OpenAlpha

AI, LLM, Self-host, Stock, Crypto, Deep research

Fintech

Crypto

Artificial Intelligence

GitHub

Visit WebsiteSee on Product HuntGithub ⧉

I'm Alex, and OpenAlpha is my first product. I built it to fix my own investing habits: buying on tips, FOMO, holding losers too long, and trying to catch falling knives. I knew the rules, but I kept breaking them at the moment I had to click. So I wrote an assistant that makes the plan before I'm emotional, then holds me to it. What it does: - Screens the market every morning and sends a brief to Telegram or Zalo - Has 7 AI analysts debate each idea, a bull side and a bear side, before anything is suggested - Computes entry, stop-loss, target and position size in code. The AI only argues why, so it can't invent numbers - Lets stop-losses move up only, never down - Tracks its past calls with /reflect and gets less confident when it's wrong It's strongest on Vietnamese stocks (9 local data sources), and it also covers crypto and gold. Two design choices I care about: 1. It never places a trade on its own. You decide and you click. 2. The AI never sees your money. It sees portfolio weights, not balances or position sizes. The code has no function that would return them. It's open source (MIT) and runs on your own machine. It works with 53 LLM providers, including free models, or local ones via Ollama. To be upfront: this is a first release. There's no test suite yet, and the setup still assumes macOS/Linux and a terminal. I'd love your honest feedback, especially on: - Would you trust an AI research tool more if it's not allowed to touch the numbers? - Which market should come next? GitHub: https://github.com/alex2loop/ope...