Core Viewpoint - The article presents a comprehensive analysis of major asset classes based on macroeconomic and technical perspectives, indicating a mixed outlook for equities, bonds, industrial products, and gold [1][3][19]. Macroeconomic Perspective - The macroeconomic view suggests a neutral stance on equity assets, a favorable outlook for bond assets, a negative outlook for industrial products, and a positive outlook for gold assets [5][19]. - Specific macro indicators such as PMI, CPI, and social financing are analyzed to assess their impact on asset performance [6][19]. Technical Perspective - The technical analysis indicates a downward trend for equity and bond assets, while industrial products and gold show an upward trend [9][10][19]. - The article employs various trend indicators to measure the performance of different asset classes, with historical data supporting the current trends [7][10]. Asset Valuation - The equity risk premium (ERP) for the CSI 800 index is reported at 73.74%, indicating that equity valuations are relatively low [12][13]. - The analysis of fund flow shows a net inflow of 915 million yuan into equity assets, suggesting a positive sentiment among investors [15][16]. Asset Allocation Performance Tracking - Historical performance data indicates that a fixed ratio combined with macro and technical indicators yielded a return of 1.06% as of June 2025, with an annualized return of 11.86% since April 2006 [2][20][24]. - Different asset allocation strategies, including risk parity and volatility control, have been evaluated, showing varying returns and risk profiles [25][24]. Summary of Asset Class Scores - The combined scores from macro and technical indicators show equities at 0, bonds at 2, industrial products at 0, and gold at 4, reflecting the overall market sentiment and expected performance [18][19].

Market Performance - The recent five trading days saw the Sci-Tech 50 Index increase by 3.17%, the ChiNext Index by 5.69%, the large-cap value by 1.52%, the large-cap growth by 2.61%, the SSE 50 by 1.27%, and the small-cap represented by the CSI 2000 by 4.94% [1] - The sectors showing strong performance include computers and national defense, while oil, petrochemicals, and food and beverages lagged behind [1] Risk Premium Analysis - The risk premium, calculated as the inverse of the static PE of the CSI All Index minus the yield of ten-year government bonds, indicates that the implied returns of equity and bond assets are at historically high levels, reaching 4.17% on April 26, 2022, and 4.08% on October 28, 2022 [1] - As of January 19, 2024, the indicator stood at 4.11%, marking the fifth occurrence since 2016 to exceed 4% [1] Valuation Levels - As of June 27, 2025, the CSI All Index's PETTM percentile is at 59%, with the SSE 50 and CSI 300 at 66% and 57% respectively, while the ChiNext Index is close to 19% [2] - The ChiNext Index's valuation is relatively low compared to historical averages [2] Long-term Market Trends - The technical analysis of the Deep 100 Index suggests a cyclical pattern of bear markets every three years, followed by bull markets, with significant declines observed in previous cycles [2] - The current adjustment phase, which began in Q1 2021, appears to have sufficient time and space for a potential upward cycle [2] Fund Flow and Trading Activity - In the last five trading days, ETF funds experienced an outflow of 1.3 billion yuan, while margin trading increased by approximately 17 billion yuan, with an average daily trading volume of 1.4528 trillion yuan across the two markets [4] AI and Machine Learning Applications - The use of convolutional neural networks (CNN) for modeling price and volume data has been explored, with the latest focus on sectors such as banking and artificial intelligence [3][11]

Market Performance - The recent five trading days saw the Sci-Tech 50 Index decline by 1.55%, the ChiNext Index by 1.66%, while the large-cap value stocks rose by 1.07% and large-cap growth stocks fell by 0.54% [1] - The banking and telecommunications sectors performed well, whereas the beauty and textile sectors lagged behind [1] Risk Premium Analysis - The static PE of the CSI All Share Index indicates a risk premium, with the EP minus the ten-year government bond yield showing historical extremes at two standard deviations above the mean during previous market bottoms [1] - As of January 19, 2024, the risk premium indicator reached 4.11%, marking the fifth occurrence since 2016 to exceed 4% [1] Valuation Levels - As of June 20, 2025, the CSI All Share Index's TTM PE is at the 53rd percentile, with the SSE 50 and CSI 300 at 65% and 54% respectively, indicating that the ChiNext Index is relatively undervalued historically [2] Long-term Market Trends - The Deep 100 Index has historically experienced bear markets every three years, followed by bull markets, with the current adjustment since Q1 2021 showing sufficient time and space for a potential upward cycle [2] Fund Flow and Trading Activity - In the last five trading days, ETF inflows totaled 17 billion yuan, with margin trading increasing by approximately 310 million yuan, and the average daily trading volume across both markets was 1.1863 trillion yuan [4] AI and Machine Learning Applications - A convolutional neural network is utilized to model price and volume data, mapping learned features to industry themes, with a focus on sectors such as non-ferrous metals and banking [3][10]

