Market Overview - The S&P 500 index increased by 0.6% this week, while the Nasdaq rose by 1.5%[3] - Growth stocks outperformed value stocks, with Russell 1000 Growth up by 1.5% and Russell 1000 Value down by 0.2%[3] Sector Performance - The automotive and auto parts sector led gains with an increase of 4.3%, followed by semiconductors at 3.1% and software and services at 2.1%[3] - The energy sector experienced the largest decline, down by 3.8%, followed by healthcare equipment and services at -2.9%[3] Fund Flows - Estimated fund inflows for S&P 500 components were $4.55 billion this week, reversing last week's outflow of $0.57 billion[4] - Semiconductor products and equipment saw the highest inflow at $2.35 billion, while healthcare equipment and services faced an outflow of $1.37 billion[4] Earnings Forecast - The dynamic F12M EPS forecast for S&P 500 components was revised up by 0.6% this week, following a 0.3% increase last week[5] - The banking sector saw the most significant upward revision at +2.7%, while healthcare equipment and services were revised down by -1.0%[5] Risk Factors - Key risks include uncertainties in economic fundamentals, international political situations, U.S. fiscal policy, and Federal Reserve monetary policy[5]

Investment Rating - The investment rating for the company is "Outperform the Market" (maintained) [2][7] Core Views - The establishment of China Yarlung Group Co., Ltd. is expected to promote the development of hydropower resources in the Yarlung Zangbo River, leading to an acceleration in domestic hydropower equipment orders, which will benefit Dongfang Electric as a leading hydropower equipment manufacturer in China [3][6] - The total investment for the Yarlung Zangbo River downstream hydropower project is approximately 1.2 trillion yuan, with the construction of five cascade power stations [3][4] Financial Forecasts - The profit forecasts for Dongfang Electric for 2025-2027 are 4.34 billion yuan, 5.52 billion yuan, and 6.09 billion yuan, representing year-on-year growth of 48.5%, 27.2%, and 10.3% respectively [3][6] - The current price-to-earnings (PE) ratios for the years 2025, 2026, and 2027 are projected to be 13.9, 11.0, and 9.9 times respectively [3][6] Market Data - The company has a market share of 41.6% in pumped storage and 45% in conventional hydropower, indicating a strong competitive position in the energy equipment sector [5] - The company has achieved significant technological advancements, including the development of the world's largest 500 MW hydraulic turbine model and the first domestically produced 300 MW variable-speed pumped storage unit [5]

Quantitative Models and Factor Construction Quantitative Models and Construction Methods - **Model Name**: Maximized Factor Exposure (MFE) Portfolio **Model Construction Idea**: The MFE portfolio is designed to test the effectiveness of individual factors under realistic constraints, such as industry exposure, style exposure, stock weight limits, and turnover constraints. This approach ensures that the factors deemed "effective" can genuinely contribute to return prediction in the final portfolio[41][42]. **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 \( f^{T} w \) represents the weighted exposure of the portfolio to the factor \( f \), and \( w \) is the stock weight vector. - **Constraints**: 1. **Style Exposure**: \( X \) represents the factor exposure matrix for stocks, \( w_b \) is the benchmark weight vector, and \( s_l, s_h \) are the lower and upper bounds for style factor exposure[42]. 2. **Industry Exposure**: \( H \) is the industry exposure matrix, and \( h_l, h_h \) are the lower and upper bounds for industry deviations[42]. 3. **Stock Weight Deviation**: \( w_l, w_h \) are the lower and upper bounds for stock weight deviations relative to the benchmark[42]. 