Quantitative Models and Construction Methods - **Model Name**: Neutral Constraint Maximum Factor Exposure Portfolio **Construction Idea**: The model aims to maximize the exposure of target factors in the portfolio while maintaining neutrality in industry and style exposures relative to the benchmark index[62][63][64] **Construction Process**: The optimization model is defined as follows: $ \begin{array}{l}\mbox{\it Max}\qquad\quad w^{\prime}\;X_{target}\\ \mbox{\it s.t.}\qquad\quad(w-\;w_{b})^{\prime}X_{ind}=\;0\\ \mbox{\it(w-\;w_{b})}^{\prime}\;X_{Beta}=\;0\\ \mbox{\it|w-\;w_{b}|\leq1\%}\\ \mbox{\it w\geq0}\\ \mbox{\it w^{\prime}B=1}\\ \mbox{\it w^{\prime}1=1}\end{array} $ - **Explanation**: - \( w \): Portfolio weight vector - \( w_b \): Benchmark portfolio weight vector - \( X_{target} \): Factor load matrix for the target factor - \( X_{ind} \): Industry exposure matrix (binary variables) - \( X_{Beta} \): Style factor exposure matrix (e.g., size, valuation, growth) - Constraints ensure neutrality in industry and style exposures, limit deviations from benchmark weights, prohibit short selling, and require full allocation within benchmark constituents[62][63][64] **Evaluation**: The model effectively balances factor exposure maximization with risk control through neutrality constraints[62][63][64] Quantitative Factors and Construction Methods - **Factor Name**: Volatility Factor **Construction Idea**: Captures the performance of stocks with varying volatility levels[16][17] **Construction Process**: - Volatility Factor = \( \frac{DASTD + CMRA + HSIGMA}{3} \) - **Sub-factor Definitions**: - \( DASTD \): Standard deviation of excess returns over 250 trading days, calculated using a half-life of 40 days - \( CMRA \): Cumulative range of log returns over 12 months - \( HSIGMA \): Standard deviation of residuals from beta regression[16][17] **Evaluation**: Demonstrates strong differentiation between high and low volatility stocks, with recent data showing low volatility stocks outperforming high volatility stocks[16][17] - **Factor Name**: Growth Factor **Construction Idea**: Measures growth potential based on revenue and earnings trends[16][17] **Construction Process**: - Growth Factor = \( \frac{SGRO + EGRO}{2} \) - **Sub-factor Definitions**: - \( SGRO \): Regression slope of revenue growth over the past five fiscal years, normalized by average revenue - \( EGRO \): Regression slope of earnings growth over the past five fiscal years, normalized by average earnings[16][17] **Evaluation**: Provides insights into companies with strong growth trajectories, though sensitivity to financial reporting quality is noted[16][17] Factor Backtesting Results - **Volatility Factor**: - Recent one-week multi-long-short return: -2.90% - Recent one-month multi-long-short return: -1.53%[19][20] - **Growth Factor**: - Recent one-week multi-long-short return: 0.24% - Recent one-month multi-long-short return: 3.27%[19][20] Index Enhancement Portfolio Performance - **Portfolio Name**: CSI 1000 Enhanced Portfolio - Recent one-week excess return: 2.04% - Recent one-month excess return: 2.76% - Recent one-year excess return: 17.07%[57][58] - **Portfolio Name**: CSI 500 Enhanced Portfolio - Recent one-week excess return: 0.03% - Recent one-month excess return: -1.56% - Recent one-year excess return: -8.56%[57][58] - **Portfolio Name**: CSI 800 Enhanced Portfolio - Recent one-week excess return: -0.42% - Recent one-month excess return: -0.26% - Recent one-year excess return: 8.40%[57][58] - **Portfolio Name**: CSI 300 ESG Enhanced Portfolio - Recent one-week excess return: -0.11% - Recent one-month excess return: 0.25% - Recent one-year excess return: 6.90%[57][58] - **Portfolio Name**: CSI 300 Enhanced Portfolio - Recent one-week excess return: -0.71% - Recent one-month excess return: 0.51% - Recent one-year excess return: 10.25%[57][58] Annualized Performance Metrics - **CSI 1000 Enhanced Portfolio**: - Annualized excess return: 15.50% - Information ratio: 2.97[59][60] - **CSI 500 Enhanced Portfolio**: - Annualized excess return: 8.70% - Information ratio: 2.07[59][60] - **CSI 800 Enhanced Portfolio**: - Annualized excess return: 7.11% - Information ratio: 2.18[59][60] - **CSI 300 ESG Enhanced Portfolio**: - Annualized excess return: 5.64% - Information ratio: 1.75[59][60] - **CSI 300 Enhanced Portfolio**: - Annualized excess return: 6.39% - Information ratio: 2.33[59][60]

