Quantitative Factors and Construction Factor Name: GAN_GRU Factor - **Construction Idea**: The GAN_GRU factor is derived by processing volume-price time-series features using a Generative Adversarial Network (GAN) model, followed by encoding these time-series features with a Gated Recurrent Unit (GRU) model to generate a stock selection factor [4][13][41] - **Construction Process**: 1. **Input Features**: 18 volume-price features such as closing price, opening price, turnover, and turnover rate are used as input data. These features are sampled every 5 trading days over the past 400 days, resulting in a feature matrix of shape (40,18) [14][17][18] 2. **Data Preprocessing**: - Outlier removal and standardization are applied to each feature over the 40-day time series - Cross-sectional standardization is performed at the stock level [18] 3. **GAN Model**: - **Generator**: An LSTM-based generator is used to preserve the sequential nature of the input features. The generator takes random noise (e.g., Gaussian distribution) as input and generates data that mimics the real data distribution [23][33][37] - **Discriminator**: A CNN-based discriminator is employed to classify real and generated data. The discriminator uses convolutional layers to extract features from the 2D volume-price time-series "images" [33][35] - **Loss Functions**: - Generator Loss: $$ L_{G} = -\mathbb{E}_{z\sim P_{z}(z)}[\log(D(G(z)))] $$ where \( z \) represents random noise, \( G(z) \) is the generated data, and \( D(G(z)) \) is the discriminator's output probability for the generated data being real [24] - Discriminator Loss: $$ L_{D} = -\mathbb{E}_{x\sim P_{data}(x)}[\log D(x)] - \mathbb{E}_{z\sim P_{z}(z)}[\log(1-D(G(z)))] $$ where \( x \) is real data, \( D(x) \) is the discriminator's output probability for real data, and \( D(G(z)) \) is the discriminator's output probability for generated data [27] 4. **GRU Model**: - Two GRU layers (GRU(128,128)) are used to encode the time-series features, followed by an MLP (256,64,64) to predict future returns [22] 5. **Factor Output**: The predicted returns (\( pRet \)) from the GRU+MLP model are used as the stock selection factor. The factor is neutralized for industry and market capitalization effects and standardized [22] Factor Evaluation - The GAN_GRU factor effectively captures the sequential and cross-sectional characteristics of volume-price data, leveraging the strengths of GANs for feature generation and GRUs for time-series encoding [4][13][41] --- Factor Backtesting Results GAN_GRU Factor Performance Metrics - **IC Mean**: 11.43% (2019-2025), 10.97% (last year), 9.27% (latest month) [41][42] - **ICIR**: 0.89 [42] - **Turnover Rate**: 0.82 [42] - **Annualized Return**: 38.52% [42] - **Annualized Volatility**: 23.82% [42] - **IR**: 1.62 [42] - **Maximum Drawdown**: 27.29% [42] - **Annualized Excess Return**: 24.86% [41][42] GAN_GRU Factor Industry Performance - **Top 5 Industries by IC (Latest Month)**: - Home Appliances: 27.00% - Non-Bank Financials: 23.08% - Retail: 20.01% - Steel: 14.83% - Textiles & Apparel: 13.64% [41][42] - **Top 5 Industries by IC (Last Year)**: - Utilities: 14.43% - Retail: 13.33% - Non-Bank Financials: 13.28% - Steel: 13.23% - Telecommunications: 12.36% [41][42] GAN_GRU Factor Long Portfolio Performance - **Top 5 Industries by Excess Return (Latest Month)**: - Textiles & Apparel: 5.19% - Utilities: 3.62% - Automobiles: 3.29% - Non-Bank Financials: 2.56% - Pharmaceuticals: 1.47% [2][43] - **Top 5 Industries by Average Monthly Excess Return (Last Year)**: - Home Appliances: 5.44% - Building Materials: 4.70% - Textiles & Apparel: 4.19% - Agriculture: 4.09% - Utilities: 3.92% [2][43]

