金工定期报告：从微观出发的五维行业轮动月度跟踪202501

• Model Name: Five-dimensional Industry Rotation Model - Model Construction Idea: Based on the Dongwu Financial Engineering multi-factor stock selection system, the model is constructed from the micro-level of individual stocks, utilizing style indicators to classify stocks within the industry[8] - Model Construction Process: The micro factors are divided into five categories: volatility, fundamentals, trading volume, sentiment, and momentum. Using the style preferences of the large category stock selection factors as a reference, industry internal discrete indicators and industry internal traction indicators are constructed, resulting in five types of composite industry factors. The model includes volatility, fundamentals, trading volume, sentiment, and momentum[8] - Model Evaluation: The model effectively captures the style differences within the industry and constructs comprehensive industry factors based on these differences[8] - Model Name: Five-dimensional Industry Rotation Model Performance Tracking - Model Construction Idea: The model's performance is tracked by constructing the model's return capability score and stability score based on the excess returns of the long group and the average industry ranking of the long group[9] - Model Construction Process: The return capability score and stability score are constructed based on the excess returns of the long group and the average industry ranking of the long group from June 1, 2024, to December 31, 2024[9] - Model Evaluation: The momentum factor has performed the best recently[11] - Model Name: Five-dimensional Industry Rotation Model Backtest Performance - Model Construction Idea: The model's performance is tracked by constructing the model's return capability score and stability score based on the excess returns of the long group and the average industry ranking of the long group[9] - Model Construction Process: The model's performance is tracked by constructing the model's return capability score and stability score based on the excess returns of the long group and the average industry ranking of the long group from June 1, 2024, to December 31, 2024[9] - Model Evaluation: The momentum factor has performed the best recently[11] Model Backtest Results - Five-dimensional Industry Rotation Model, annualized return: 22.15%, annualized volatility: 10.92%, IR: 2.03, monthly win rate: 73.68%, maximum historical drawdown: 13.30%[13] - Five-dimensional Industry Rotation Model, long hedge equal-weighted industry portfolio, annualized return: 10.78%, annualized volatility: 6.66%, IR: 1.62, monthly win rate: 71.05%, maximum historical drawdown: 9.36%[17] Factor Backtest Results - Volatility factor, annualized return: 11.70%, annualized volatility: 10.24%, IR: 1.14, win rate: 59.66%, maximum drawdown: 14.27%[14] - Fundamental factor, annualized return: 6.32%, annualized volatility: 10.01%, IR: 0.63, win rate: 55.46%, maximum drawdown: 21.50%[14] - Trading volume factor, annualized return: 7.48%, annualized volatility: 12.31%, IR: 0.61, win rate: 57.98%, maximum drawdown: 17.86%[14] - Sentiment factor, annualized return: 8.77%, annualized volatility: 13.05%, IR: 0.67, win rate: 66.39%, maximum drawdown: 13.56%[14] - Momentum factor, annualized return: 12.25%, annualized volatility: 10.91%, IR: 1.12, win rate: 61.02%, maximum drawdown: 13.52%[14] - Composite factor, annualized return: 22.15%, annualized volatility: 10.92%, IR: 2.03, win rate: 73.68%, maximum drawdown: 13.30%[14] Monthly Performance - December 2024, long portfolio return: 0.71%, short portfolio return: -2.53%, long-short hedge return: 3.24%[22] Index Enhancement Strategy - Model Name: Five-dimensional Industry Rotation Model Index Enhancement Strategy - Model Construction Idea: The model is applied to the index enhancement strategy, using the CSI 300 as an example[26] - Model Construction Process: At the end of each month, the top five industries are selected as enhancement industries, the bottom five industries are excluded, and the remaining industries remain unchanged. The stocks belonging to the excluded industries are removed based on their weight in the CSI 300, and the weights of the excluded industry stocks are proportionally assigned to the enhancement industry stocks, forming a new CSI 300 enhancement portfolio with monthly rebalancing[26] - Model Evaluation: The strategy effectively enhances the CSI 300 index by selecting and excluding industries based on the model's scores[26] Index Enhancement Strategy Backtest Results - Five-dimensional Industry Rotation Model Index Enhancement Strategy, annualized return: 8.16%, annualized volatility: 20.34%, IR: 40.11%, win rate: 56.52%, maximum drawdown: 26.66%[28] - CSI 300, annualized return: -1.16%, annualized volatility: 19.98%, IR: -5.81%, win rate: 53.04%, maximum drawdown: 39.92%[28] - Excess return, annualized return: 9.18%, annualized volatility: 7.59%, IR: 120.94%, win rate: 70.43%, maximum drawdown: 12.74%[28]