4. **Constituent Weight**: \( B_b \) is a binary vector indicating whether a stock is part of the benchmark, and \( b_l, b_h \) are the lower and upper bounds for constituent weights[42]. 5. **No Short Selling**: Ensures non-negative weights and limits individual stock weights to \( l \)[42]. 6. **Full Investment**: Ensures the portfolio is fully invested with \( \mathbf{1}^{T} w = 1 \)[43]. - **Implementation**: 1. Define constraints for style, industry, and stock weights. For example, for CSI 500 and CSI 300 indices, industry exposure is neutralized, and stock weight deviations are capped at 1%[45]. 2. Construct the MFE portfolio at the end of each month based on the constraints[45]. 3. Backtest the portfolio, accounting for transaction costs (0.3% per side), and calculate performance metrics relative to the benchmark[45]. **Model Evaluation**: The MFE portfolio effectively tests factor performance under realistic constraints, making it a robust tool for evaluating factor predictability in practical scenarios[41][42]. Quantitative Factors and Construction Methods - **Factor Name**: DELTAROA **Factor Construction Idea**: Measures the change in return on assets (ROA) compared to the same quarter in the previous year, capturing improvements in asset utilization efficiency[16]. **Factor Construction Process**: $ DELTAROA = ROA_{current\ quarter} - ROA_{same\ quarter\ last\ year} $ Where \( ROA = \frac{Net\ Income}{Total\ Assets} \)[16]. **Factor Evaluation**: DELTAROA is a growth-oriented factor that has shown strong performance in multiple sample spaces, particularly in the CSI A500 index[19][25]. - **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[16]. **Factor Construction Process**: $ SUE = \frac{Actual\ Earnings - Expected\ Earnings}{Standard\ Deviation\ of\ Expected\ Earnings} $[16]. **Factor Evaluation**: SUE is a profitability factor that performs well in growth-oriented indices like CSI 1000 and CSI A500[19][23][25]. - **Factor Name**: One-Year Momentum **Factor Construction Idea**: Captures the trend-following behavior of stocks by measuring price momentum over the past year, excluding the most recent month[16]. **Factor Construction Process**: $ Momentum = \frac{Price_{t-12} - Price_{t-1}}{Price_{t-1}} $ Where \( t-12 \) and \( t-1 \) represent the stock price 12 months and 1 month ago, respectively[16]. **Factor Evaluation**: Momentum is a widely used factor that has shown consistent performance in large-cap indices like CSI 300 and CSI 500[19][21]. Factor Backtesting Results - **CSI 300 Sample Space**: - **Best-Performing Factors (1 Week)**: Single-quarter revenue growth, DELTAROA, single-quarter ROE[19]. - **Worst-Performing Factors (1 Week)**: Three-month volatility, one-month volatility, three-month turnover[19]. - **CSI 500 Sample Space**: - **Best-Performing Factors (1 Week)**: One-year momentum, standardized unexpected revenue, standardized unexpected earnings[21]. - **Worst-Performing Factors (1 Week)**: SPTTM, single-quarter SP, dividend yield[21]. - **CSI 1000 Sample Space**: - **Best-Performing Factors (1 Week)**: Three-month reversal, standardized unexpected revenue, single-quarter surprise magnitude[23]. - **Worst-Performing Factors (1 Week)**: Dividend yield, one-month volatility, BP[23]. - **CSI A500 Sample Space**: - **Best-Performing Factors (1 Week)**: DELTAROA, standardized unexpected earnings, single-quarter ROA[25]. - **Worst-Performing Factors (1 Week)**: Three-month volatility, one-month turnover, one-month volatility[25]. - **Public Fund Heavyweight Index Sample Space**: - **Best-Performing Factors (1 Week)**: One-year momentum, standardized unexpected revenue, expected net profit QoQ[27]. - **Worst-Performing Factors (1 Week)**: Dividend yield, one-month volatility, three-month volatility[27].

