Quantitative Models and Construction Methods - **Model Name**: Maximized Factor Exposure (MFE) Portfolio **Model Construction Idea**: The MFE portfolio aims to maximize the exposure of a single factor while controlling for constraints such as industry exposure, style exposure, stock weight deviation, and turnover rate. This approach evaluates the effectiveness of factors under realistic constraints in enhanced index portfolios [56][57][59] **Model Construction Process**: The optimization model is formulated as follows: $ \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} $ - **Objective Function**: Maximize single-factor exposure, where \( f \) represents factor values, and \( w \) is the stock weight vector - **Constraints**: 1. Style exposure deviation (\( X \)): \( s_{l} \) and \( s_{h} \) are the lower and upper bounds for style factor deviation 2. Industry exposure deviation (\( H \)): \( h_{l} \) and \( h_{h} \) are the lower and upper bounds for industry deviation 3. Stock weight deviation (\( w_{l} \) and \( w_{h} \)): Limits on individual stock weight deviation relative to the benchmark 4. Component weight limits (\( b_{l} \) and \( b_{h} \)): Constraints on the weight of benchmark components 5. No short selling and upper limits on stock weights 6. Full investment constraint: \( 1^{T}w=1 \) 7. Turnover constraint: \( \Sigma|w-w_{0}|\leq to_{h} \), where \( w_{0} \) is the previous period's weight [56][57][59] **Model Evaluation**: The model effectively balances factor exposure and practical constraints, ensuring stable returns and avoiding excessive concentration in specific stocks [60] --- Quantitative Factors and Construction Methods - **Factor Name**: Six-Month UMR **Factor Construction Idea**: The six-month UMR factor measures risk-adjusted momentum over a six-month window, capturing medium-term momentum trends [19][8][44] **Factor Construction Process**: - The UMR (Up-Minus-Down Ratio) is calculated as the ratio of upward movements to downward movements in stock prices over a specified period - The six-month UMR specifically uses a six-month window to compute this ratio, adjusted for risk [19][8][44] **Factor Evaluation**: This factor demonstrates strong performance in various index spaces, particularly in the CSI 500 and CSI All Share indices, indicating its effectiveness in capturing medium-term momentum [8][44] - **Factor Name**: Three-Month UMR **Factor Construction Idea**: Similar to the six-month UMR, this factor focuses on shorter-term momentum trends over a three-month window [19][8][44] **Factor Construction Process**: - The three-month UMR is calculated using the same methodology as the six-month UMR but with a three-month window for data aggregation [19][8][44] **Factor Evaluation**: This factor shows consistent performance across multiple indices, including the CSI 500 and CSI All Share indices, making it a reliable short-term momentum indicator [8][44] - **Factor Name**: Pre-Tax Earnings to Total Market Value (EPTTM) **Factor Construction Idea**: This valuation factor evaluates the earnings yield of a stock, providing insights into its relative valuation [19][8][44] **Factor Construction Process**: - EPTTM is calculated as the ratio of pre-tax earnings to the total market value of a stock, with adjustments for rolling time windows (e.g., one year) [19][8][44] **Factor Evaluation**: EPTTM consistently ranks among the top-performing valuation factors, particularly in the CSI 300 and CSI 800 indices, reflecting its robustness in identifying undervalued stocks [8][44] --- Backtesting Results of Models - **MFE Portfolio**: - The MFE portfolio demonstrates strong performance under various constraints, with backtesting results showing significant alpha generation relative to benchmarks like CSI 300, CSI 500, and CSI 1000 [60][61] --- Backtesting Results of Factors - **Six-Month UMR**: - CSI 500: Weekly return of 0.99%, monthly return of 1.65%, annualized return of -4.07% [26] - CSI All Share: Weekly return of 1.23%, monthly return of 1.59%, annualized return of 7.43% [44] - **Three-Month UMR**: - CSI 500: Weekly return of 0.94%, monthly return of 1.31%, annualized return of 0.68% [26] - CSI All Share: Weekly return of 1.02%, monthly return of 1.63%, annualized return of 5.64% [44] - **EPTTM**: - CSI 300: Weekly return of 0.74%, monthly return of 1.42%, annualized return of 3.89% [22] - CSI 800: Weekly return of 1.00%, monthly return of 1.91%, annualized return of 2.87% [30]