Quantitative Models and Construction Methods - **Model Name**: Value Smart Beta Portfolio **Model Construction Idea**: The model selects stocks based on the value style, aiming for high beta elasticity and long-term stable excess returns[8] **Model Construction Process**: The Value Smart Beta Portfolio includes two sub-portfolios: Value 50 Portfolio and Value Balanced 50 Portfolio. These portfolios are constructed by selecting stocks with low historical correlation and aligning with the value style. The detailed construction process is based on the methodology outlined in the October 5, 2024, report on Smart Beta portfolio construction[8] - **Model Name**: Growth Smart Beta Portfolio **Model Construction Idea**: The model focuses on the growth style, targeting high beta elasticity and long-term stable excess returns[8] **Model Construction Process**: The Growth Smart Beta Portfolio includes two sub-portfolios: Growth 50 Portfolio and Growth Balanced 50 Portfolio. Stocks are selected based on their alignment with the growth style and low historical correlation. The methodology follows the principles outlined in the October 5, 2024, report on Smart Beta portfolio construction[8] - **Model Name**: Small-Cap Smart Beta Portfolio **Model Construction Idea**: The model emphasizes the small-cap style, aiming for high beta elasticity and long-term stable excess returns[8] **Model Construction Process**: The Small-Cap Smart Beta Portfolio includes two sub-portfolios: Small-Cap 50 Portfolio and Small-Cap Balanced 50 Portfolio. Stocks are chosen based on their small-cap characteristics and low historical correlation. The construction process adheres to the methodology described in the October 5, 2024, report on Smart Beta portfolio construction[8] Model Backtesting Results - **Value Smart Beta Portfolio** - **Value 50 Portfolio**: - Weekly Absolute Return: 0.09% - Weekly Excess Return: -1.24% - Monthly Absolute Return: 3.51% - Monthly Excess Return: 0.03% - Year-to-Date Absolute Return: 14.88% - Year-to-Date Excess Return: 9.41% - Maximum Relative Drawdown: 2.34%[9] - **Value Balanced 50 Portfolio**: - Weekly Absolute Return: 1.72% - Weekly Excess Return: 0.39% - Monthly Absolute Return: 5.31% - Monthly Excess Return: 1.82% - Year-to-Date Absolute Return: 10.67% - Year-to-Date Excess Return: 5.20% - Maximum Relative Drawdown: 3.99%[9] - **Growth Smart Beta Portfolio** - **Growth 50 Portfolio**: - Weekly Absolute Return: 1.67% - Weekly Excess Return: -1.12% - Monthly Absolute Return: 5.00% - Monthly Excess Return: -1.20% - Year-to-Date Absolute Return: 6.07% - Year-to-Date Excess Return: 1.69% - Maximum Relative Drawdown: 3.61%[9] - **Growth Balanced 50 Portfolio**: - Weekly Absolute Return: 1.19% - Weekly Excess Return: -1.60% - Monthly Absolute Return: 2.09% - Monthly Excess Return: -4.12% - Year-to-Date Absolute Return: 10.52% - Year-to-Date Excess Return: 6.14% - Maximum Relative Drawdown: 6.11%[9] - **Small-Cap Smart Beta Portfolio** - **Small-Cap 50 Portfolio**: - Weekly Absolute Return: 1.30% - Weekly Excess Return: -0.82% - Monthly Absolute Return: 9.84% - Monthly Excess Return: 4.30% - Year-to-Date Absolute Return: 34.84% - Year-to-Date Excess Return: 18.00% - Maximum Relative Drawdown: 6.23%[9] - **Small-Cap Balanced 50 Portfolio**: - Weekly Absolute Return: 1.82% - Weekly Excess Return: -0.29% - Monthly Absolute Return: 5.54% - Monthly Excess Return: 0.00% - Year-to-Date Absolute Return: 27.99% - Year-to-Date Excess Return: 11.15% - Maximum Relative Drawdown: 4.56%[9]

1. Report Industry Investment Rating - No relevant content provided 2. Core View of the Report - This week, the market sentiment recovered, with the Sci - tech Innovation Board leading the rise. Small and micro - cap stocks remained active, and dividend assets with relatively strong defensive attributes also achieved positive returns. Among the dividend sub - sectors, the central and state - owned enterprise dividend series index had a more prominent performance, with an average increase of about 2.44%. The A - share industries continued to diverge, and the commodity market was strong under the "anti - involution" market. In the electronics sector, semiconductor equipment and optical components led the gains. The dividend and electronics enhanced portfolios had weak excess performance, but the electronics balanced allocation enhanced portfolio outperformed the median return of active technology funds [2][7]. 