Quantitative Models and Construction Methods Model Name: Guosen Golden Engineering Hong Kong Stock Selection Portfolio - **Model Construction Idea**: The model aims to perform dual-layer selection based on fundamental and technical aspects from the analyst-recommended stock pool[15][17]. - **Model Construction Process**: - **Step 1**: Construct the analyst-recommended stock pool based on events such as analyst upward earnings forecast revisions, first-time analyst attention, and analyst report titles exceeding expectations[17]. - **Step 2**: Perform fundamental and technical selection from the analyst-recommended stock pool to pick stocks with both fundamental support and technical resonance[17]. - **Backtesting Period**: 2010-01-01 to 2025-06-30, considering transaction costs, the portfolio's annualized return is 19.11%, with an excess return of 18.48% relative to the Hang Seng Index[17]. - **Model Evaluation**: The model effectively combines fundamental and technical analysis to achieve significant excess returns over the Hang Seng Index[17]. Model Backtesting Results - **Guosen Golden Engineering Hong Kong Stock Selection Portfolio**: - **Absolute Return**: 49.59% (2025)[2][18] - **Excess Return Relative to Hang Seng Index**: 25.83% (2025)[2][18] - **Annualized Return**: 19.11%[17] - **Excess Return Relative to Hang Seng Index**: 18.48%[17] - **Information Ratio (IR)**: 1.22[20] - **Tracking Error**: 14.55%[20] - **Maximum Drawdown**: 23.73%[20] - **Return-to-Drawdown Ratio**: 0.78[20] Quantitative Factors and Construction Methods Factor Name: Stable New High Stock Selection - **Factor Construction Idea**: The factor aims to identify stocks that have recently reached new highs and exhibit stable price paths, leveraging momentum and trend-following strategies[21][23]. - **Factor Construction Process**: - **Formula**: $ 250 \text{ Day New High Distance} = 1 - \frac{\text{Close}_{t}}{\text{ts\_max(Close, 250)}} $ - **Explanation**: $\text{Close}_{t}$ represents the latest closing price, and $\text{ts\_max(Close, 250)}$ represents the maximum closing price over the past 250 trading days[23]. - **Selection Criteria**: - **Analyst Attention**: At least 5 buy or hold ratings in the past 6 months[24]. - **Relative Strength**: Top 20% in terms of price change over the past 250 days[24]. - **Price Stability**: Comprehensive scoring based on price path smoothness and new high persistence over the past 120 days[24]. - **Trend Continuation**: Average 250-day new high distance over the past 5 days, selecting the top 50 stocks[24]. - **Factor Evaluation**: The factor effectively captures stocks with strong momentum and stable price paths, which are likely to continue their upward trends[21][23][24]. Factor Backtesting Results - **Stable New High Stock Selection**: - **Absolute Return**: 267.4% (康方生物)[23][29] - **250-Day New High Distance**: 0.0% (康方生物)[23][29] - **Past 250-Day Price Change**: 52.9% (康方生物)[23][29] - **Past 20-Day Price Change**: 52.9% (康方生物)[23][29] - **Analyst Report Count**: 46 (康方生物)[23][29]

Core Insights - The report highlights the performance of various active quantitative strategies, indicating that the "Growth and Stability Portfolio" has achieved a year-to-date return of 33.13% [1][2] - The "Excellent Fund Performance Enhancement Portfolio" has shown an absolute return of 10.32% this year, ranking in the 45.63 percentile among active equity funds [1][2] - The "Expected Selection Portfolio" has outperformed the benchmark with a year-to-date return of 24.40%, ranking in the 11.53 percentile among active equity funds [1][2] - The "Brokerage Golden Stock Performance Enhancement Portfolio" has achieved a year-to-date return of 14.13%, ranking in the 31.39 percentile among active equity funds [1][2] Summary by Sections Excellent Fund Performance Enhancement Portfolio - The portfolio aims to outperform the median return of active equity funds by utilizing a quantitative approach based on the holdings of top-performing funds [3][19] - This week, the portfolio achieved an absolute return of 2.75%, with a year-to-date return of 10.32%, underperforming the benchmark by 0.31% [1][23] - The portfolio's ranking among active equity funds is 1583 out of 3469, placing it in the 45.63 percentile [1][23] Expected Selection Portfolio - This portfolio selects stocks based on expected performance and analyst profit revisions, focusing on both fundamental and technical criteria [4][24] - The portfolio achieved an absolute return of 3.68% this week and 24.40% year-to-date, outperforming the benchmark by 0.62% [1][33] - It ranks 400 out of 3469 in the active equity fund category, placing it in the 11.53 percentile [1][33] Brokerage Golden Stock Performance Enhancement Portfolio - The strategy utilizes a stock pool from brokerage recommendations, aiming to optimize the portfolio while