wjhccc

wjhccc / TradingAgents-Studio

Public

可视化多智能体 LLM 交易研究平台 — 看见 Agent 怎么想、怎么辩、怎么决策,而不是只看最后一个 BUY/SELL。

agent python quantitative-trading
23
1
85% credibility
Found May 23, 2026 at 25 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

TradingAgents-Studio is an open-source visual research platform that uses multiple AI agents to analyze stocks and financial markets. The system coordinates different AI specialists—a market analyst, sentiment analyst, news analyst, and fundamentals analyst—whose findings feed into a debate between Bull and Bear researchers. A trading agent then proposes specific actions, which are further scrutinized by risk management agents with different investment styles. Finally, a portfolio manager synthesizes everything into a clear investment decision with ratings, price targets, and stop-loss levels. The platform supports both US/global markets and Chinese A-shares, includes a web interface for visualizing the AI debate in real-time, offers holdings tracking with live quotes, paper trading for virtual portfolio simulation, and historical backtesting. It integrates free data sources by default and clearly states it is for educational research only, not investment advice.

How It Works

1
🔍 You discover a stock research tool

You hear about a platform where AI agents work together to analyze stocks, debate investment ideas, and show their reasoning visually.

2
💻 You open the web dashboard

The platform greets you with a clean dashboard where you can track your holdings, run new analyses, and review past research reports.

3
🎯 You ask about a stock in plain words

Simply type 'research Tesla short-term' or '研究茅台短期' and the system understands your intent, fills in the stock details, and prepares your analysis automatically.

4
🤖 AI agents start their research

Multiple AI analysts simultaneously examine market data, news, social sentiment, and financial details for your chosen stock.

5
You watch the AI debate unfold
📈
Bull perspective argues for buying

Highlights growth opportunities, positive indicators, and favorable market conditions

📉
Bear perspective argues caution

Points out risks, competitive threats, and potential downsides

6
📊 You receive a clear decision

The system produces a structured recommendation with a rating (Buy/Hold/Sell), entry price, target, stop-loss, and the reasoning behind it all.

7
💼 You can practice with paper trading

If you want to try it out, you can simulate trades based on the AI's recommendations using a virtual account to see how the strategy would have performed.

✅ You have a research companion

You now have a tool that shows you not just what to think about a stock, but how AI agents analyze, debate, and arrive at their conclusions.

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 25 to 23 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is TradingAgents-Studio?

TradingAgents-Studio is a visual research workbench that shows you how multiple AI agents debate their way to a trading decision, rather than just handing you a BUY/SELL signal at the end. Built in Python, it layers a web UI and Chinese A-share support onto the upstream TradingAgents framework. The architecture chains together specialist analysts (market, news, sentiment, fundamentals), bull and bear researchers who argue opposing positions, a trader who proposes positions, and a risk management team that debates sizing and stops. Everything flows through LangGraph, with outputs streamed live to a Vue 3 dashboard where you watch the debate unfold as chat bubbles and causal chain cards. You can also run it headless via CLI or Docker.

Why is it gaining traction?

The hook is transparency: most LLM trading tools give you a wall of markdown and expect you to reverse-engineer the logic. This project inverts that by making the agent reasoning the product. The A-share support is a differentiator, with four China-specific analysts (cn_social, event, capital_flow, macro) and free data via AKShare that auto-detects Chinese ticker formats. The multi-LLM factory supports everything from DeepSeek to Claude to local Ollama, so you're not locked into one provider. Decision Replay backtesting replays stored agent calls with zero LLM cost, letting you stress-test historical performance.

Who should use this?

Python developers building trading research pipelines who want visibility into AI reasoning. Quants exploring multi-agent llm agent frameworks for market analysis. Retail traders who want to understand what an AI system is actually doing before trusting a signal. Educators teaching llm agent architecture through a concrete, hands-on example. The web UI lowers the barrier for non-CLI users; the Python API lets you embed it into larger systems.

Verdict

TradingAgents-Studio scores 0.8500000238418579% on credibility. The codebase is well-structured with clear documentation, and the Apache 2.0 license invites community contribution. At 23 stars it's early-stage, so production use requires caution. If you're building with llm github models and want a research platform that shows its work, this is worth a weekend experiment.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.