Quantitative Factors and Models Summary Quantitative Factors and Their Construction - **Factor Name**: Standardized Unexpected Revenue (SUR) - **Construction Idea**: Measures the deviation of actual revenue from analysts' expectations, standardized by the standard deviation of expected revenue[20][27][31] - **Construction Process**: $ SUR = \frac{Actual\ Revenue - Expected\ Revenue}{Standard\ Deviation\ of\ Expected\ Revenue} $ - The numerator represents the difference between actual and expected revenue - The denominator is the standard deviation of expected revenue, ensuring comparability across stocks[20][27][31] - **Evaluation**: Demonstrated strong performance across multiple indices, indicating its effectiveness in capturing unexpected revenue trends[8][27][31] - **Factor Name**: Delta ROA - **Construction Idea**: Tracks the year-over-year change in Return on Assets (ROA) to capture profitability trends[20][31][39] - **Construction Process**: $ \Delta ROA = ROA_{Current\ Quarter} - ROA_{Same\ Quarter\ Last\ Year} $ - ROA is calculated as $ \frac{Net\ Income}{Total\ Assets} $ - The factor highlights improvements or deteriorations in asset efficiency[20][31][39] - **Evaluation**: Consistently strong performance, particularly in small-cap indices like the CSI 1000 and CSI 2000, suggesting its relevance in growth-oriented stocks[8][39][43] - **Factor Name**: Standardized Unexpected Earnings (SUE) - **Construction Idea**: Similar to SUR, measures the deviation of actual earnings from analysts' expectations, standardized by the standard deviation of expected earnings[20][31][39] - **Construction Process**: $ SUE = \frac{Actual\ Earnings - Expected\ Earnings}{Standard\ Deviation\ of\ Expected\ Earnings} $ - The numerator captures the earnings surprise - The denominator ensures standardization for comparability[20][31][39] - **Evaluation**: Strong performance in indices like CSI 500 and CSI 800, indicating its ability to capture earnings surprises effectively[8][27][31] - **Factor Name**: Delta ROE - **Construction Idea**: Measures the year-over-year change in Return on Equity (ROE) to identify shifts in shareholder profitability[20][31][39] - **Construction Process**: $ \Delta ROE = ROE_{Current\ Quarter} - ROE_{Same\ Quarter\ Last\ Year} $ - ROE is calculated as $ \frac{Net\ Income}{Shareholders'\ Equity} $ - Highlights changes in equity efficiency over time[20][31][39] - **Evaluation**: Demonstrated strong performance in growth-oriented indices, particularly the CSI 1000 and Growth Enterprise Market (GEM) indices[8][39][43] Factor Backtesting Results - **Standardized Unexpected Revenue (SUR)** - CSI 500: Weekly return 1.43%, monthly return 1.66%, annualized return 12.83%[27] - CSI 800: Weekly return 1.36%, monthly return 2.61%, annualized return 4.26%[31] - CSI All Share: Weekly return 1.37%, monthly return 1.95%, annualized return 6.91%[47] - **Delta ROA** - CSI 1000: Weekly return 0.56%, monthly return 1.67%, annualized return 11.51%[35] - CSI 2000: Weekly return 1.90%, monthly return 1.90%, annualized return 27.67%[39] - CSI All Share: Weekly return 1.10%, monthly return 2.33%, annualized return 7.78%[47] - **Standardized Unexpected Earnings (SUE)** - CSI 500: Weekly return 1.39%, monthly return 2.75%, annualized return 7.19%[27] - CSI 800: Weekly return 0.52%, monthly return 1.33%, annualized return 3.04%[31] - CSI All Share: Weekly return 1.09%, monthly return 2.46%, annualized return 0.72%[47] - **Delta ROE** - CSI 1000: Weekly return 0.30%, monthly return 1.59%, annualized return 8.89%[35] - CSI 2000: Weekly return 1.24%, monthly return 1.21%, annualized return 90.84%[39] - GEM: Weekly return 1.03%, monthly return 2.76%, annualized return 21.85%[43] Quantitative Model Construction - **Model Name**: Maximized Factor Exposure (MFE) Portfolio - **Construction Idea**: Constructs portfolios that maximize exposure to a single factor while controlling for industry, style, and stock-specific constraints[62][63][66] - **Construction Process**: $ \begin{array}{ll} max & f^{T}w \\ s.t. & s_{l} \leq X(w-w_{b}) \leq s_{h} \\ & h_{l} \leq H(w-w_{b}) \leq h_{h} \\ & w_{l} \leq w-w_{b} \leq w_{h} \\ & b_{l} \leq B_{b}w \leq b_{h} \\ & 0 \leq w \leq l \\ & 1^{T}w = 1 \\ & \Sigma|w-w_{0}| \leq to_{h} \end{array} $ - Maximizes factor exposure $ f^{T}w $ - Constraints include style, industry, stock-specific deviations, and turnover limits[62][63][66] - **Evaluation**: Effective in isolating factor performance under realistic portfolio constraints, widely used in index enhancement strategies[62][63][66] Model Backtesting Results - **MFE Portfolio** - CSI 300: Weekly excess return max 1.67%, min -0.65%, median 0.21%[54] - CSI 500: Weekly excess return max 1.13%, min -0.76%, median 0.24%[57] - CSI 1000: Weekly excess return max 1.11%, min -0.52%, median 0.24%[61]