3. Summary by Related Catalogs 3.1 Introduction of Active Quantitative Products - Since July 2023, the Yangtze River Quantitative Finance team has launched multiple active quantitative products such as the dividend selection strategy and the industry high - win - rate strategy. The active quantitative product weekly report is launched to track the performance of active quantitative strategies, including new strategy releases and the return performance of existing strategies [6][13]. 3.2 Strategy Tracking 3.2.1 Dividend Series - Market performance: The market sentiment recovered, with the Sci - tech Innovation 50 and the Sci - tech Innovation Composite Index rising about 4.63% and 3.95% respectively this week. Small and micro - cap stocks were active. Dividend assets achieved positive returns, and the central and state - owned enterprise dividend series index had an average increase of about 2.44%. - Strategy performance: Although the central and state - owned enterprise high - dividend 30 portfolio achieved positive returns, affected by the cyclical product market, both dividend portfolios failed to outperform the CSI Dividend Total Return Index. Since the beginning of 2025, the offensive and defensive dividend 50 portfolio has an excess return of about 1.91% and ranks at about the 44th percentile among all dividend - type funds [7][15][21]. 3.2.2 Electronics Series - Market performance: A - share industries continued to diverge. The commodity market was strong, with raw materials and energy rising about 5.25% and 4.97% respectively. The public utilities and financial sectors significantly corrected. In the electronics sector, semiconductor equipment and optical components rose about 6.59% and 5.10% respectively, far ahead of other sub - tracks. - Strategy performance: This week, the electronics balanced allocation enhanced portfolio achieved positive returns and outperformed the median return of active technology funds, but both electronics portfolios failed to outperform the electronics total return index. Since the beginning of 2025, both portfolios have outperformed the electronics industry index, with excess returns of about 1.96% and 2.98% for the electronics balanced allocation enhanced portfolio and the electronics sector preferred enhanced portfolio respectively [7][24][31].

Quantitative Models and Construction Methods 1. Model Name: Interest Rate Price-Volume Multi-Cycle Timing Strategy - **Model Construction Idea**: This strategy uses kernel regression algorithms to identify support and resistance lines in interest rate trends. It combines signals from long, medium, and short investment cycles to form a composite timing view[10][22] - **Model Construction Process**: 1. **Signal Generation**: - Use kernel regression to capture the shape of interest rate trends and identify breakout signals for long, medium, and short cycles[10] - Long-cycle signals switch monthly, medium-cycle signals switch bi-weekly, and short-cycle signals switch weekly[10] 2. **Portfolio Allocation Rules**: - If at least two cycles show downward breakouts and the trend is not upward, allocate fully to long-duration bonds - If at least two cycles show downward breakouts but the trend is upward, allocate 50% to medium-duration bonds and 50% to long-duration bonds - If at least two cycles show upward breakouts and the trend is not downward, allocate fully to short-duration bonds - If at least two cycles show upward breakouts but the trend is downward, allocate 50% to medium-duration bonds and 50% to short-duration bonds - In other cases, allocate equally across short, medium, and long durations[22] 3. **Benchmark**: Equal-weighted duration strategy (1/3 short, 1/3 medium, 1/3 long duration)[22] 4. **Stop-Loss Mechanism**: Adjust to equal-weight allocation if daily