Investment Rating - The report maintains an "Accumulate" rating for the real estate industry [1]. Core Insights - The report emphasizes the integration of AI and quantitative analysis in stock selection for the real estate sector, utilizing a dual-model framework that covers A-shares, Hong Kong stocks, and US stocks [4][10]. - The system employs a three-model architecture (DeepSeek V3.2, GLM-5, Kimi K2.5) to enhance stock rating accuracy by incorporating real-time policy updates and market dynamics [9][10]. - The report highlights the importance of a multi-source data collection strategy to ensure stable data availability for stock evaluations [4][10]. Summary by Sections 1. Project Overview - The project aims to create an intelligent stock selection system that combines AI technology with traditional quantitative and fundamental analysis to provide timely stock ratings in the real estate sector [9]. - The system is designed to address challenges such as information overload and rapid policy changes that traditional research methods struggle to manage [9]. 2. Real Estate Stock AI Selection Model - The system features two independent rating models: "Quantitative AI Selection" and "Dongwu Real Estate Selection," catering to different investor preferences [10]. - As of March 3, 2026, the system covers 61 real estate-related stocks across three major markets [10][17]. 3. Data Collection and Multi-source Downgrade - The system utilizes a four-layer data downgrade architecture to ensure data collection stability, switching to backup sources when primary data is unavailable [4][10]. - The AI model integrates real-time policy searches to reflect significant policy changes in stock ratings, with policy impact accounting for up to 35% of the rating [4][10]. 4. Technical Architecture Overview - The system employs a classic front-end and back-end separation architecture, ensuring efficient data processing and user interaction [11][13]. 5. Stock Coverage - The system covers a diverse range of real estate stocks, including developers, property management, and REITs, with a focus on financially healthy companies [18]. 6. AI Model Scoring - The AI scoring model is the most critical component, accounting for 50% of the overall rating, and is designed to analyze stocks as a seasoned analyst would [43][50]. - The model assesses various dimensions, including policy impact, company fundamentals, technical analysis, and market sentiment [50].
房地产行业深度报告:基于量化分析与大模型的房地产股票AI选股系统
Soochow Securities·2026-03-05 06:29