Core Viewpoint - The article discusses the style rotation effect in the Hong Kong stock market, utilizing a multi-factor model based on five categories of information, achieving an annualized return of 19.50% and an annualized excess return of 8.98% compared to an equal-weighted benchmark [2][7]. Group 1: Characteristics and Influencing Factors of Hong Kong Stock Style Rotation - The technology and dividend styles in the Hong Kong stock market exhibit a long-term negative correlation in returns, indicating a clear rotation phenomenon, with short-term advantages rotating monthly and long-term advantages lasting over six months [4][11]. - The excess returns of the Hang Seng and CSI Hong Kong Stock Connect technology and dividend indices have maintained a strong positive correlation since mid-2020, while the excess returns of the dividend and technology styles have shown a strong negative correlation, making it suitable for constructing style rotation strategies [4][11]. - The rotation speed between technology and dividend styles is moderate, with short-term advantages rotating monthly and long-term advantages extending beyond six months [4][11]. Group 2: Multi-Factor Style Rotation Model Framework - A ridge logistic regression method with a collinearity penalty term is employed to confirm parameters and factors through rolling regression, ultimately determining the style rotation viewpoint through a simple binary classification [5][6]. - The model selects 101 indicators from five categories to assess the predictive ability of individual indicators, using a two-year rolling window for training [5][25]. - The model generates style holding views based on predicted probabilities, employing both a simple binary classification method and a buffer zone method to manage position adjustments [6][32]. Group 3: Model Performance and Insights - The simple binary classification model achieves an annualized return of 19.50%, with an annualized excess return of 8.98% and a relative volatility of 13.62%, demonstrating significant return enhancement capabilities [7][34]. - The model's average monthly return for predicting technology style outperformance is 1.78%, while for predicting dividend style outperformance, it is 1.17% [7][34]. - The model's performance has shown positive excess returns in most years, with notable years being 2020 and 2024, where excess returns exceeded 20% [7][39]. Group 4: Factors Influencing Style Rotation - Five dimensions are identified as influencing factors for style rotation: macroeconomic, funding, market, fundamental, and valuation indicators, with all indicators standardized to a monthly frequency [19][21]. - Macroeconomic indicators reflect the growth, inflation, and liquidity status of China, Hong Kong, and the U.S., highlighting the interconnectedness of these economies [19][20]. - Funding indicators include net inflows and preferences of southbound funds and global public funds, which are crucial for understanding style performance [21][22]. Group 5: Current Recommendations and Future Outlook - As of September 30, 2025, the latest recommendation is to allocate to the Hong Kong dividend style, primarily due to the negative correlation of selected volatility ratio factors with the probability of technology style outperformance [35][39]. - The analysis indicates that the dividend style has certain advantages in risk aversion, especially as the volatility of the technology style remains higher than that of the dividend style [35][39].