Algorithmic Trading

A trading platform that runs 24/7 and manages multiple accounts at the same time. The backend runs on Node.js; signal generation combines XGBoost with online logistic regression.

Scope

• Node.js backend and multi-account trade execution
• Model-based signal generation (XGBoost, online logistic regression)
• Risk and position limits
• Monitoring, daily logs and error alerts

Approach

Models are trained offline while the online layer updates its coefficients with new data. Orders are constrained by per-account limits, and the system is monitored for continuous operation with safeguards that trigger on failure.

My role

Architecture, Node.js backend, model integration and development.