A-Share Market Data + LLM Analysis for FinRL — open source tool

[ZiwayZhao]

Problem

FinRL supports US market data (Yahoo Finance, FMP, WRDS) but lacks native support for Chinese A-share market — 5000+ listed companies on Shanghai, Shenzhen, and Beijing exchanges.

What we built

CoWorker Fin-Agent — an open-source A-share analysis tool:

• Real-time data for all A-share stocks via AKShare
• Technical indicators: MA60 position, volume-price patterns (9 rules), monthly golden eye
• LLM analysis: Deep research reports via DeepSeek with proprietary methodology
• Structured output: JSON format ready to feed into RL agents as features

# Example: analyze Kweichow Moutai
from examples.fin_agent import deep_analysis
report = deep_analysis("600519")
print(report["metrics"])  # {price, ma60_position, volume_pattern, golden_eye, ...}

Why it matters for FinRL

1. Data gap: A-share users currently build their own pipeline
2. Features: Technical indicators can serve as RL agent observation space
3. LLM signals: Research reports can augment decision making

MIT licensed, pip install agent-coworker. Would love feedback from the community!

GitHub: https://github.com/ZiwayZhao/agent-coworker