Core Viewpoint - The article discusses the increasing application of machine learning in quantitative stock selection, particularly focusing on GBDT and neural network models, as traditional factors have become less effective [1][4]. Group 1: Model Selection - Machine learning has been widely adopted in quantitative stock selection, with GBDT models (including LGBM, XGBoost, and CatBoost) and neural networks (including GRU, TCN, and Transformer) being the primary focus [1]. - GBDT models are effective for handling manually constructed features, while neural networks excel in capturing temporal changes in features [2]. Group 2: Feature Data Preparation - Different model types require different feature types; tree models handle price and fundamental features well, while neural networks perform better with high-frequency data [22][27]. - Feature selection methods, particularly SHAP, can effectively reduce the number of features while maintaining model performance [2][31]. - Standardization of features before feeding them into models is crucial for improving model performance [2][35]. Group 3: Loss Function Adjustment and Prediction Target Processing - Besides the common MSE loss function, investors often use IC as a loss function, with various ranking loss functions showing improved performance [2][37]. - Using cross-sectional normalization helps the model focus on differences in cross-sectional returns, enhancing factor performance [3][50]. Group 4: Machine Learning Models - GBDT is highlighted as a superior algorithm due to its iterative approach of updating target values based on residuals from previous trees [10][11]. - Neural networks, including RNN, LSTM, GRU, CNN, TCN, and Transformer, are discussed for their effectiveness in various domains, particularly in time series prediction [12][19]. Group 5: Index Enhancement Strategies - The article presents the performance of various index enhancement strategies, with the CSI 300 index showing an annualized excess return of 10.03% and a maximum drawdown of -5.42% [3]. - The CSI 500 index strategy has a slightly lower annualized excess return of 8.41% with a maximum drawdown of -10.78%, while the CSI 1000 index strategy shows a more stable performance with an annualized excess return of 11.44% and a maximum drawdown of -7.95% [3].

Core Viewpoint - The rapid development of ETFs in the A-share market has led to a significant increase in their scale and number, surpassing actively managed funds, indicating a growing preference for passive investment strategies among investors [4][5]. Group 1: ETF Growth and Market Dynamics - As of June 15, 2025, the total scale of stock ETFs (including off-market linked funds) reached 3.81 trillion yuan, with the number of ETFs totaling 2,031, exceeding the scale of actively managed funds at 2.84 trillion yuan [4][5]. - The A-share market exhibits significant industry and style differentiation, suggesting that merely holding a single ETF for the long term may not yield optimal investment experiences [4][6]. - The investment objective of ETFs is to closely track the net value performance of specific indices, making the choice of index crucial for investors seeking substantial returns [6][10]. Group 2: ETF Rotation Strategy Development - A common method for constructing ETF rotation strategies involves aggregating effective stock factors at the index level, allowing for index rotation effects [2][11]. - The use of the AGRU model based on daily K-line volume and price data has resulted in the identification of high-performing stock selection factors in the A-share market [12][16]. - Monthly rebalancing of the strategy yielded an average IC of 7.80%, with an annualized excess return of 4.92% and a maximum drawdown of -14.02% [31][39]. Group 3: Performance of Fixed Number ETF Rotation Strategies - Limiting the number of held ETFs to 5, 10, or 15 resulted in varying annualized excess returns: 12.34% for 5 ETFs, 8.75% for 10 ETFs, and 8.13% for 15 ETFs, with corresponding maximum drawdowns of -12.17%, -8.83%, and -8.66% respectively [59][65]. - The strategy consistently achieved positive excess returns annually, with a notable 8.74% excess return year-to-date [63][65]. Group 4: Factor Testing and Adjustments - The factor's performance was enhanced through the adjustment of the loss function, leading to improved multi-directional return performance [17][19]. - The AGRU factor demonstrated strong stock selection effects across various stock pools, with annualized excess returns of 21.97% for the CSI 300 pool and 11.46% for the CSI 500 pool [64][65]. Group 5: MMR Algorithm and Risk Diversification - The MMR (Maximum Marginal Relevance) algorithm was employed to reduce the correlation among selected investment targets, enhancing the stability of the strategy's performance [45][50]. - The strategy's annualized excess return improved from 7.94% to 8.43% after implementing the MMR adjustments, with a corresponding increase in the information ratio [50][52].

