Back to projects
In active development

OpenClaw

Automated algorithmic trading platform

PythonLangGraphFastAPIPostgreSQLInteractive Brokers API

OpenClaw is an automated algorithmic trading platform that connects to Interactive Brokers and runs the full loop — from idea to execution to monitoring — without manual intervention. It's built around agent workflows that research candidates, size positions against risk limits, and place orders, with every decision recorded for later analysis.

The system is organized into two sleeves: a core book of higher-conviction holdings and a discovery book for exploratory ideas. A FastAPI service orchestrates the LangGraph agents and persists portfolio state to PostgreSQL, while risk controls cap exposure and guard against runaway behaviour. Marked-to-market snapshots are published every few minutes during US market hours.

The dashboard below is a live, read-only view of a paper-trading account driven by that publisher — equity, P&L, and current holdings as the backend computes them. OpenClaw is in active development; the strategy and tooling continue to evolve.

Key capabilities

  • Portfolio monitoring
  • Risk management
  • Agent workflows
  • Execution automation
  • Trade analytics

System architecture

How the pieces fit together — from research and market data through risk checks to broker execution.

OpenClaw system architecture diagram
Decision-support pipeline with human-in-the-loop approval before every order. Isolated Docker networks separate research, market data, and broker paths; the LLM is veto-only and can never raise a score, size a position, or approve what the quant layer rejected.Open full size

Live portfolio

A read-only snapshot of a paper-trading account. Marks and P&L are computed by the backend and published every few minutes during US market hours.

Loading portfolio…