excess return falls below -0.5%[22] - **Model Evaluation**: The strategy demonstrates strong robustness with consistent positive returns and high win rates over an 18-year backtest period[22][23] --- Model Backtest Results 1. Interest Rate Price-Volume Multi-Cycle Timing Strategy - **Long-Term Performance (2007.12.31 to Latest Report Date)**: - Annualized Return: 6.15% - Maximum Drawdown: 1.52% - Return-to-Drawdown Ratio: 2.25 - Excess Annualized Return: 1.66% (relative to equal-weighted duration benchmark) - Excess Return-to-Drawdown Ratio: 1.17[22] - **Short-Term Performance (Since 2023 End)**: - Annualized Return: 6.93% - Maximum Drawdown: 1.52% - Return-to-Drawdown Ratio: 5.94 - Excess Annualized Return: 2.2% - Excess Return-to-Drawdown Ratio: 2.31[22][23] - **Win Rates (2007-2025)**: - Annual Absolute Return > 0: 100% - Annual Excess Return > 0: 100%[23] - **Yearly Performance Statistics**: - Example Years: - 2008: Absolute Return 17.08%, Excess Return 4.41% - 2014: Absolute Return 13.47%, Excess Return 2.67% - 2024: Absolute Return 9.35%, Excess Return 2.52%[26] --- Quantitative Factors and Construction Methods 1. Factor Name: Interest Rate Structure Indicators (Level, Slope, Convexity) - **Factor Construction Idea**: Transform yield-to-maturity (YTM) data of 1-10 year government bonds into structural indicators to analyze the interest rate market from a mean-reversion perspective[7][9] - **Factor Construction Process**: 1. **Level Structure**: Average YTM across all maturities 2. **Slope Structure**: Difference between long-term and short-term YTM 3. **Convexity Structure**: Second derivative of the yield curve to measure curvature[7][9] - **Factor Evaluation**: The current readings indicate a low level structure, low slope structure, and neutral-to-low convexity structure, suggesting a relatively bearish outlook for the interest rate market[9] --- Factor Backtest Results 1. Interest Rate Structure Indicators - **Current Readings**: - Level Structure: 1.6% (17th percentile over 3 years, 10th percentile over 5 years, 5th percentile over 10 years) - Slope Structure: 0.35% (18th percentile over 3 years, 11th percentile over 5 years, 14th percentile over 10 years) - Convexity Structure: 0.09% (32nd percentile over 3 years, 21st percentile over 5 years, 21st percentile over 10 years)[9]

Key Insights - The report highlights a focus on the humanoid robot industry, which has seen a significant increase in market attention, with the National Securities Robot Industry Index rising by 7.6% from July 7 to July 18, 2025 [6][8] - Major factors driving this resurgence include substantial orders from leading companies, capital acquisitions, influential statements from industry leaders, and supportive government policies aimed at fostering innovation in humanoid robotics [7][8] - The report also notes that the active equity fund median position reached 90.63% in Q2 2025, indicating a historical high and a shift towards increased allocations in TMT, Hong Kong stocks, and machinery sectors [9][10] Humanoid Robot Industry - The humanoid robot market is experiencing a revival, with key players like China Mobile placing significant orders, which serve as a validation of product functionality and market readiness [6][7] - The report identifies a trend of increased capital activity, with companies pursuing mergers and acquisitions to enhance their market positions [7] - Government initiatives are also playing a crucial role, with policies aimed at promoting the development of humanoid robots and related technologies [8] Active Equity Fund Analysis - The report indicates that the highest allocation sectors for active equity funds in Q2 2025 were TMT (23.37%), Hong Kong stocks (20.41%), and machinery (19.68%), reflecting a strategic shift in investment focus [9][10] - The report emphasizes that the current allocation levels are above historical averages for several sectors, indicating a bullish sentiment among fund managers [9][10] AI Computing Industry - The AI computing supply chain is entering a phase of maturity, driven by advancements in generative AI and large language models, leading to a closure of the demand-supply loop [11][12] - The report highlights that the infrastructure for AI computing is expected to see continued investment, with significant growth in demand for high-end AI servers [12][13] - The competition in the PCB industry is intensifying due to the rising demand for AI servers, with a projected 150% increase in demand for high-density interconnect (HDI) boards [13]

