Sentiment + Technical Analysis. Human-in-the-loop.
Institutional-grade AI research desk delivering BUY/SELL/HOLD signals with full transparency.
- See It in Action
- Project Overview
- Installation
- Configuration
- Algorithm & Functionality
- Build Custom AI Trading Agents
- Collaborate
- Roadmap
- Quick Start
- Project Structure
- Deployment
- Technical Stack
- Contact
- License
Pick a ticker → Analyze → Scout, Analyst, Risk, Signal. Full pipeline, every step visible.
python3 main.py BTC-USD
AuraIntelligence combines narrative sentiment from 50+ sources with industry-standard technical analysis to produce BUY/SELL/HOLD signals with confidence (45–92%). Built for manual traders. No auto-trading — you decide, the AI informs.
git clone https://github.com/acenjo/Aura-AI-Trading-Agent.git
cd Aura-AI-Trading-Agent
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
pip install -r requirements.txtNote: pandas-ta is commented out by default (analyst has built-in indicators). Uncomment in requirements.txt if you want to use it.
| Variable | Required | Purpose |
|---|---|---|
CHUTES_API_KEY |
No | LLM sentiment via Chutes AI. Without it: rule-based sentiment (works offline). |
Copy .env.example to .env and add your key if using LLM mode. Price data uses yfinance (free, no API key).
50 posts per ticker from X, Reddit, StockTwits, Bloomberg, Reuters, CNBC, Seeking Alpha.
- Source tiers: Tier 1 (1.35×), Tier 2 (1.15×), Tier 3 (0.95×)
- Recency: Newer posts weighted higher
- Dual mode: LLM (60%) + rule-based (40%), or pure rule-based offline
- Output: Score -1 to +1, summary, front-run flag
- Indicators: Wilder RSI(14), EMA 9/21, MACD(12,26,9), Bollinger(20,2), volume
- Weights: RSI 35%, EMA 25%, MACD 25%, Bollinger 10%, Volume 5%
- Fusion: 35% sentiment + 65% technical → combined score
- Confluence: STRONG signals require 3+ aligned indicators
- Output: STRONG BUY / BUY / HOLD / SELL / STRONG SELL
- Position size: Scaled by signal and sentiment. Stop-loss / take-profit configurable.
- Output: Size %, dollar position, SL %, TP %. Human confirmation before execution.
- Confidence: 45–92% from signal type, sentiment alignment, RSI extremes, signal strength
- Reasoning chain: Full trace stored in SQLite for audit.
Every AI trading agent is built around the trader's requirements. No generic templates. Your strategy, your instruments, your edge — encoded into agents that fit your workflow.
| Requirement | How We Build It |
|---|---|
| Indicators | Add VWAP, ATR, custom oscillators. Reweight RSI/MACD/Bollinger to your preference. |
| Data sources | Integrate your feeds, APIs, providers. Prioritize sources that match your edge. |
| Risk logic | Your position sizing, drawdown limits, portfolio constraints. |
| Instruments | Stocks, crypto, forex, commodities. Your universe, your config. |
| Timeframes | 1m, 5m, 1H, 4H, daily. Multi-timeframe when needed. |
| Strategy logic | Your rules, your confluence, your thresholds — built into Scout, Analyst, Risk. |
Active development cycle: Requirements → design → build → deploy → iterate. Your feedback drives each iteration. Custom AI trading agents, built actively from your specifications.
Collaboration is central. Traders, analysts, and builders shape AuraIntelligence. I lead toward disciplined, data-driven execution — not just tools, but direction and methodology that scale with your edge.
| You are | Engagement |
|---|---|
| Trader | Transparent signals, risk sizing, workflows. Custom agents built to your requirements. |
| Analyst | Validate methodology, co-design logic, suggest indicators and sources. |
| Builder | Custom agents, white-label, partnership. Your specifications, delivered. |
Your input directly influences priorities. Feature requests, use cases, and collaboration inquiries drive the roadmap. Reach out to collaborate — custom AI trading agents are built actively with traders.
Long-term vision. Phased execution. Your input shapes priorities.
| Phase | Focus | Status |
|---|---|---|
| Phase 1 | Core pipeline, web dashboard, CLI | ✅ Done |
| Phase 2 | Multi-timeframe, more data sources | 🚧 In progress |
| Phase 3 | Alerts, backtesting, signal backfill | 📋 Planned |
| Phase 4 | Collaboration layer — share signals, peer validation | 📋 Planned |
| Phase 5 | Custom agent builder — define your own logic | 🔮 Vision |
| Phase 6 | API, broker integrations, Telegram/Discord bots | 🔮 Vision |
| Phase 7 | Institutional — multi-strategy, teams, audit | 🔮 Vision |
| Phase 8 | Community marketplace — configs, indicators, workflows | 🔮 Vision |
# Web Dashboard (port 8501)
streamlit run interface/dashboard.py
# Open http://localhost:8501
# CLI
python3 main.py BTC-USD
python3 main.py AAPL
python3 main.py ETH-USDAura-AI-Trading-Agent/
├── agents/
│ ├── scout_agent.py # Sentiment (50 feeds, LLM + rule-based)
│ ├── analyst_agent.py # RSI, EMA, MACD, Bollinger Bands
│ └── risk_agent.py # Position sizing, stop-loss, take-profit
├── core/
│ ├── engine.py # Pipeline: Scout → Analyst → Risk → Aggregate
│ └── database.py # SQLite (signals table)
├── interface/
│ └── dashboard.py # Streamlit web UI
├── media/ # Demo GIFs (dashboard-demo.gif, cli-demo.gif)
├── main.py # CLI entry point
├── run_streamlit.sh # Start dashboard on port 8501
├── requirements.txt
└── .env.example
Ubuntu / Linux VPS:
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
./run_streamlit.shOpen http://YOUR_SERVER_IP:8501
Python 3.9+, Streamlit, yfinance, Plotly, SQLAlchemy, Chutes AI (optional). Rule-based mode works offline.
Telegram: @AuraTerminal
Custom AI trading agents built actively from your requirements. Collaboration, customization, and support — reach out to discuss your needs.
Proprietary. All rights reserved.
AuraIntelligence — AI Trading Intelligence for Manual Traders
📱 Contact: @AuraTerminal
Collaborate. Customize. Build AI trading agents together.