minimizing deviations from the benchmark [5][34] - This week, the portfolio achieved an absolute return of 1.91%, with a year-to-date return of 14.13%, outperforming the benchmark by 1.67% [1][41] - The portfolio ranks 1089 out of 3469 in the active equity fund category, placing it in the 31.39 percentile [1][41] Growth and Stability Portfolio - The portfolio employs a two-dimensional evaluation system for growth stocks, prioritizing those with upcoming earnings announcements [6][42] - This week, the portfolio achieved an absolute return of 2.15%, with a year-to-date return of 29.61%, outperforming the benchmark by 17.15% [1][50] - It ranks 252 out of 3469 in the active equity fund category, placing it in the 7.26 percentile [1][50]

Quantitative Models and Construction Methods 1. Model Name: 250-Day New High Distance - **Model Construction Idea**: This model tracks the distance of a stock or index from its 250-day high to identify market trends and hotspots. It is based on the premise that stocks nearing their 52-week high tend to outperform, as supported by prior research (e.g., George@2004, William O'Neil's CANSLIM system, and Mark Minervini's "Stock Market Wizard").[11][18][21] - **Model Construction Process**: The 250-day new high distance is calculated as follows: $ 250 \text{ Day New High Distance} = 1 - \frac{\text{Close}_{t}}{\text{ts\_max(Close, 250)}} $ - $\text{Close}_{t}$: Latest closing price - $\text{ts\_max(Close, 250)}$: Maximum closing price over the past 250 trading days If the latest closing price reaches a new high, the distance is 0. If the price has fallen from the high, the distance is positive, indicating the percentage drop.[11] - **Model Evaluation**: This model effectively identifies market leaders and trends, aligning with momentum and trend-following strategies.[11][18] 2. Model Name: Stable New High Stock Screening - **Model Construction Idea**: This model focuses on identifying stocks with stable momentum characteristics, emphasizing smooth price paths and consistent new highs. Research suggests that smoother momentum stocks outperform those with jumpy price paths (e.g., Bali et al., 2011; Da et al., 2012).[25][27] - **Model Construction Process**: Stocks are screened from the pool of those hitting 250-day highs in the past 20 trading days based on the following criteria: - **Analyst Attention**: At least 5 "Buy" or "Overweight" ratings in the past 3 months - **Relative Strength**: Top 20% in 250-day price performance - **Price Stability**: Evaluated using two metrics: - **Price Path Smoothness**: Ratio of price displacement to price path length - **New High Consistency**: Average 250-day new high distance over the past 120 days - **Trend Continuation**: Average 250-day new high distance over the past 5 days Stocks meeting these criteria are ranked, and the top 50% are selected.[25][27] - **Model Evaluation**: This model captures stocks with strong and stable momentum, leveraging underreaction to smooth price paths for enhanced returns.[25][27] --- Model Backtesting Results 1. 250-Day New High Distance - **Indices**: - Shanghai Composite: 0.00% - Shenzhen Component: 5.06% - CSI 300: 4.64% - CSI 500: 3.65% - CSI 1000: 0.91% - CSI 2000: 0.00% - ChiNext Index: 10.71% - STAR 50 Index: 10.59%[12][13][15] 2. Stable New High Stock Screening - **Selected Stocks**: 46 stocks, including Shenghong Technology, Borui Pharmaceutical, and Shijia Photon[28][33] - **Sector Distribution**: - Manufacturing: 14 stocks (e.g., machinery) - Technology: 13 stocks (e.g., computers)[28][33] --- Quantitative Factors and Construction Methods 1. Factor Name: 250-Day New High Distance - **Factor Construction Idea**: Measures the relative position of a stock's price to its 250-day high, serving as a momentum indicator.[11] - **Factor Construction Process**: $ 250 \text{ Day New High Distance} = 1 - \frac{\text{Close}_{t}}{\text{ts\_max(Close, 250)}} $ - $\text{Close}_{t}$: Latest closing price - $\text{ts\_max(Close, 250)}$: Maximum closing price over the past 250 trading days[11] - **Factor Evaluation**: Effectively identifies momentum leaders and market trends.[11][18] 2. Factor Name: Price Path Smoothness - **Factor Construction Idea**: Quantifies the smoothness of a stock's price trajectory, with smoother paths indicating stronger momentum.[25][27] - **Factor Construction Process**: $ \text{Price Path Smoothness} = \frac{\text{Price Displacement}}{\text{Price Path Length}} $ - **Price Displacement**: Net change in price over a period - **Price Path Length**: Sum of absolute daily price changes over the same period[27] - **Factor Evaluation**: Highlights stocks with stable momentum, leveraging underreaction to smooth price paths.[25][27] 3. Factor Name: New High Consistency - **Factor Construction Idea**: Measures the average proximity to 250-day highs over a specified period, indicating sustained momentum.