Group 1: Core Insights - The report introduces a quantitative model solution for addressing the value-growth style switching issue based on odds and win rates [1][8] - The latest growth style investment expectation is calculated at 0.14, while the value style investment expectation is at -0.04, recommending a shift towards growth style [4][18] Group 2: Odds - The estimated odds for the growth style is 1.11, while for the value style it is 1.08, indicating a negative correlation between relative valuation levels and expected odds [2][14] - The report emphasizes that the relative valuation level of market styles is a key influencing factor for expected odds [2][14] Group 3: Win Rates - Among seven win rate indicators, four point towards growth and three towards value, resulting in a current win rate of 53.87% for growth and 46.13% for value [3][16][17] Group 4: Investment Expectations and Strategy Returns - The annualized return of the style rotation model strategy from 2013 to present is 27.35%, with a Sharpe ratio of 1.01 [4][19] - The total return for the growth style is 544.78%, while for the value style it is 605.02%, indicating a strong performance of both styles [19]

Quantitative Models and Construction Methods 1. Model Name: Dividend Forecast Model - **Model Construction Idea**: The model aims to predict the impact of dividends on index futures contracts by estimating the dividend points based on historical and current financial data of index constituent stocks[7][18][21] - **Model Construction Process**: 1. **Estimate Net Profit**: Use annual reports, earnings forecasts, and other financial disclosures to estimate the net profit of constituent stocks[19][21] 2. **Calculate Pre-Tax Dividend Total**: Assume a constant dividend payout ratio (dividend amount/net profit) to calculate the pre-tax dividend total for each stock[21][22] 3. **Impact on Index**: - Calculate the dividend yield: $$\text{Dividend Yield} = \frac{\text{Post-Tax Dividend Total}}{\text{Latest Market Value}}$$ - Calculate the dividend points' impact on the index: $$\text{Dividend Points Impact (\%)} = \text{Stock Weight} \times \text{Dividend Yield}$$ - Adjust stock weights using the formula: $$\mathrm{w_{it}={\frac{w_{i0}\times\mathrm{\(1+R\)}}{\sum_{1}^{n}w_{i0}\times\mathrm{\(1+R\)}}}}$$ where \(w_{i0}\) is the initial weight, and \(R\) is the stock's return over the period[22] 4. **Forecast Dividend Impact on Contracts**: - Estimate ex-dividend dates based on historical patterns or announced schedules - Aggregate dividend impacts before the contract's settlement date to calculate the total dividend points and percentage impact on the futures contract[23][24][26] - **Model Evaluation**: The model provides a systematic approach to quantify dividend impacts, but its accuracy depends on assumptions about dividend payout ratios and ex-dividend dates[18][21][24] 2. Model Name: Futures Pricing Model with Discrete Dividends - **Model Construction Idea**: This model calculates the theoretical price of index futures by incorporating the present value of discrete dividend distributions during the contract period[27] - **Model Construction Process**: 1. Assume the following parameters: - \(F_t\): Futures price at time \(t\) - \(S_t\): Spot price at time \(t\) - \(D\): Present value of dividends during the contract period - \(r\): Risk-free rate over the contract period 2. Calculate the present value