CryptoEdge & TradeV1 ML
Binance Algorithmic Trading
Automated Binance crypto trading engine and machine learning price forecasting pipeline with cloud CI/CD.
Overview
CryptoEdge is an institutional-grade quantitative trading bot and machine learning forecasting platform connected to the Binance Spot and Futures exchange. It combines technical indicator momentum models (MACD, RSI, Bollinger Bands, Order Flow) with supervised ML models to generate high-probability trade signals, execute risk-managed orders, and log execution metrics.
Design decisions
The choices that shaped the system, and why.
- 1
Automated Cloud Build CI/CD (`cloudbuild.yaml`)
Implemented Google Cloud Build automation that triggers unit tests (`pytest`), container builds, and security scans on every Git push before staging deployment.
- 2
Strict Risk Management Circuit Breakers
Built automated kill-switches enforcing maximum drawdown limits per 24 hours to protect capital from flash crashes or exchange API latency.
Architecture
Real-Time Market Data Ingestion
WebSocket streaming of Binance order books, trade ticks, and OHLCV kline intervals.
Feature Engineering Pipeline
Computes 40+ quantitative indicators, volatility metrics, and moving average spreads in vectorized Pandas arrays.
Algorithmic Execution Engine
Implements stop-loss, take-profit trailing stops, and position sizing based on portfolio Kelly criterion.
Cloud-Native Architecture
Dockerized with Google Cloud Build CI/CD pipelines for automated testing and container deployment.
DevOps & deployment
Configured for Google Cloud Build with container registry storage and deployment to GCP Compute Engine or AWS EC2 with low-latency exchange peering.
How it ships · Google Cloud Run · Secret Manager · keyless Workload Identity Federation
- 1Push to main
- 2pytest + import/route smoke check
- 3Auth to GCP with OIDC (no JSON key)
- 4Build & push to Artifact Registry
- 5Cloud Run deploy with --min-instances=1 and secrets from Secret Manager
Features
- Live paper trading and real-money Binance API key execution.
- Automated risk management preventing drawdown exceeding preset thresholds.
- Predictive ML model predicting candle direction with historical backtesting engine.
- Comprehensive trade journal and PnL reporting dashboard.
Related writing
DevOps · 4 min
Cloud Run Problems I Actually Hit — and the Morning My Demo Was Gone
Missing service accounts, region-locked models, disks that vanish, one-writer SQLite, schedulers that stop — and the build-time check that found Nora Labs returning Google's 404.
29 Oct 2026
DevOps · 4 min
The $67.88 Lesson: What Two Trading Bots Taught Me About Automation
61 real trades, a profitable_trades.csv full of test rows, and a bot that bought 182 times but sold 6 — why logs aren't outcomes, and the guardrails that replaced them.
27 Oct 2026
Deep dive · 4 min
CryptoEdge: Explainable Market Signals Instead of a Black Box
Seven plain questions, 240 indicator readings, an order-flow lab and paper mode by default — plus the cleanest CI/CD pipeline in my portfolio.
20 Jul 2026
More in ML, Vision & Quant
ML & Vision
Quintrix ITSO Engine
Content-aware retention for CCTV: score every 15-second segment by what happens in it, keep incidents lossless and empty corridors as keyframes. 60.0% less storage on the evaluated footage.
ML & Vision
Machine Learning Playbook & Deep Learning Lab
Interactive machine learning educational playground and hands-on laboratory covering computer vision, regression, and deep neural nets.