validation · Python 3.12 · FastAPI · PostgreSQL · TimescaleDB · Claude API · APScheduler · Flutter
AI Trader
Macro-news trading system with hourly LLM decisions.
Repo: moinsen-dev/ai_trader
AI Trader
Read news hourly, interpret them with Claude, suggest probabilistic crypto positions, and measure them honestly against reality.
How It Works
The system runs an hourly decision loop:
- ingest news and macro events
- interpret the situation with Claude
- derive probabilistic portfolio actions for BTC and ETH
Two systems run in parallel:
- Direction Trader: long, short, or hold signal
- Volatility Oracle: DVOL index prediction
Both are continuously validated against reality. Measurement first, story later.
Personal Anchor
I have been active in markets since 2008. In Zurich, 2,000 euros once became almost 200,000 over four years. Not one lucky trade, but discipline, patience, and a market that moved at the right time.
AI Trader is the next step: turning trading intuition into code that can stay with the task when I cannot.
Tech
- Stack: Python 3.12, FastAPI, PostgreSQL, TimescaleDB, Claude API, APScheduler, Flutter dashboard
- Repo:
moinsen-dev/ai_trader(private) - Status: simulation and validation since March 25, 2026