What problem does it solve? After a daily market review, investors still need to manually sift through thousands of A-share stocks to find candidates likely to strengthen in the coming days. This Skill automates that selection by combining market review outputs, model inference scores, L2 order-book microstructure factors, position signals, sector strength, and news sentiment into one ranked candidate list with deep per-stock analysis. ## Core Features & Use Cases - Multi-dimensional candidate scoring: Weights L2 microstructure (40%), model fusion score, L1 momentum, position signals, sector strength, and news sentiment into a composite ranking, with hard filters for ST stocks, negative fusion scores, and position gates. - Cross-day aggregation and backtesting: Supports --window N aggregation across inference days and a no-lookahead backtest script that validates picks against the Shanghai Composite Index with T+1/T+3/T+5 returns and stop-loss simulation. - End-to-end report delivery: Runs 9-layer deep analysis on Top picks, writes a Markdown report, converts it to PDF, and publishes both to the stock reports directory visible in the frontend. - Use Case: After the market closes, ask for today's stock recommendations; the Skill runs the daily review data collection, scores the full market, deep-dives the Top 5, and delivers a PDF report with candidate rankings, risk statements, and a next-day verification checklist. ## Quick Start Ask the assistant to run today's multi-dimensional stock picks for the A-share market and generate the recommendation report with Top 5 deep analysis as a PDF.
