📊 Automated Multi-Agentic Trading System

An elite, high-performance full-stack portal built with Python FastAPI, SQLAlchemy ORM, Pydantic Validation, and a Bloomberg-inspired, industrial-utilitarian dashboard. This platform implements a decoupled, asynchronous multi-agent orchestration architecture powered by the OpenAI Agents SDK and managed with uv.

---

🏛️ System Architecture

Rather than relying on a single monolithic prompt, this system enforces a strict separation of concerns among specialized execution nodes operating in sandboxed contexts. Data flows strictly via immutable message states.

       [ USER / OPERATOR ]
               │
               ▼
┌──────────────────────────────┐
│  Master Trading Desk Router  │  ◄─── [ FastAPI Web Gateway & Session Control ]
└──────────────┬───────────────┘
               │
               ▼  (Extracts parameters & timeframes)
┌──────────────────────────────┐
│ Intraday/Swing Market Analyst│  ◄─── [ technical metrics, indicators: RSI, EMA ]
└──────────────┬───────────────┘
               │
               ▼  (Proposes trading parameters)
┌──────────────────────────────┐
│      Risk Manager Agent      │  ◄─── [ compliance audits: APPROVED / REJECTED ]
└──────────────┬───────────────┘
               │
               ▼  (Deterministic Overrides applied)
┌──────────────────────────────┐
│      Neon PostgreSQL DB      │  ◄─── [ Persists transaction logs & history ]
└──────────────────────────────┘

🔹 1. Master Trading Desk Router (Orchestrator)

• Role: Conversational Gateway, State Initiator & Interface Controller.
• Objective: Ingests raw user intent, validates asset token parameters, and guides the user through execution preferences.
• Rules: Standardizes tickers (AAPL for stocks, BTC-USD for crypto) and restricts data fetching until both the asset identity and timeframe targets are locked in.

🔹 2. Intraday & Swing Market Analyst

• Role: Quantitative & Technical Signal Synthesizer.
• Objective: Consumes mathematical indicator matrices, calculates moving averages (EMA/SMA), computes momentum vectors, and outputs high-fidelity summaries.
• Rules: Holds zero portfolio execution or transactional decision authority; never outputs buy/sell signals directly.

🔹 3. Risk Manager Agent

• Role: Capital Preservation & Compliance Audit Guardrail.
• Objective: Final automated gatekeeper that audits proposed operations against structural safety rules.
• Rules: Prefixes audits with [APPROVED] or [REJECTED] and appends a single sentence of quantitative compliance justification.

---

⚡ Key Features

• Premium Bloomberg-Inspired UI: Radial dark slate background (#0B0F17), sleek panel styling (#131C2E), and absolute neon-accent high-contrast color codes.
• Interactive Live SVG Candle Charting: Fetches actual real-time price candlesticks (using yfinance) and dynamically wicks, scales, and plots candles inside the SVG container with a 10-second polling live update.
• Quick-Select Asset Badges: Instantly targets leading tech assets (AAPL, NVDA, TSLA, MSFT) and cryptocurrencies (BTC-USD, ETH-USD, SOL-USD) at the click of a button.
• Strict Parameter Overrides: Integrates real-time compliance blocks. If the Risk Manager issues a [REJECTED] verdict, the portal forces a strict HOLD override and updates position sizing allocations to 0% to preserve capital.
• State-of-the-Art Langfuse Telemetry: Tracks global traces the moment the user interacts with the desk, nests execution spans for each agent run, and captures final database insertion payloads.
• Secure Operator Portal: Full SQLAlchemy-backed authentication featuring custom-validated sign-ups (Pakistani/International cell validation, flexible 6+ character passwords) and password hashing with passlib[bcrypt].

---

📊 Dynamic Strategy Matrix

Data lookbacks and execution calculations automatically shift based on the strategy selected in the dashboard:

Strategy Category	Timeframe	Lookback	Data Interval	Primary Indicators	Exit Bounds (Stop Loss / Take Profit)
Intraday Scalp	15-Minute (15m)	7 Days (7d)	15m	9-EMA, 21-EMA, RSI (14)	-1.0% / +2.0%
Intraday Swing	1-Hour (1h)	30 Days (30d)	1h	9-EMA, 21-EMA, RSI (14)	-2.5% / +5.0%
Medium Swing	4-Hour (4h)	60 Days (60d)	4h	50-MA, MACD, RSI (14)	-4.0% / +8.0%
Macro Position	1-Day (1d)	180 Days (180d)	1d	50-MA, 200-MA, Macro RSI	-8.0% / +15.0%

---

⚙️ Installation & Setup

Ensure you have Python 3.12+ and uv installed on your system.

1. Clone & Initialize Workspace

git clone https://github.com/Ahmadthealdo/Automated-Multi-Agentic-Trading-System.git
cd Automated-Multi-Agentic-Trading-System

2. Install Project Dependencies

Run the following uv command to create the virtual environment and install all web-hosting, validation, and multi-agent libraries:

uv pip install -r pyproject.toml
# Or install explicitly:
uv add fastapi uvicorn "pydantic[email]" "passlib[bcrypt]" psycopg sqlalchemy yfinance openai langfuse

3. Configure Environment Variables

Create a .env file in the root directory:

# Relational Neon PostgreSQL Connection String
DATABASE_URL=postgresql://<user>:<password>@<host>/<database>?sslmode=require

# OpenAI API Key for Multi-Agent Brain
OPENAI_API_KEY=sk-proj-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

# Langfuse Telemetry Tracing Keys
LANGFUSE_PUBLIC_KEY=pk-lf-xxxxxxxxxxxxxxxxxxxx
LANGFUSE_SECRET_KEY=sk-lf-xxxxxxxxxxxxxxxxxxxx
LANGFUSE_HOST=https://cloud.langfuse.com

4. Database Setup & Migrations

Upon application startup, FastAPI dynamically verifies relational tables and automatically handles any database column migrations (such as upgrading mobile_number to verified_phone on Neon PostgreSQL), making database setup 100% zero-configuration.

---

🚀 Running the Application

Start the FastAPI Dev Server

Launch the application locally in reloading development mode:

uvicorn main:app --reload

The application will boot and host the interactive portal locally at http://127.0.0.1:8000.

🧪 Programmatic Verification Run

Run our end-to-end multi-agent pipeline validation test in your terminal to verify databases, indicators, and agent outputs instantly:

python .agents/brain/8169d2e7-195f-482a-bf14-b5bf777313c9/scratch/validate_system.py

---

📁 File Structure

├── .agents/                    # System Agent workspace instructions & skills
├── static/
│   ├── app.js                  # Scaled SVG candle drawer & client logic
│   └── style.css               # Utilitarian Bloomberg layout styles
├── templates/                  # Frontend views
├── index.html                  # Core single-page operator dashboard
├── main.py                     # FastAPI server routes & multi-agent execution pipeline
├── schemas.py                  # Pydantic validation boundaries & schemas
├── tools.py                    # yfinance quantitative extraction tools
├── pyproject.toml              # UV dependency declarations
└── README.md                   # System documentation

---

---

👥 Authors & Contributors

• Abdul Qadeer Khan
• Ahmad Abdullah

---

🛡️ License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.