of dividends: $$\mathbf{D}=\sum_{\mathrm{i=1}}^{\mathrm{m}}\mathbf{D}_{\mathrm{i}}\,/(1+\phi)$$ where \(\phi\) is the risk-free rate for the interval between dividend payments[27] 3. Derive the futures price using the no-arbitrage pricing formula: $$F_t = (S_t - D)(1 + r)$$[27] - **Model Evaluation**: This model is effective for scenarios with discrete dividend distributions but may require adjustments for continuous dividend flows or irregular dividend schedules[27] 3. Model Name: Futures Pricing Model with Continuous Dividends - **Model Construction Idea**: This model assumes dividends are distributed continuously and uniformly over the contract period, simplifying the pricing process[28] - **Model Construction Process**: 1. Assume the following parameters: - \(F_t\): Futures price at time \(t\) - \(S_t\): Spot price at time \(t\) - \(d\): Annualized dividend yield - \(r\): Annualized risk-free rate - \(T-t\): Time to maturity 2. Derive the theoretical futures price: $$F_t = S_t e^{(r-d)(T-t)}$$[28] - **Model Evaluation**: This model is suitable for markets with frequent and evenly distributed dividends but may oversimplify real-world scenarios with irregular dividend patterns[28] --- Model Backtesting Results 1. Dividend Forecast Model - **Dividend Points Prediction for August Contracts**: - **SSE 50 (IH)**: 3.62 points - **CSI 300 (IF)**: 7.76 points - **CSI 500 (IC)**: 9.18 points - **CSI 1000 (IM)**: 6.25 points[3][8][10] - **Annualized Hedging Costs (Excluding Dividends)**: - **SSE 50 (IH)**: -3.44% - **CSI 300 (IF)**: -1.03% - **CSI 500 (IC)**: 7.79% - **CSI 1000 (IM)**: 11.11%[3][8][10] 2. Futures Pricing Model with Discrete Dividends - **Remaining Dividend Impact on August Contracts**: - **SSE 50 (IH)**: 0.13% - **CSI 300 (IF)**: 0.19% - **CSI 500 (IC)**: 0.15% - **CSI 1000 (IM)**: 0.10%[11][18][24] 3. Futures Pricing Model with Continuous Dividends - **Not explicitly tested in the report** --- Quantitative Factors and Construction Methods 1. Factor Name: Dividend Yield Factor - **Factor Construction Idea**: Measures the dividend yield of index constituent stocks to assess their contribution to the overall index dividend impact[22] - **Factor Construction Process**: 1. Calculate the dividend yield for each stock: $$\text{Dividend Yield} = \frac{\text{Post-Tax Dividend Total}}{\text{Latest Market Value}}$$ 2. Aggregate the weighted dividend yields of all constituent stocks to determine the index-level dividend yield[22] - **Factor Evaluation**: Provides a direct measure of dividend contributions but may be sensitive to changes in stock weights and market values[22] --- Factor Backtesting Results 1. Dividend Yield Factor - **Dividend Yield Impact on August Contracts**: - **SSE 50 (IH)**: 3.62 points - **CSI 300 (IF)**: 7.76 points - **CSI 500 (IC)**: 9.18 points - **CSI 1000 (IM)**: 6.25 points[3][8][10]

Quantitative Models and Construction Methods - **Model Name**: Index Enhancement Strategy (80% Component Constraint) **Model Construction Idea**: The model aims to replicate the performance of typical index enhancement products by adjusting the distribution of components across different market capitalization domains[4][33][34] **Model Construction Process**: 1. The model constrains the component weight to 80% while adjusting the allocation in micro-cap stocks. 2. Specific constraints include: - Industry exposure: 0.1% - Weight cap for CSI 2000 components: 0.2% - Weight cap for micro-cap stocks: 0.1% - Monthly rebalancing frequency 3. Performance metrics such as excess return, tracking error, IR, and maximum drawdown are calculated for different micro-cap allocations (0%, 5%, 10%)[34][35] **Model Evaluation**: The model demonstrates that higher micro-cap allocations can enhance excess returns, albeit with slightly increased tracking error and drawdown[35] - **Model Name**: Index Enhancement Strategy (Relaxed Component Constraint) **Model Construction Idea**: This model explores the impact of relaxing the component weight constraint to 40% while varying micro-cap allocations and market capitalization exposures[33][39] **Model Construction Process**: 1. The component weight constraint is relaxed to 40%, and micro-cap allocations are adjusted (0%, 10%, 20%). 