[27] - **Factor Construction Process**: $ \text{New High Consistency} = \text{Mean(250 Day New High Distance over 120 Days)} $[27] - **Factor Evaluation**: Captures stocks with consistent momentum, emphasizing sustainability.[27] 4. Factor Name: Trend Continuation - **Factor Construction Idea**: Measures short-term proximity to 250-day highs, indicating recent momentum strength.[27] - **Factor Construction Process**: $ \text{Trend Continuation} = \text{Mean(250 Day New High Distance over 5 Days)} $[27] - **Factor Evaluation**: Identifies stocks with strong short-term momentum, complementing longer-term factors.[27] --- Factor Backtesting Results 1. 250-Day New High Distance - **Indices**: - Shanghai Composite: 0.00% - Shenzhen Component: 5.06% - CSI 300: 4.64% - CSI 500: 3.65% - CSI 1000: 0.91% - CSI 2000: 0.00% - ChiNext Index: 10.71% - STAR 50 Index: 10.59%[12][13][15] 2. Price Path Smoothness - **Selected Stocks**: 46 stocks, including Shenghong Technology, Borui Pharmaceutical, and Shijia Photon[28][33] 3. New High Consistency - **Selected Stocks**: 46 stocks, with top sectors being manufacturing (14 stocks) and technology (13 stocks)[28][33] 4. Trend Continuation - **Selected Stocks**: 46 stocks, with top sectors being manufacturing (14 stocks) and technology (13 stocks)[28][33]

1. Report Industry Investment Rating No relevant content provided. 2. Core Viewpoints - Measuring the duration of bond funds helps investors assess risks and estimate risk - return characteristics, including annualized returns and net - value changes during interest - rate fluctuations. In a volatile interest - rate market, long - duration bond funds face greater price - fluctuation risks, while short - duration ones are more stable [2][16]. - Over the past three years, the overall duration of funds has been on an upward trend. In a low - interest - rate environment, fund managers have increased allocations to long - duration interest - rate bonds to lock in higher returns. There is a significant negative correlation between duration changes and the 10 - year Treasury yield [4]. 3. Summary by Relevant Catalogs 3.1. Duration Essence Re - cognition - Duration is an approximate indicator of a bond's sensitivity to interest - rate changes, representing the approximate percentage change in value for a 100 - basis - point interest - rate change, which determines interest - rate risk exposure. It can also be seen as a measure of time, quantifying the sensitivity of bond prices to interest - rate changes by calculating the weighted cash - flow recovery time [1][13]. - Modified duration is the approximate percentage change in bond price for a 100 - basis - point yield change under the assumption of constant expected cash flows [1][13]. 3.2. Bond Fund Duration Calculation Methods - The three mainstream methods for calculating bond fund duration are the top - holding weighted method, the interest - rate sensitivity method, and the asset - portfolio method. In high - frequency calculation scenarios, the asset - portfolio method is the most mainstream and has the best comprehensive performance [16][17]. 3.3. Bond Fund Duration Estimation Steps 3.3.1. Sample Selection - Select mid - long - term and short - term pure - bond funds according to the Wind fund classification standard, excluding funds with less than 1 - year data, amortized - cost - valuation funds, and non - original funds. The observation period is from early 2017 to the present [18][19][20]. 3.3.2. Duration Calculation - Based on top - holdings: Calculate the weighted - average duration using the top five heavy - holding bonds in the fund report, but this method may have large deviations [22][23]. - Based on interest - rate sensitivity: Derive the fund's duration from the interest - rate risk analysis in the fund report, but the data frequency is low [24]. - Asset - portfolio method: Use factor - return regression to estimate the duration of pure - bond funds. Set non - negative constraints and a coefficient - sum constraint to ensure the rationality of the results [28][29][30]. 3.3.3. Model Selection - The least absolute deviation regression (LAD) is chosen as the core regression method for daily duration estimation of bond funds to ensure the robustness and interpretability of the model output [39]. 3.3.4. Factor Processing - Select a factor system covering interest - rate and credit - bond index factors. Use the variance inflation factor (VIF) analysis to screen factors and remove variables with significant multicollinearity [40][42]. 3.4. Bond Fund Duration Calculation and Tracking 3.4.1. Pure - Bond Fund Duration Calculation - The calculated duration is generally consistent with the reported duration. From 2017 to 2025, the average and median durations of pure - bond funds were relatively stable, while the maximum duration increased significantly. The median duration of pure - bond funds and the 10 - year Treasury yield were mostly negatively correlated [45][48][50]. 