Vibe-Trading Docker setup with Ollama configuration

This tutorial covers a self-hosted Vibe-Trading research workspace using Docker. The speaker describes the open source framework as a system that turns natural language finance questions into executable Python code and backtests. The speaker's setup calls for Docker and Git, with provider credentials for hosted LLMs. It also explains the endpoint setting for a local Ollama instance.

The walkthrough starts with cloning the HKUDS repository and inspecting docker-compose.yaml. Host volume mappings preserve session history and settings across container rebuilds. The speaker advises keeping these mappings and creating a root .env file for provider credentials. For Ollama running on the host, the guide uses host.docker.internal instead of localhost, which refers to the container itself in this context.

Configuration also covers API_AUTH_KEY authentication and search backend overrides using Google and Bing. Deployment uses docker compose pull followed by docker compose up -d, with docker ps as the check for running frontend and backend containers. The React interface opens at localhost:5899.

A request to fetch the latest NVDA data serves as the connection test. The tutorial then introduces Alpha Zoo, which the speaker describes as a library of 452 quantitative factors. Claims about preventing look-ahead data are the speaker's explanation, rather than an independent validation of the factors or backtest results.