broker-research-analyst

Aggregates and analyzes Chinese broker research reports from Eastmoney into structured Markdown reports.

6.4k|1.2k|Updated Jun 9, 2022
One-click install
npx skills add https://github.com/charliedream1/ai_quant_trade --skill broker-research-analyst
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: broker-research-analyst
Source: https://github.com/charliedream1/ai_quant_trade/tree/main/egs_skill/broker-research-analyst
Command: npx skills add https://github.com/charliedream1/ai_quant_trade --skill broker-research-analyst

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests, markitdown, pdfplumber, PyPDF2, PyMuPDF, Pillow, akshare, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

Manually collecting and comparing sell-side research reports across multiple brokerages is slow and error-prone. This Skill automates the full pipeline of fetching broker research reports, extracting ratings and earnings forecasts, comparing institutional views, and flagging sell-side bias and stale reports.

Core Features & Use Cases

  • Research Report Aggregation: Fetches stock, industry, strategy, and macro research reports from the Eastmoney public API with no API key required, normalizing ratings, target prices, and EPS/PE forecasts.
  • PDF Parsing and Chart Extraction: Downloads report PDFs and parses them through a layered chain (MarkItDown, pdfplumber, PyPDF2, optional MinerU), extracting chart images via PyMuPDF for multimodal LLM analysis.
  • Quality Gate and Bias Detection: Filters reports by freshness (90 days), a 25-broker whitelist, and rating completeness, and flags sell-side conflicts of interest when buy ratings dominate.
  • Multi-Agent Analysis: Coordinates six expert agents (auditor, rating extractor, fundamental extractor, risk spotter, consensus aggregator, supervisor) to produce a cited Markdown report with hallucination checks.
  • Use Case: Ask for a summary of all broker opinions on Kweichow Moutai (600519) over the last 90 days and receive a report with rating distribution, EPS consensus, institutional divergence, and risk warnings.

Quick Start

Ask the assistant to summarize the last 90 days of broker research reports for stock 600519 and generate an analysis report with ratings, target prices, and risk warnings.

Frequently Asked Questions about broker-research-analyst

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I aggregate broker research reports for a Chinese stock?

Run report_router.py with the stock subcommand, passing the 6-digit stock code, name, and lookback window in days. It fetches report metadata from the Eastmoney public API, applies quality gates, and generates a Markdown analysis report.

What data source does broker research report aggregation use?

The primary source is the Eastmoney research report center public API at reportapi.eastmoney.com, which requires no API key. AKShare stock_research_report_em serves as a secondary source, with WebSearch as a fallback.

Which Python libraries parse broker research report PDFs?

The parser chain tries MarkItDown first for LLM-friendly Markdown with tables, then falls back to pdfplumber and PyPDF2. An optional MinerU path offers higher accuracy for Chinese PDFs, and PyMuPDF extracts chart images independently.

Does the Eastmoney research report API require an API key?

No, the Eastmoney report list API and PDF download endpoints are public and need no key. Requests should be throttled to at least one second apart to avoid rate limiting or anti-bot challenge pages.

Why does PDF download from Eastmoney sometimes fail?

Eastmoney's pdf.dfcfw.com returns a JavaScript anti-bot challenge page instead of the PDF. The downloader detects this page, executes its JavaScript with Node.js to solve the cookie, retries, and verifies the response starts with %PDF before saving.

How does the skill detect sell-side bias in research reports?

The quality gate counts rating distribution across reports. If buy or overweight ratings exceed 80 percent with zero sell ratings, it flags the set as sell-side generally bullish and warns about conflicts of interest in the report.