3.4.2. Mid - Long - Term vs. Short - Term Bond Fund Duration Calculation - The durations of mid - long - term and short - term pure - bond funds generally moved in the same direction, with the former having a larger fluctuation range. The duration of mid - long - term pure - bond funds showed obvious stratification, while that of short - term ones was more concentrated [57][59]. 3.4.3. Interest - Rate Bond vs. Credit - Bond Fund Duration Calculation - The durations of both credit - bond and interest - rate bond funds showed an upward trend, with the latter having a larger increase. Both types of funds' durations were negatively correlated with the 10 - year Treasury yield, and the interest - rate bond funds' duration adjustment was more sensitive [79][81][93]. 3.4.4. Single - Fund Duration Calculation - Taking "Huatai Baoxing Anyue A" as an example, the regression result was basically consistent with the reported data. The fund's duration increased significantly after June 2023 [100].

Core Insights - The report highlights the significant growth potential in the measurement and calibration industry, driven by new policies aimed at enhancing manufacturing capabilities in China [6][7][8] - The renewable energy sector, particularly in electric power equipment, is poised for growth due to supportive policies in the UK and increasing demand for energy storage solutions [13] - The pharmaceutical industry is recommended for investment, focusing on innovative drugs and their supply chains, with strong support from health insurance reforms [14] Industry and Company Analysis - **Measurement and Calibration Industry**: The first policy document from the Ministry of Industry and Information Technology emphasizes the need for precise measurement to drive innovation in manufacturing. This includes establishing a service network and digital transformation paths, with a focus on high-level calibration institutions and digital measurement software [6][7] - **Renewable Energy Sector**: The UK government has restarted subsidies for electric vehicles and charging infrastructure, indicating a robust market for energy storage systems. The report suggests focusing on companies involved in battery production and charging infrastructure, such as Ningde Times and Keda Technology [13] - **Pharmaceutical Industry**: The report continues to recommend innovative drug sectors, highlighting the recent adjustments in health insurance and commercial insurance that favor high-value innovative drugs. Companies like Kelun-Biotech and Innovent Biologics are noted for their strong potential in both domestic and international markets [14] - **Oil and Gas Sector**: China National Offshore Oil Corporation (CNOOC) has made significant advancements in oil and gas exploration, achieving record production levels in both domestic and international operations. The report anticipates continued growth in production capacity, particularly in the Stabroek block in Guyana [20][21][24]

- The report discusses the market performance of various indices, highlighting that the CSI 1000 index performed well with a 1.14% increase, while the SSE 50 index rose by 0.12%[6] - The report notes that the defense, communications, electronics, pharmaceuticals, and steel industries performed well, with returns of 3.06%, 2.45%, 2.14%, 1.75%, and 1.55%, respectively[8] - The report mentions that the ETF with the highest premium on July 16, 2025, was the Online Consumption ETF with a premium of 0.95%, while the ETF with the highest discount was the Penghua GEM New Energy ETF with a discount of 0.97%[23]

1. Report Industry Investment Rating No relevant content provided. 2. Core Viewpoints of the Report - Observing the duration of bond funds helps investors assess their risks. By calculating the duration, one can estimate the risk - return characteristics of the fund, including the annualized return and net - value changes during interest - rate fluctuations. In a volatile interest - rate market, bond funds with longer durations face greater price - fluctuation risks, while those with shorter durations are relatively more stable [2]. - Over the past three years, the overall duration of funds has been on an upward trend, which is an active choice of institutions to optimize risk - return in a low - interest - rate environment. There is an obvious negative correlation between duration changes and the yield of the ten - year Treasury bond. In an interest - rate downward cycle, extending the duration becomes the core strategy to increase returns [4]. 