Quantitative Factors and Construction Methods - **Factor Name**: Single-quarter ROE **Construction Idea**: This factor measures the return on equity (ROE) for a single quarter, reflecting the profitability of a company relative to its equity base[2][18] **Construction Process**: The formula for single-quarter ROE is: $ Quart\_ROE = \frac{Net\ Income \times 2}{Beginning\ Equity + Ending\ Equity} $ Here, "Net Income" represents the net profit for the quarter, and "Beginning Equity" and "Ending Equity" are the equity values at the start and end of the quarter, respectively[18] **Evaluation**: This factor performed well in the CSI All Share Index space during the past week, indicating its effectiveness in identifying profitable stocks[2][42] - **Factor Name**: Single-quarter ROA **Construction Idea**: This factor evaluates the return on assets (ROA) for a single quarter, assessing how efficiently a company utilizes its assets to generate profits[18] **Construction Process**: The formula for single-quarter ROA is: $ Quart\_ROA = \frac{Net\ Income \times 2}{Beginning\ Assets + Ending\ Assets} $ "Net Income" is the quarterly net profit, while "Beginning Assets" and "Ending Assets" are the total assets at the start and end of the quarter, respectively[18] **Evaluation**: This factor also demonstrated strong performance in the CSI All Share Index space over the past week, highlighting its utility in asset efficiency analysis[2][42] - **Factor Name**: Standardized Unexpected Earnings (SUE) **Construction Idea**: This factor captures the deviation of actual earnings from expected earnings, standardized by the standard deviation of expected earnings, to measure earnings surprises[18] **Construction Process**: The formula for SUE is: $ SUE = \frac{Actual\ Earnings - Expected\ Earnings}{Standard\ Deviation\ of\ Expected\ Earnings} $ "Actual Earnings" refers to the reported earnings, while "Expected Earnings" and their standard deviation are derived from analyst forecasts[18] **Evaluation**: This factor showed significant positive performance in the National SME Index (CSI 2000) and the ChiNext Index spaces, indicating its effectiveness in identifying earnings surprises[36][39] Factor Backtesting Results - **Single-quarter ROE**: - CSI All Share Index: Weekly return of 1.46%, monthly return of 1.95%, annualized return over the past year of -1.73%, and historical annualized return of 4.88%[42][43] - **Single-quarter ROA**: - CSI All Share Index: Weekly return of 1.09%, monthly return of 1.33%, annualized return over the past year of 0.27%, and historical annualized return of 4.14%[42][43] - **Standardized Unexpected Earnings (SUE)**: - National SME Index (CSI 2000): Weekly return of 6.41%, monthly return of 19.22%, annualized return over the past year of 32.33%, and historical annualized return of 10.98%[36] - ChiNext Index: Weekly return of 7.76%, monthly return of 26.34%, annualized return over the past year of 44.74%, and historical annualized return of 7.82%[39] Composite Factor Portfolio Construction - **MFE Portfolio Construction**: **Idea**: The Maximized Factor Exposure (MFE) portfolio is designed to maximize the exposure to a single factor while controlling for constraints such as industry and style exposures, stock weight deviations, and turnover[55][59] **Optimization Model**: The optimization problem is formulated as: $ \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} $ Here, $f$ represents the factor values, $w$ is the weight vector, and the constraints include style, industry, stock weight, and turnover limits[55][58] **Evaluation**: The MFE portfolio approach ensures that factor effectiveness is tested under realistic constraints, making it a robust method for evaluating factor performance[55][59] MFE Portfolio Backtesting Results - **CSI 300 Index**: - Weekly excess return: Maximum 1.05%, minimum -0.81%, median 0.00%[46][49] - Monthly excess return: Maximum 3.00%, minimum -1.15%, median 0.30%[46][49] - **CSI 500 Index**: - Weekly excess return: Maximum 1.00%, minimum -0.08%, median 0.40%[50][52] - Monthly excess return: Maximum 2.73%, minimum -0.42%, median 0.99%[50][52] - **CSI 1000 Index**: - Weekly excess return: Maximum 0.82%, minimum -0.26%, median 0.28%[53][54] - Monthly excess return: Maximum 3.52%, minimum -0.08%, median 1.72%[53][54]