量化点评报告：九月配置建议：利用估值价差定位风格轮动的大周期

Quantitative Models and Construction Valuation Spread-Based Style Factor Model - Model Name: Valuation Spread-Based Style Factor Model - Model Construction Idea: This model uses the valuation spread between long and short groups of a factor (e.g., BP) to construct a style factor's odds indicator. It emphasizes a "left-side" signal characteristic, meaning "buy when it drops significantly, sell when it rises significantly" [7][11] - Model Construction Process: 1. Select a factor and sort stocks by factor values into five groups (industry-neutral, grouping within each industry and then merging) to determine the long and short groups [11] 2. Calculate the median log(BP) for the long and short groups, where log transformation ensures BP follows a normal distribution [11] 3. Compute the raw valuation spread as the difference between the median log(BP) of the long group and the short group [11] 4. Standardize the raw valuation spread using a rolling six-year z-score [11] - Model Evaluation: The valuation spread demonstrates a certain degree of differentiation in predicting the future one-year returns of style factors. It is more suitable for identifying left-side signals [7] --- Quantitative Factors and Construction Dividend Yield Factor - Factor Name: Dividend Yield Factor - Factor Construction Idea: Represents the valuation spread of the dividend yield factor, indicating the relative attractiveness of value-oriented stocks [14] - Factor Construction Process: Constructed using the valuation spread methodology described above, with the dividend yield as the underlying factor [14] PB Factor - Factor Name: PB Factor - Factor Construction Idea: Represents the valuation spread of the price-to-book ratio (PB), reflecting the relative valuation of value-oriented stocks [14] - Factor Construction Process: Constructed using the valuation spread methodology described above, with PB as the underlying factor [14] PE Factor - Factor Name: PE Factor - Factor Construction Idea: Represents the valuation spread of the price-to-earnings ratio (PE), indicating the relative valuation of value-oriented stocks [14] - Factor Construction Process: Constructed using the valuation spread methodology described above, with PE as the underlying factor [14] Quality Factor - Factor Name: Quality Factor - Factor Construction Idea: Represents the valuation spread of the ROE factor, typically associated with "core assets" like high-quality companies [20] - Factor Construction Process: Constructed using the valuation spread methodology described above, with ROE as the underlying factor [20] Low Volatility Factor - Factor Name: Low Volatility Factor - Factor Construction Idea: Represents the valuation spread of low-volatility stocks, often linked to "stable assets" [20] - Factor Construction Process: Constructed using the valuation spread methodology described above, with volatility as the underlying factor [20] Momentum Factor - Factor Name: Momentum Factor - Factor Construction Idea: Represents the valuation spread of momentum stocks, often associated with stocks heavily held by public funds [20] - Factor Construction Process: Constructed using the valuation spread methodology described above, with momentum as the underlying factor [20] --- Backtesting Results of Factors Valuation Spread-Based Style Factor Model - Odds Indicator: Demonstrates differentiation in predicting future one-year returns of style factors, with higher odds indicating better opportunities [7][8] Dividend Yield Factor - Odds: 0.47 standard deviations, categorized as medium odds [14] PB Factor - Odds: 0.63 standard deviations, categorized as medium-high odds [14] PE Factor - Odds: 0.82 standard deviations, categorized as medium-high odds [14] Quality Factor - Odds: 1.17 standard deviations, categorized as high odds [20] Low Volatility Factor - Odds: 1.75 standard deviations, categorized as very high odds [20] Momentum Factor - Odds: -1.36 standard deviations, categorized as low odds [20]