning — parallel scanning across 50+ cryptocurrency pairs
• Output formats — JSON signals, CSV exports, visualization charts

Tech Stack

Component	Technology
Deep Learning	PyTorch
Data Processing	pandas, numpy
Exchange Data	Binance API (python-binance)
Technical Analysis	ta-lib, pandas-ta
Visualization	matplotlib, plotly
Config Management	Python dataclasses
Testing	pytest

Project Structure

nexus/
├── config.py        # Configuration (API keys via environment variables)
├── analyze.py       # Main analysis orchestrator
├── scanner/         # Multi-asset market scanner
├── layers/          # 7 signal layers (each in separate module)
├── ml/              # PyTorch model definitions and training
├── models/          # Saved model checkpoints
├── scoring/         # Ensemble scoring and signal generation
├── data/            # Data fetching and caching
└── output/          # Signal exports and charts

Setup

git clone https://github.com/aria-tjr/111.git nexus
cd nexus
pip install -r requirements.txt

export BINANCE_API_KEY="your_binance_api_key"
export BINANCE_SECRET="your_binance_secret"
export ALTFINS_API_KEY="your_altfins_key"   # optional

python analyze.py

All API keys are loaded from environment variables. See config.py for the full configuration schema. Never hardcode credentials.

Signal Output

Each generated signal includes:

• Asset pair (e.g. BTC/USDT)
• Direction (LONG / SHORT / NEUTRAL)
• Confidence score (0–100)
• Contributing layer scores
• Recommended entry zone
• Invalidation level

Disclaimer

This tool is provided for research and educational purposes. Cryptocurrency markets are highly volatile; signals are not financial advice. Never trade more than you can afford to lose.

License

MIT