Core Viewpoint - The article discusses the potential of Reinforcement Learning (RL) in quantitative investment, particularly in developing timing strategies that can maximize cumulative returns through trial and error learning mechanisms [1][2]. Summary by Sections 1. Introduction to Reinforcement Learning - Reinforcement Learning (RL) is a machine learning method that enables decision-making systems to learn optimal actions in specific situations to maximize cumulative rewards. This method is particularly suitable for environments with clear goals but no direct guidance on achieving them [6][12]. 2. Timing Strategy - The article focuses on the Double Deep Q-Network (DDQN) model, which uses 10-minute frequency price and volume data as input. The goal is for the model to learn to provide buy/sell/hold signals at various time points to maximize end-period returns. The backtesting phase outputs timing signals every 10 minutes, adhering to a t+1 trading rule [2][3]. 3. Empirical Analysis - The strategy was tested on various liquid ETFs and stocks from January 1, 2023, to May 31, 2025. The results showed that the strategy generated 72, 30, 73, and 188 timing signals for different assets, with average win rates of 52.8%, 53.3%, 54.8%, and 51.6%, respectively. Cumulative returns outperformed benchmark assets by 10.9%, 35.5%, 64.9%, and 37.8% [3][74][80]. 4. Summary and Outlook - Despite the impressive performance of RL in various fields, challenges such as stability issues remain in the quantitative investment domain. Future reports will explore more RL algorithms to develop superior strategies [5]. 5. Data Description - The timing strategy was applied to the CSI 300 Index, CSI 500 Index, CSI 1000 Index, and a specific stock, utilizing liquid ETFs corresponding to these indices. The training data spanned from January 1, 2014, to December 31, 2019, with validation and testing periods defined [74][75]. 6. Performance Metrics - The performance metrics for the RL timing strategy included total returns, annualized returns, maximum drawdown, annualized volatility, Sharpe ratio, information ratio, and return-to-drawdown ratio, demonstrating the strategy's effectiveness compared to benchmark assets [77][80].

Core Viewpoint - The report discusses the "Leading Stock Diffusion Effect" as a mechanism driving sector trends in the A-share market, emphasizing the importance of constructing optimal investment portfolios based on improved factors like economic conditions and capital flows [1][2][3]. Research Background - The demand for industry-level beta timing has increased due to the development of flexible allocation funds and the growing industry ETF system, making sector rotation a core asset allocation need [6]. - The A-share market has seen accelerated sector rotation, which poses challenges to traditional rotation models, necessitating a reevaluation and improvement of these models [7]. Mechanism of Diffusion Effect - The diffusion effect in the A-share market typically involves capital migrating from core leading stocks to related targets, driven by policy triggers, active capital inflows, cognitive dissemination, and expectation overshoot leading to differentiation [2][16]. - The process includes vertical and horizontal expansions within the industry, market capitalization descent, and valuation arbitrage, ultimately leading to a broader sector rally [17]. Performance of Improved Factors - The report presents improved factors based on the previous discussion, showing significant performance enhancements in the revised SUE and active large order factors, with annualized excess returns of 7.9% and 10.3% respectively [21][22]. - The improved factors demonstrate better stability and lower volatility compared to traditional models, particularly in recent years [64]. Optimal Industry Portfolio - The optimal industry portfolio, constructed using a common condition screening method based on component factors, has shown superior historical performance with an annualized return of 26.0% and an annualized excess return of 19.1% since 2013 [3][64]. - The portfolio has maintained stable excess growth since 2022, with an annualized excess return of 11.7% and a maximum drawdown of 9.2% [74]. Comparison of Multi-Headed Construction Methods - The report compares two multi-headed construction methods: composite factor multi-headed and component factor common condition screening, concluding that the latter offers lower volatility and more stable excess returns [42][64]. - The composite factor multi-headed approach has shown stagnation in excess returns in recent years, while the optimal industry portfolio continues to outperform [53][64].

Core Viewpoint - The article emphasizes the significance of the China Securities Hong Kong Stock Connect Technology Index, which focuses on large-cap technology companies with high R&D investment and revenue growth within the Hong Kong Stock Connect framework [1][3]. Group 1: Index Characteristics - The index selects 50 large-cap technology companies to reflect the overall performance of technology leaders in the Hong Kong Stock Connect [3]. - The index has a balanced industry distribution, with major allocations in internet, automotive, and innovative pharmaceuticals, avoiding overcrowded sectors like electronics and media [11]. - The index includes industry leaders such as BYD in automotive and BeiGene in innovative pharmaceuticals, reducing exposure to second-tier internet companies [12]. Group 2: Market Performance - The index has shown high elasticity in market performance, outperforming similar indices during various market cycles since 2014 [22]. - Since the end of 2014, the annualized return of the index has exceeded that of other similar indices, such as the National Securities Hong Kong Stock Connect Technology Index and the Hang Seng Technology Index [22]. - The index's performance has been driven by significant contributions from the automotive and biopharmaceutical sectors, with distinct phases of market activity observed [22]. Group 3: Fund Introduction - The Southern China Securities Hong Kong Stock Connect Technology ETF (code: 159269) is set to track the index and will be issued starting June 18 [37]. - The fund aims to closely replicate the index's performance, minimizing tracking deviation and error [37].

