If you're like me, relying on expensive financial data terminals like Bloomberg can feel limiting — both in terms of cost and customization. Recently, I started exploring TradingAgents, a fully open-source trading framework that literally allows you to run a sophisticated trading desk right on your own machine without monthly fees. TradingAgents is built on advanced large language models (LLMs) and modular multi-agent architecture. It decomposes trading tasks across various specialized agents — including fundamentals, sentiment, news, and technical analysts — plus bullish and bearish researchers who debate the investment merit. The consensus feeds into a trader agent that issues actionable trading decisions. What impressed me most is the wide range of supported LLM providers: OpenAI's GPT, Google's Gemini, Anthropic's Claude, xAI Grok, MiniMax, and even local models via Ollama. This flexibility makes it easy to tailor the system depending on your API keys or preference for local hosting. The framework covers markets globally, including stocks across US, Hong Kong, London, India, China, Australia, Canada, and crypto markets like BTC-USD and ETH-USD — all via standardized ticker formats from Yahoo Finance. This makes it versatile for diverse trading approaches. Persistence and checkpoint resume features are invaluable for maintaining continuity in research even after interruptions. Each run logs decisions and market reflections, so the system 'learns' from prior outcomes. Plus, customizable sampling temperature settings help control model output variability, ensuring consistency crucial for financial research. The community is active with regular releases adding features like sentiment analysis improvements, new data vendors, language support, and enhanced system stability. This makes TradingAgents not only a tool but a growing platform for collaborative trading research. I run the system through Docker on my local box, and the interactive CLI interface is intuitive — letting me select tickers, research depth, and providers quickly. If you want to step away from costly terminals and experiment with AI-powered trading research, this framework offers a powerful starting point. Note: While TradingAgents is built for research and experimentation, be cautious using it for live trading decisions without thorough validation — it’s not financial advice but a research scaffold. Still, it’s a remarkable open-source breakthrough for traders and researchers wanting full control of their desk and data pipelines.