Quantitative Models and Construction Methods 1. Model Name: Maximized Factor Exposure (MFE) Portfolio - **Model Construction Idea**: The MFE portfolio is designed to test the effectiveness of single factors under realistic constraints, such as industry exposure, style exposure, stock weight deviation, and turnover rate. This approach ensures that the factors deemed "effective" can genuinely contribute to return prediction in the final portfolio[38][39]. - **Model Construction Process**: The MFE portfolio is constructed using the following optimization model: $ \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} \\ & \mathbf{0}\leq w\leq l \\ & \mathbf{1}^{T}\ w=1 \end{array} $ - **Objective Function**: Maximize single-factor exposure, where \( f \) represents factor values, \( f^{T}w \) is the weighted exposure, and \( w \) is the stock weight vector. - **Constraints**: 1. **Style Exposure**: \( X \) is the factor exposure matrix, \( w_b \) is the benchmark weight vector, and \( s_l, s_h \) are the lower and upper bounds for style exposure[39]. 2. **Industry Exposure**: \( H \) is the industry exposure matrix, and \( h_l, h_h \) are the lower and upper bounds for industry deviation[39]. 3. **Stock Weight Deviation**: \( w_l, w_h \) are the lower and upper bounds for stock weight deviation[39]. 4. **Constituent Weight Control**: \( B_b \) is a 0-1 vector indicating benchmark constituents, and \( b_l, b_h \) are the lower and upper bounds for constituent weights[39]. 5. **No Short Selling**: Ensures non-negative weights and limits individual stock weights[39]. 6. **Full Investment**: Ensures the portfolio is fully invested (\( \mathbf{1}^{T}w = 1 \))[40]. - **Implementation**: - Constraints are set monthly, and the MFE portfolio is rebalanced accordingly. - Historical returns are calculated, and transaction costs of 0.3% (double-sided) are deducted to evaluate the portfolio's performance relative to the benchmark[42]. - **Model Evaluation**: The MFE portfolio effectively identifies factors that can predict returns under realistic constraints, making it a robust tool for factor validation[38][39]. --- Quantitative Factors and Construction Methods 1. Factor Name: Standardized Unexpected Earnings (SUE) - **Factor Construction Idea**: Measures the deviation of actual earnings from expected earnings, standardized by the standard deviation of expected earnings, to capture earnings surprises[15]. - **Factor Construction Process**: $ SUE = \frac{\text{Actual Net Profit} - \text{Expected Net Profit}}{\text{Standard Deviation of Expected Net Profit}} $ - **Parameters**: - Actual Net Profit: Reported quarterly net profit. - Expected Net Profit: Consensus analyst forecast for the quarter. - Standard Deviation: Variability in analyst forecasts[15]. 2. Factor Name: Delta ROA (DELTAROA) - **Factor Construction Idea**: Tracks the change in return on assets (ROA) compared to the same quarter in the previous year to capture profitability trends[15]. - **Factor Construction Process**: $ \Delta ROA = \text{ROA}_{\text{current quarter}} - \text{ROA}_{\text{same quarter last year}} $ - **Parameters**: - ROA: \( \frac{\text{Net Income} \times 2}{\text{Average Total Assets}} \)[15]. 3. Factor Name: Standardized Unexpected Revenue (SUR) - **Factor Construction Idea**: Measures the deviation of actual revenue from expected revenue, standardized by the standard deviation of expected revenue, to capture revenue surprises[15]. - **Factor Construction Process**: $ SUR = \frac{\text{Actual Revenue} - \text{Expected Revenue}}{\text{Standard Deviation of Expected Revenue}} $ - **Parameters**: - Actual Revenue: Reported quarterly revenue. - Expected Revenue: Consensus analyst forecast for the quarter. - Standard Deviation: Variability in analyst forecasts[15]. --- Factor Backtesting Results 1. **Performance in CSI 300 Universe** - **Top-Performing Factors (1 Week)**: Single-quarter ROA (1.09%), Standardized Unexpected Revenue (0.73%), Single-quarter Revenue Growth (0.71%)[17]. - **Underperforming Factors (1 Week)**: Specificity (-0.93%), 3-Month Reversal (-0.53%), 1-Month Volatility (-0.46%)[17]. 2. **Performance in CSI 500 Universe** - **Top-Performing Factors (1 Week)**: Standardized Unexpected Revenue (1.07%), Single-quarter Net Profit Growth (1.00%), Standardized Unexpected Earnings (0.99%)[19]. - **Underperforming Factors (1 Week)**: 3-Month Volatility (-1.08%), BP (-0.28%), 1-Month Volatility (-1.14%)[19]. 