策略深度报告：基于XGBoost模型的AI行业配置系统

Group 1 - The report constructs an AI industry allocation system based on the "cycle-signal-momentum" framework, utilizing the XGBoost model to enhance industry allocation strategies in response to accelerated market rotation and changing investment styles [2][3][21] - The "cycle-signal-momentum" framework categorizes industries into eight major sectors, including cyclical resources, real estate, public utilities, essential consumption, discretionary consumption, hard technology, soft technology, and finance, each with distinct macro or micro signals [3][18][21] - The report emphasizes the need for a data-driven approach in strategy research, moving away from subjective experience to models that provide predictive capability, interpretability, and practical applicability [2][21] Group 2 - The AI industry allocation model has shown a cumulative return of 122.31% from January 2014 to March 2025, outperforming the equal-weighted industry index return of 80.26% during the same period [3][4] - The model incorporates 15 core factors, including macroeconomic indicators, market sentiment variables, and industry profitability metrics, to capture the multidimensional drivers of industry returns [3][67] - The report highlights the importance of liquidity and industry cycles in driving the performance of technology and high-end manufacturing sectors, with significant trends observed during the 3G, 4G, and 5G cycles [43][44][46] Group 3 - Essential consumption shows strong defensive characteristics, with its market performance closely correlated with CPI, indicating that price movements significantly influence this sector [36][37] - Discretionary consumption is driven by income and consumer expectations, with historical data showing a high correlation with retail sales growth, reflecting consumer confidence and spending behavior [41][42] - The financial sector's performance is influenced by financial cycles, with key indicators such as loan growth and net interest margins playing a crucial role in investment decisions [48][52]