2. Additional constraints include: - Industry exposure: 0.1% - Weight cap for CSI 2000 components: 0.2% - Weight cap for micro-cap stocks: 0.1% - Monthly rebalancing frequency 3. Performance metrics such as excess return, tracking error, IR, and maximum drawdown are calculated for different scenarios[39][40] **Model Evaluation**: Relaxing the component constraint significantly improves excess returns, especially with higher micro-cap allocations, though it introduces higher tracking error and drawdown risks[40] Model Backtesting Results - **Index Enhancement Strategy (80% Component Constraint)**: - CSI 300 (0% micro-cap): Excess Return: 7.97%, Tracking Error: 3.34%, IR: 5.38, Max Drawdown: -1.16% - CSI 300 (5% micro-cap): Excess Return: 8.52%, Tracking Error: 3.45%, IR: 5.58, Max Drawdown: -1.19% - CSI 300 (10% micro-cap): Excess Return: 8.70%, Tracking Error: 3.57%, IR: 5.51, Max Drawdown: -1.33% - CSI 500 (0% micro-cap): Excess Return: 7.55%, Tracking Error: 3.87%, IR: 4.38, Max Drawdown: -1.52% - CSI 500 (5% micro-cap): Excess Return: 8.23%, Tracking Error: 3.88%, IR: 4.78, Max Drawdown: -1.38% - CSI 500 (10% micro-cap): Excess Return: 9.20%, Tracking Error: 3.98%, IR: 5.24, Max Drawdown: -1.39% - CSI 1000 (0% micro-cap): Excess Return: 10.12%, Tracking Error: 4.28%, IR: 5.40, Max Drawdown: -1.50% - CSI 1000 (5% micro-cap): Excess Return: 9.76%, Tracking Error: 4.31%, IR: 5.16, Max Drawdown: -1.69% - CSI 1000 (10% micro-cap): Excess Return: 9.76%, Tracking Error: 4.31%, IR: 5.16, Max Drawdown: -1.69%[35] - **Index Enhancement Strategy (Relaxed Component Constraint)**: - CSI 300 (0% micro-cap): Excess Return: 10.87%, Tracking Error: 4.35%, IR: 5.73, Max Drawdown: -1.29% - CSI 300 (10% micro-cap): Excess Return: 13.96%, Tracking Error: 7.01%, IR: 4.64, Max Drawdown: -3.02% - CSI 500 (0% micro-cap): Excess Return: 10.25%, Tracking Error: 6.65%, IR: 3.52, Max Drawdown: -2.19% - CSI 500 (20% micro-cap): Excess Return: 17.08%, Tracking Error: 7.98%, IR: 5.07, Max Drawdown: -2.43% - CSI 1000 (0% micro-cap): Excess Return: 10.84%, Tracking Error: 6.24%, IR: 3.98, Max Drawdown: -1.54% - CSI 1000 (20% micro-cap): Excess Return: 16.81%, Tracking Error: 7.38%, IR: 5.38, Max Drawdown: -2.04%[40] Quantitative Factors and Construction Methods - **Factor Name**: Market Capitalization (Size) **Factor Construction Idea**: Market capitalization is used as a linear factor to segment stocks into deciles, with smaller-cap stocks expected to deliver higher excess returns[19][22] **Factor Construction Process**: 1. Divide the market into 10 deciles based on market capitalization. 2. Calculate the excess return for each decile. 3. Analyze the trend of excess returns across deciles[22] **Factor Evaluation**: The smallest decile (G01) delivers the highest excess return (22.4%), while returns decrease progressively with increasing market capitalization[22] - **Factor Name**: Mid-Cap (Nonlinear Size) **Factor Construction Idea**: Mid-cap is modeled as a cubic function to capture the performance of stocks outside the large-cap and small-cap domains[2][25] **Factor Construction Process**: 1. Define mid-cap stocks using a cubic function of market capitalization. 2. Analyze the overlap between mid-cap and market capitalization groups. 3. Evaluate the excess return of mid-cap groups[25][26] **Factor Evaluation**: Mid-cap stocks exhibit significant overlap with small-cap stocks, and the smallest mid-cap group (G01) delivers high excess returns (21.6%)[22][25] Factor Backtesting Results - **Market Capitalization (Size)**: - G01: 22.4%, G02: 15.0%, G03: 22.6%, G04: 20.4%, G05: 13.6%, G06: 13.2%, G07: 10.9%, G08: 6.9%, G09: 3.9%, G10: -5.6%[22] - **Mid-Cap (Nonlinear Size)**: - G01: 21.6%, G02: 13.7%, G03: -0.5%, G04: 0.0%, G05: 1.5%, G06: 0.8%, G07: 0.5%, G08: -2.1%, G09: -0.2%, G10: -2.7%[22]