Market Performance - The Sci-Tech 50 Index decreased by 1.89% over the last five trading days, while the ChiNext Index increased by 0.22%. The large-cap value index rose by 0.10%, and the large-cap growth index fell by 0.16%. The Shanghai 50 Index declined by 0.46%, and the small-cap index represented by the CSI 2000 dropped by 0.74%. The non-ferrous metals and oil & petrochemical sectors performed well, whereas household appliances and food & beverage sectors lagged behind [1]. Risk Premium Analysis - The static PE of the CSI All Index minus the yield of 10-year government bonds indicates a risk premium. Historical extreme bottoms have shown this data to be at two standard deviations above the mean, with notable peaks in 2012, 2018, and 2020. As of April 26, 2022, the risk premium reached 4.17%, and on October 28, 2022, it rose to 4.08%. The latest reading on January 19, 2024, was 4.11%, marking the fifth instance since 2016 exceeding 4%. As of June 13, 2025, the indicator was at 3.83%, with the two standard deviation boundary at 4.75% [1]. Valuation Levels - As of June 13, 2025, the CSI All Index's P/E TTM percentile was at 54%. The Shanghai 50 and CSI 300 indices were at 62% and 52%, respectively. The ChiNext Index was close to 13%, while the CSI 500 and CSI 1000 indices were at 30% and 22%. The ChiNext Index's valuation is relatively low compared to historical averages [2]. Long-term Market Trends - The technical analysis of the Deep 100 Index indicates a bear market every three years, followed by a bull market. Historical declines ranged from 40% to 45%, with the current adjustment starting in Q1 2021 showing sufficient time and space for a potential upward cycle [2]. Fund Flow and Trading Activity - Over the last five trading days, ETF funds saw an outflow of 17 billion yuan, while margin financing increased by approximately 9.4 billion yuan. The average daily trading volume across both markets was 1.3392 trillion yuan [2]. Neural Network Analysis - A convolutional neural network (CNN) was utilized to model price and volume data, mapping learned features to industry themes. The latest recommended themes include non-ferrous metals and banking sectors [9].

Core Viewpoint - The report aims to construct a multi-factor weighted ETF rotation strategy and optimize the stock factor mapping framework to test for marginal improvement in performance [1][5]. Group 1: Background and Market Overview - The concept of index-based investment has gained recognition among investors, with ETFs becoming a significant tool for asset allocation due to their transparency, low fees, and ease of trading. As of April 2025, the number of ETFs listed on domestic exchanges reached 1,141, with a total market value of 4.04 trillion yuan, an increase from 3.73 trillion yuan at the end of 2024 [4]. - Previous reports have constructed product rotation strategies based on various factors such as Level 2 data, redemption data, and neural networks, while also mapping individual stock factors to indices [4]. Group 2: Single Factor Stock Selection and ETF Rotation Comparison - The report compares the performance of single-factor stock selection and ETF rotation, highlighting the differences in factor characteristics. The factors include low-frequency price-volume, fundamentals, ETF redemption fund flows, Level 2 data, and neural network-related factors [6][7]. - Backtesting results indicate that the long position in stocks achieved significant excess returns compared to broad indices, while the RANK_IC and annualized returns of the ETF rotation factors showed marginal declines [11][15]. Group 3: ETF Rotation Backtesting Framework Adjustments - The report attempts to retain only the top-weighted components by setting an upper limit on weight thresholds and applying equal-weight "non-linear mapping." The results show that adjusting to equal-weight mapping covering all components led to a marginal decline in the performance of the single-factor long position [2]. - The ETF selection process identifies ETFs with similar holdings but different names, filtering out products with lower factor values when the overlap of constituent stocks exceeds a threshold. When an 80% overlap threshold is set, the performance of the single-factor long position improves [2]. Group 4: Multi-Factor Weighted ETF Rotation Empirical Testing - The empirical testing of the multi-factor weighted ETF rotation strategy, starting from January 2021, shows that the RANK_IC and ICIR of the portfolio improved marginally with monthly rebalancing. The equal-weighted top 5 long position achieved an annualized return of 18.6%, while the IC-weighted and ICIR-weighted portfolio achieved approximately 20% annualized returns, indicating more stable performance compared to the equal-weighted portfolio [2][15]. Group 5: ETF Rotation Backtesting Empirical Results - The backtesting results for ETF rotation indicate that the RANK_IC and annualized returns of the factors showed a noticeable marginal decline, with the ICIR significantly lower than that of the stock selection backtesting results [15].