Vikeur

Vikeur is a system I built on my own to study and test automated crypto trading. It collects market data, generates trading signals, and can execute trades — in a simulated ("paper") mode or, once a strategy is proven, in a controlled live mode.

Automated trading can go wrong fast, in a way that loses real money. I wanted a system where mistakes are caught early, and where a strategy cannot trade with real funds until it has proven itself with data.

I built a backend in Python and FastAPI that collects live market data from HTX and Binance, runs several independent decision engines, and combines their signals through a calibration step. A separate risk engine checks every decision before it can be executed, and can stop all trading immediately through a kill switch.

I designed and built the full system alone: the data pipeline, the decision engines, the risk engine, the strategy lifecycle logic, the Next.js dashboard, and the deployment setup.

• Live market data from HTX and Binance (spot and futures)
• Several independent decision engines, merged through probability calibration
• A risk engine with a hard kill switch
• A governance rule that blocks live trading until a strategy is statistically validated
• Automatic suspension of strategies that stop performing
• Telegram notifications
• A Next.js dashboard to monitor the system, protected behind authentication

I built this in stages, adding one capability at a time — data, then signals, then risk, then execution, then monitoring — and testing each stage before moving to the next.

The system is feature-complete and running in paper mode. I have not moved real funds into live trading yet — the governance rule is doing exactly what it is designed to do.