3. Summaries According to Relevant Catalogs 3.1 Duration Essence Re - understanding - Duration is an approximate indicator to measure the sensitivity of bond value to interest - rate changes. It is the approximate percentage of value change when the interest rate changes by 100 basis points, determining the interest - rate risk exposure. It can also be regarded as a time measure, reflecting the elasticity of bond prices relative to interest - rate changes. Longer duration means higher sensitivity to interest - rate changes [1][13]. - Modified duration is a form of duration used by industry insiders. It is the approximate percentage of bond - price change when the yield changes by 100 basis points under the assumption that the expected cash flow of the bond remains unchanged [1][13]. 3.2 Bond Fund Duration Calculation Methods - The mainstream bond - fund duration calculation methods in the market include the top - holding weighted method, the interest - rate sensitivity method, and the asset - portfolio method. In high - frequency calculation scenarios, the asset - portfolio method is the most mainstream and has the best comprehensive performance, combining data frequency (daily/weekly) and dynamic adjustment ability [2][16][17]. 3.3 Bond Fund Duration Estimation Steps 3.3.1 Sample Selection - Focus on pure - bond funds. Select medium - long - term and short - term pure - bond funds respectively, excluding funds with less than one - year data, funds using the amortized - cost valuation method, and non - original funds. The observation time is set from the beginning of 2017 to the present [18][19][20]. 3.3.2 Duration Calculation - Based on top - holding bonds: Calculate the weighted - average portfolio duration using the top five heavy - holding bonds disclosed in the fund's regular reports. However, this method may lead to large deviations [22][23]. - Based on interest - rate sensitivity: Derive the fund's duration from the interest - rate risk sensitivity analysis in the fund's regular reports. But the data update frequency is low [24]. - Using the asset - portfolio method: Introduce a method based on factor - return regression estimation. Take the average daily return of pure - bond public funds as the explained variable and the daily returns of different bond indices as explanatory variables. After regression analysis, the fund's portfolio duration can be reasonably estimated [28]. 3.4 Model Selection - After comparing the advantages and disadvantages of OLS, Lasso regression, Ridge regression, Stepwise Regression, and LAD, LAD is selected as the core regression method for estimating the daily duration of bond funds to ensure the robustness and interpretability of the model output in complex market environments [33][39]. 3.5 Factor Processing - Select a factor system covering interest - rate bond index factors and credit - bond index factors. Use the Variance Inflation Factor (VIF) analysis method to screen variables, removing variables with high VIF values to ensure the stability of model parameters [40][42]. 3.6 Bond Fund Duration Calculation and Tracking 3.6.1 Pure - Bond Fund Duration Calculation - The calculated duration and the reported duration of pure - bond funds generally have the same trend and small differences. From 2017 to 2025, the average and median durations of pure - bond funds were relatively stable, while the maximum duration continued to rise, and the minimum duration was basically below 1 year. There is a negative correlation between the median duration of pure - bond funds and the yield of the ten - year Treasury bond [45][48][50]. 3.6.2 Medium - Long - Term Bond Fund vs. Short - Term Bond Fund Duration Calculation - The duration of medium - long - term and short - term pure - bond funds is on an upward trend. The duration of medium - long - term pure - bond funds fluctuates more widely, and the distribution is more stratified. The duration of short - term pure - bond funds is more concentrated, and the risk - preference differentiation is limited. There is a negative correlation between the duration of both types of funds and the yield of the ten - year Treasury bond [57][59][66]. 3.6.3 Interest - Rate Bond Fund vs. Credit - Bond Fund Duration Calculation - The durations of both credit - bond funds and interest - rate bond funds show an upward trend, with the latter having a larger increase. The duration adjustment of credit - bond funds is relatively stable, while that of interest - rate bond funds is more volatile. There is a negative correlation between the duration of both types of funds and the yield of the ten - year Treasury bond, and the interest - rate bond funds respond more directly to interest - rate changes [79][81][93]. 3.6.4 Single - Fund Duration Calculation - Taking "Huatai Baoxing Anyue A" as an example, the regression result is basically consistent with the reported data. The fund's duration remained stable below 2.5 years from 2020 to June 2023 and then increased sharply [100].