Migrating a Legacy AmiBroker System to a Cloud-Based Python Bot

The Problem

Challenge

The client had been running an AmiBroker-based auto-trading system for several years using a third-party broker plugin that was no longer maintained. The plugin crashed regularly, requiring manual restarts. Worse, after the broker migrated to a new API version, the plugin stopped working entirely for an extended stretch — costing the client missed trading days. They needed a stable, maintainable solution that didn't depend on unmaintained software.

Our Approach

Solution

We rebuilt the entire system as a Python application connecting directly to the client's broker's REST API (the client's broker had migrated to a new API version). The strategy logic was ported from AFL to Python with a 1:1 logic match, validated against historical signals before going live. The bot runs as a systemd service on an Ubuntu VPS, auto-restarts on crash, and sends a Telegram alert if it goes down. We added a WebSocket-based market data feed (replacing the AmiBroker data plugin) and a simple browser dashboard for monitoring.

Outcomes

Results

• Unplanned downtime was eliminated during the observed live-operation period
• Order execution latency improved substantially compared to the old plugin
• The client's ongoing IT overhead for the trading system dropped to near zero
• Strategy performance was preserved — the Python port produced identical signals on historical data
• The system automatically resumes trading after VPS restarts without any manual intervention