Algorithmic trading with explainable AI signals
A platform for algorithmic trading: the signal engine combines rule-based technical analysis, news sentiment and fundamentals, shows why every recommendation was made, and validates itself through a matrix backtest and paper trading via a broker API.
Dozens of tickers, news and indicators, every day.
Trading a stock market by hand means tracking dozens of tickers, reading the news and weighing indicators, every single day. An early-stage fintech wanted a system that does this itself — and above all shows why it made each decision.
Signals that show why they fired.
We built a .NET backend with realtime delivery and full telemetry — traces, metrics and logs — into its own observability stack. The signal engine combines rule-based technical analysis, language-model scoring of news sentiment and fundamentals into a morning recommendation with an audit trail of which indicators fired and how the model reasoned. On top of that: a matrix backtest across multiple tickers, paper trading through a broker API and a monitor comparing internal validation against real runs.
A morning recommendation that explains itself.
The main application carries 729 commits in roughly 4.5 months — a working proof of concept spanning the signal engine to paper trading via a broker API.