Hey HN,

  I built ACIS Trading (https://acis-trading.com) after getting frustrated that robo-advisors wanted to manage my money instead of helping me with what I already owned.

  ACIS connects to your brokerage (Schwab, E*Trade, Webull, Alpaca) or imports CSV, then uses LightGBM models trained on 10 years of daily data to identify rebalancing opportunities. It tells you exactly what to trade ("Sell 33 shares of NVDA, buy 75 shares of JNJ")
  based on concentration risk, volatility, and ML scores.

  Tech stack: FastAPI, React, PostgreSQL, DuckDB feature store, LightGBM (GPU), JAX PPO for position sizing.

  Would love feedback from the HN community. Happy to answer questions about the ML approach or architecture.

  ---
  Want me to save these somewhere in your repo for reference?