Pulse Terminal

Pulse Terminal is a full-stack, real-time cryptocurrency trading interface and market data platform. The system streams L2 order book data from Binance WebSockets, processes and aggregates order book parameters, synchronizes state via Redis, and renders a high-performance terminal UI.

Architecture

                       +-------------------+
                       | Binance WebSocket |
                       +---------+---------+
                                 | (wss://stream.binance.us)
                                 v
                       +---------+---------+
                       |  binance_ws.py    |
                       +----+---------+----+
                            |         | (Every 60s)
               (Redis Pub)  |         v
                            |  +------+------+
                            |  | PostgreSQL  |
                            v  +------+------+
                       +----+----+    |
                       |  Redis  |    | (GET /api/ohlcv)
                       | Pub/Sub |    |
                       +----+----+    |
                            |         |
               (Redis Sub)  |         |
                            v         v
                       +----+---------+----+
                       |    FastAPI WS     | (main.py)
                       +---------+---------+
                                 | (WebSocket)
                                 v
                       +---------+---------+
                       |   React Client    | (Pulse UI)
                       +-------------------+

---

Features

• Real-Time Order Book Engine: Connects to the Binance L2 depth stream and pushes real-time bid and ask arrays to the client with sub-100ms latency.
• Microstructure Metrics: Computes key market-microstructure metrics on every packet:
  • Mid Price: Average of best bid and best ask.
  • Spread: Difference between best ask and best bid.
  • CVD (Cumulative Volume Delta): Volume differences (bid volume minus ask volume) across the top 20 price levels to track net buying and selling pressure.
• Redis Integration: Uses a Redis pub/sub broker to fan out updates to multiple clients.
• Memory Optimization: Employs capped local order book state tracking, pruning zero-quantity levels and restricting stored records to the top 20 entries.
• Pulse UI: A dark-themed trading desk dashboard with proportional depth bars and instant visual updates.

---

Tech Stack

Component	Technology
Backend	Python 3.11, FastAPI, Uvicorn
Streaming	Binance WebSocket, websockets library
Pub/Sub Broker	Redis 7
Database	PostgreSQL 15, asyncpg
Frontend	React, Bootstrap
Infrastructure	Docker, Docker Compose

---

Local Development

To run the backend services (FastAPI, Redis, PostgreSQL) locally:

docker compose up --build

Start the React client separately from the project root:

npm install --legacy-peer-deps
npm start

---

Production Deployment

To keep resource utilization under 512 MB:

1. Database: Use a managed serverless PostgreSQL database (e.g., Supabase). Use the Supavisor connection pooler URI as the DATABASE_URL.
2. Redis: Use a managed Redis instance (e.g., Render Key Value). Set the internal URL as REDIS_URL.
3. Backend: Host on Render as a Web Service. Set environment variables:
   • DATABASE_URL
   • REDIS_URL
4. Frontend: Host on Vercel as a static site. Set environment variables:
   • REACT_APP_API_BASE (Backend URL)
   • REACT_APP_WS_URL (Backend WebSocket URL)