3. **Performance in CSI 1000 Universe** - **Top-Performing Factors (1 Week)**: Standardized Unexpected Revenue (1.07%), Standardized Unexpected Earnings (1.00%), Single-quarter Revenue Growth (0.90%)[21]. - **Underperforming Factors (1 Week)**: 1-Month Volatility (-1.14%), 3-Month Volatility (-1.08%), 3-Month Reversal (-1.02%)[21]. 4. **Performance in CSI A500 Universe** - **Top-Performing Factors (1 Week)**: Single-quarter ROA (1.14%), Delta ROA (1.12%), Delta ROE (1.02%)[23]. - **Underperforming Factors (1 Week)**: Specificity (-0.65%), Non-Liquidity Shock (-0.64%), 1-Month Volatility (-0.62%)[23]. 5. **Performance in Public Fund Heavyweight Index** - **Top-Performing Factors (1 Week)**: Delta ROA (1.12%), Expected PEG (0.94%), Standardized Unexpected Earnings (0.99%)[25]. - **Underperforming Factors (1 Week)**: 3-Month Volatility (-0.60%), 1-Month Volatility (-0.62%), 1-Month Reversal (-0.37%)[25].

- The report tracks the performance of Guosen JinGong's index enhancement portfolios and public fund index enhancement products, alongside monitoring the performance of common stock selection factors across different stock selection spaces [12][13][16] - Guosen JinGong's index enhancement portfolios are constructed based on three main components: return prediction, risk control, and portfolio optimization. These portfolios are benchmarked against indices such as CSI 300, CSI 500, CSI 1000, and CSI A500 [13][15] - The MFE (Maximized Factor Exposure) portfolio is used to test the effectiveness of individual factors under real-world constraints. The optimization model maximizes single-factor exposure while controlling for style, industry, stock weight deviations, and other constraints. The formula for the optimization model is: $\begin{array}{ll}max&f^{T}\ w\\ s.t.&s_{l}\leq X(w-w_{b})\leq s_{h}\\ &h_{l}\leq H(w-w_{b})\leq h_{h}\\ &w_{l}\leq w-w_{b}\leq w_{h}\\ &b_{l}\leq B_{b}w\leq b_{h}\\ &\mathbf{0}\leq w\leq l\\ &\mathbf{1}^{T}\ w=1\end{array}$ where `f` represents factor values, `w` is the stock weight vector, and constraints include style exposure (`X`), industry exposure (`H`), stock weight deviation (`w`), and component stock weight limits (`B_b`) [40][41][42] - The factor library includes over 30 factors categorized into valuation, reversal, growth, profitability, liquidity, corporate governance, and analyst dimensions. Examples include BP (Net Asset/Market Cap), single-quarter EP (Net Profit/Market Cap), and EPTTM (TTM Net Profit/Market Cap) [17][18] - Factor performance varies across different stock selection spaces. For CSI 300, factors like single-quarter EP, EPTTM, and expected EPTTM performed well recently, while factors like three-month volatility and expected net profit QoQ performed poorly [19][20] - For CSI 500, factors such as single-quarter ROE, DELTAROE, and single-quarter EP showed strong performance recently, whereas factors like one-year momentum and three-month reversal underperformed [21][22] - In the CSI 1000 space, factors like standardized unexpected earnings, EPTTM, and single-quarter EP performed well, while factors like non-liquidity impact and three-month institutional coverage lagged [23][24] - For CSI A500, factors such as expected EPTTM, single-quarter ROE, and expected PEG showed strong performance, while factors like one-year momentum and expected net profit QoQ underperformed [25][26] - In the public fund heavy index space, factors like expected PEG, expected EPTTM, and single-quarter EP performed well recently, while factors like one-month reversal and one-month volatility performed poorly [27][28] - Public fund index enhancement products are tracked for their excess returns relative to benchmarks. For CSI 300 products, the highest weekly excess return was 1.02%, and the lowest was -0.37%, with a median of 0.08% [29][33] - CSI 500 products showed a weekly excess return range of 1.87% to -0.44%, with a median of 0.38% [34][35] - CSI 1000 products had a weekly excess return range of 1.06% to -0.43%, with a median of 0.38% [36][37] - CSI A500 products showed a weekly excess return range of 0.73% to -0.19%, with a median of 0.17% [38][39]