Machine-Learning on Jonathan Franklin

Automated Trading System

Mon, 07 Sep 2026 00:00:00 +0000

Problem

Pricesearcher’s automated FX/CFD trading operation needed more than strategies. Running through MetaTrader 5 meant connecting research and backtesting to live execution, account-level risk settings, data collection and reporting. The software also needed to handle market hours and the practical constraints of a Windows-based trading terminal.

Architecture

MQL5 Expert Advisors ran inside MetaTrader 5, while a Python/FastAPI trade engine used the MT5 connector alongside the terminal on a Windows VPS. S3 held exported trade history and the OHLCV pipeline supplied bars for research and backtesting. AWS monitoring fed alerts into Slack, and an internal dashboard exposed strategy, account and commission views.

Socketrade Trading Bot

Thu, 18 Jun 2026 00:00:00 +0000

Problem

MetaTrader 5’s Python API is request/response based. For Socketrade, I wanted the terminal to push ticks, closed bars and trade transactions to Python strategies, with commands travelling back for order execution.

The runtime also needed to manage concurrent setups—each a strategy, symbol or pair, timeframe and risk configuration. Changing a setup should not require restarting the bot, and disabling new entries should not abandon positions already open.

Architecture

An MQL5 Expert Advisor and Python asyncio hub form a newline-delimited JSON socket bridge. The hub feeds terminal events into a strategy runtime whose setups are defined by YAML configuration and model files. A backtester feeds historical bars through the same strategy runtime, with a small web UI generating forms from the configuration schema and charting results.

Quantrade Research System

Tue, 16 Jun 2026 00:00:00 +0000

Problem

I built Quantrade as a personal research system for machine-learning trading strategies. I wanted to compare ideas on consistent data and evaluation rules, rather than judge a model by prediction accuracy or a favourable chart alone. Trading fees, drawdown and the order of training and test data all change how useful a result is.

Architecture

A Python research library connects market-data providers, feature preparation, PyTorch trainers and fee-aware trade simulations. A YAML-driven multiprocessing runner executes comparable benchmark plans and writes result tables, while streaming feature and account models connect the same research ideas to an event-driven strategy interface.