- The report introduces several active quantitative strategies launched by the Changjiang Golden Engineering team since July 2023, including the Dividend Selection Strategy and the Industry High Win Rate Strategy[4][12][13] - The report tracks the performance of two dividend portfolios and two electronic portfolios, highlighting their significant excess returns relative to their respective benchmarks since the beginning of the year[4][13][20] - The Dividend Series includes the "Central State-owned Enterprises High Dividend 30 Portfolio" and the "Balanced Dividend 50 Portfolio," while the Industry Enhancement Series focuses on the electronics sector, including the "Electronic Balanced Configuration Enhancement Portfolio" and the "Electronic Sector Preferred Enhancement Portfolio"[13][14][20] - The "Electronic Sector Preferred Enhancement Portfolio" outperformed the electronics industry index by approximately 0.52% on a weekly basis, and since the beginning of 2025, the "Electronic Balanced Configuration Enhancement Portfolio" and the "Electronic Sector Preferred Enhancement Portfolio" have exceeded the electronics industry index by approximately 3.97% and 5.78%, respectively[5][23][30]

Quantitative Models and Construction Methods Model Name: MFE (Maximized Factor Exposure) Portfolio - **Model Construction Idea**: The MFE portfolio aims to maximize the exposure to a single factor while controlling for various constraints such as industry exposure, style exposure, and stock weight limits[75][76]. - **Model Construction Process**: - The optimization model is formulated as follows: $$ \begin{array}{ll} \text{max} & f^{T}w \\ \text{s.t.} & s_{l} \leq X(w-w_{b}) \leq s_{h} \\ & h_{l} \leq H(w-w_{b}) \leq h_{h} \\ & w_{l} \leq w-w_{b} \leq w_{h} \\ & b_{l} \leq B_{b}w \leq b_{h} \\ & 0 \leq w \leq l \\ & 1^{T}w = 1 \\ & \Sigma|w-w_{0}| \leq to_{h} \end{array} $$ - **Explanation**: - \( f \): Factor values - \( w \): Stock weight vector to be solved - Constraints include style exposure, industry exposure, stock weight deviation, component stock weight limits, and turnover rate[75][76][77]. - The model is solved using linear programming to efficiently determine the optimal weights[76]. - **Model Evaluation**: The MFE portfolio is evaluated based on its historical performance relative to the benchmark index, considering constraints such as industry and style exposures[78][79]. Quantitative Factors and Construction Methods Factor Name: Trend - **Factor Construction Idea**: The Trend factor captures the momentum of stock prices over different time horizons[12][17]. - **Factor Construction Process**: - **Trend_120**: $$ \text{EWMA}(\text{halflife}=20) / \text{EWMA}(\text{halflife}=120) $$ - **Trend_240**: $$ \text{EWMA}(\text{halflife}=20) / \text{EWMA}(\text{halflife}=240) $$ - **Factor Evaluation**: The Trend factor showed a positive return of 2.15% this week, indicating a strong market preference for trend-following strategies[12]. Factor Name: Single Quarter Net Profit YoY Growth - **Factor Construction Idea**: This factor measures the year-over-year growth in net profit for a single quarter[2][8]. - **Factor Construction Process**: - Calculation: $$ \text{Single Quarter Net Profit YoY Growth} = \frac{\text{Current Quarter Net Profit} - \text{Previous Year Same Quarter Net Profit}}{\text{Previous Year Same Quarter Net Profit}} $$ - **Factor Evaluation**: This factor performed the best among the CSI All Share Index components this week[2][8]. Factor Backtesting Results Trend Factor - **Recent Week**: 2.15%[12] - **Recent Month**: 5.62%[14] - **Year-to-Date**: -1.74%[14] - **Last Year**: 26.90%[14] - **Historical Annualized**: 14.22%[14] Single Quarter Net Profit YoY Growth Factor - **Recent Week**: 1.69%[57] - **Recent Month**: 3.19%[57] - **Year-to-Date**: 8.08%[57] - **Last Year**: 3.65%[57] - **Historical Annualized**: 3.20%[57]