Intelligent Trading System v2: Cortex

Cortex is a next-generation automated trading system that combines high-frequency signal processing with a Retrieval-Augmented Generation (RAG) AI brain to make informed, context-aware trading decisions.

🚀 Key Features

1. AI "Brain" (RAG Architecture)

Unlike traditional bots that rely solely on math, Cortex has a Memory.

• ChromaDB: Stores past trading scenarios, market setups, and outcomes ("Playbooks").
• Ollama (LLM): Analyzes current ML signals against this historical context to detect hallucinations or weak setups.
• RAG Process:
  1. Retrieve: Finds similar past market conditions.
  2. Augment: Feeds current technicals + past outcomes to the AI.
  3. Generate: AI outputs a "Confidence Score" and reasoning (Buy/Sell/Hold).

2. Zero-Latency Dashboard

A modern, dark-mode UI built with Flask and Server-Sent Events (SSE).

• Real-Time Ticker: Live price updates without page reloads.
• Active Trade Management: visual stop-loss/take-profit tracking.
• System Health: Monitor AI inference latency and bridge connection status.

3. Intelligent Engine

• Hybrid Signal Generation: Combines classic ML (TensorFlow/Scikit-Learn) with "Intelligent Indicators".
• Safety First: "Live Guard" prevents execution if MT5 connection is unstable.
• Auto-Risk: Dynamic position sizing based on account equity and volatility (ATR).

---

🛠️ Architecture

graph TD
    A[MT5/Binance] -->|Live Data| B(Real-time Service)
    B --> C{ML Engine}
    C -->|Raw Signal| D[RAG Evaluator]
    D <-->|Query/Context| E[(ChromaDB Memory)]
    D <-->|Inference| F[Ollama AI]
    D -->|Validated Signal| G[Execution Engine]
    G --> A
    B --> H["Web Dashboard (Flask)"]
    G --> H

Loading

📦 Installation

1. Clone & Install Dependencies:

   pip install -r requirements.txt

2. Configure Credentials:

   • Copy config.example.json to config.json.
   • Enter your MT5 and Telegram credentials. (See SETUP_GUIDE.md for details).

3. Run the System:

   The system requires two separate terminals:

   Terminal 1: MT5 Bridge (Must run first!)

   cd mt5-bridge
   py main.py

   Terminal 2: Core System

   py run.py

   Access the UI at: http://127.0.0.1:5000

🧠 Training & Offline Mode

The system handles offline model training via the scripts/ module.

• python -m scripts.train -c config.json

🤝 Contributing

• Frontend: app/templates (HTML/Tailwind)
• AI Logic: service/ai_agent.py
• Trading Core: service/server.py