基本面量化系列研究之四十三：TMT拥挤度偏高，市场或继续高切低

Quantitative Models and Construction Methods 1. Model Name: Industry Rotation Model Based on Prosperity-Trend-Crowdedness Framework - Model Construction Idea: This model integrates three dimensions: industry prosperity, market trend, and crowdedness, aiming to identify industries with high prosperity, strong trends, and low crowdedness for rotation strategies [1][3][8] - Model Construction Process: 1. Prosperity Dimension: Evaluate industries based on fundamental indicators such as earnings growth and analyst expectations 2. Trend Dimension: Identify industries with strong upward momentum using technical indicators 3. Crowdedness Dimension: Measure the level of market participation and sentiment to avoid over-crowded industries 4. Combine these three dimensions into a scoring framework to rank industries and allocate weights accordingly [1][8][9] - Model Evaluation: The model has shown stable performance with consistent excess returns over benchmarks, making it a reliable tool for industry rotation [9] 2. Model Name: Industry Distress Reversal Model - Model Construction Idea: Focuses on industries in distress or recovering from past distress, with potential for long-term improvement in fundamentals and inventory cycles [17][99] - Model Construction Process: 1. Exclude industries with inventory and capital expenditure historical percentiles >80% 2. Exclude industries with gross margin and free cash flow historical percentiles <20% 3. Identify industries with improving inventory conditions and favorable macroeconomic signals 4. Allocate weights to industries meeting these criteria [17][99] - Model Evaluation: The model captures opportunities in recovering industries, delivering strong absolute and relative returns in backtests [99] 3. Model Name: PB-ROE Stock Selection Model - Model Construction Idea: Select stocks within industries based on valuation and profitability metrics, emphasizing high valuation-to-profitability efficiency [13][107] - Model Construction Process: 1. Use the industry allocation weights from the rotation model 2. Rank stocks within each industry by PB-ROE scores 3. Select the top 40% of stocks based on these scores 4. Weight selected stocks by their market capitalization and PB-ROE scores [13][107] - Model Evaluation: The model has demonstrated strong performance in both absolute and relative terms, with high information ratios and low drawdowns [107] --- Model Backtesting Results 1. Industry Rotation Model - Annualized excess return: 16.3% - Information ratio (IR): 1.74 - Maximum drawdown: -7.4% - Monthly win rate: 71% - 2023 excess return: 9.3% - 2024 excess return: 5.0% - 2025 YTD (as of April) excess return: 2.2% [9][96][97] 2. Industry Distress Reversal Model - Annualized excess return: 16.5% - Information ratio (IR): 1.76 - Maximum drawdown: -8.7% - 2023 absolute return: 13.0%, excess return: 16.6% - 2024 absolute return: 25.6%, excess return: 14.5% - 2025 YTD (as of April) excess return: 1.7% [99][101][102] 3. PB-ROE Stock Selection Model - Annualized excess return: 22.9% - Information ratio (IR): 2.02 - Maximum drawdown: -8.0% - Monthly win rate: 74% - 2022 excess return: 10.2% - 2023 excess return: 10.4% - 2024 absolute return: 14.6%, excess return: 4.6% - 2025 YTD (as of April) excess return: 1.0% [13][107][109] --- Quantitative Factors and Construction Methods 1. Factor Name: Crowdedness - Factor Construction Idea: Measures the level of market participation and sentiment to identify over-crowded industries [1][8] - Factor Construction Process: 1. Use trading volume, fund flows, and sentiment indicators to quantify crowdedness 2. Normalize the data and rank industries by crowdedness scores 3. Avoid industries with high crowdedness scores in allocation [1][8] 2. Factor Name: Inventory Cycle - Factor Construction Idea: Incorporates inventory levels and related metrics to identify industries with favorable inventory conditions [17][18] - Factor Construction Process: 1. Calculate inventory-to-sales ratios and historical percentiles 2. Identify industries with low inventory pressure and signs of restocking 3. Combine with macroeconomic indicators for final scoring [17][18] --- Factor Backtesting Results 1. Crowdedness Factor - Integrated into the industry rotation model, contributing to stable excess returns and risk control [1][8][9] 2. Inventory Cycle Factor - Integrated into the distress reversal model, enhancing its ability to capture recovery opportunities in industries [17][18][99]