Group 1 - The core viewpoint of the article is to track the performance of various active quantitative strategies developed by GuoXin JinGong, which aim to outperform the median returns of actively managed equity funds [2][3][5] - The report includes four main strategies: Excellent Fund Performance Enhancement Portfolio, Super Expected Selection Portfolio, Broker Golden Stock Performance Enhancement Portfolio, and Growth Stability Portfolio [2][3][5] Group 2 Excellent Fund Performance Enhancement Portfolio - This strategy aims to benchmark against the median returns of actively managed equity funds, utilizing quantitative methods to enhance performance based on the holdings of top-performing funds [6][36] - As of this week, the portfolio achieved an absolute return of 0.35% and a relative excess return of -0.12% compared to the mixed equity fund index [10][38] - Year-to-date, the portfolio has an absolute return of 28.00% and ranks in the 54.37th percentile among active equity funds [10][39] Super Expected Selection Portfolio - This strategy selects stocks based on super expected events and analyst profit upgrades, focusing on both fundamental and technical criteria to build a portfolio [12][42] - This week, the portfolio recorded an absolute return of 0.70% and a relative excess return of 0.23% compared to the mixed equity fund index [20][44] - Year-to-date, it has achieved an absolute return of 46.54%, ranking in the 20.61st percentile among active equity funds [20][44] Broker Golden Stock Performance Enhancement Portfolio - This strategy utilizes a stock pool from broker recommendations, optimizing the portfolio to minimize deviations from the stock pool while aiming to outperform the ordinary equity fund index [17][46] - This week, the portfolio had an absolute return of -0.54% and a relative excess return of -1.01% compared to the mixed equity fund index [21][49] - Year-to-date, it achieved an absolute return of 33.26%, ranking in the 43.07th percentile among active equity funds [21][49] Growth Stability Portfolio - This strategy focuses on growth stocks, prioritizing those with upcoming earnings announcements to capture excess returns during favorable market conditions [27][50] - This week, the portfolio achieved an absolute return of 0.26% and a relative excess return of -0.22% compared to the mixed equity fund index [30][51] - Year-to-date, it has an absolute return of 51.84%, ranking in the 15.31st percentile among active equity funds [30][51]

Core Viewpoint - The recent decline in the US stock market, particularly in semiconductor stocks, is attributed to deeper liquidity concerns rather than just surface-level factors like Federal Reserve warnings and government shutdown risks [1][3]. Group 1: Market Dynamics - The Philadelphia Semiconductor Index fell over 2%, indicating a significant downturn in technology stocks [3]. - The market's reaction is influenced by liquidity expectations, with the Federal Reserve's statements raising concerns about potential tightening of the money supply [13][14]. Group 2: Investment Insights - Understanding liquidity is crucial for investors; it is more important to know where the money is flowing than to predict short-term price movements [14][18]. - Institutions tend to position themselves in advance, as evidenced by the trading behavior of stocks across different sectors, indicating a common strategy of early investment [7][13]. Group 3: Quantitative Analysis - Quantitative models can provide insights into market behavior by analyzing trading patterns and separating transaction activities [3][13]. - Data reveals that while the market may react to negative news, the underlying liquidity concerns are the true drivers of market movements [14][16].

Core Viewpoint - The article emphasizes the importance of excess return (Alpha) in quantitative investment, highlighting the need for thorough analysis and attribution of performance to understand the sources of excess returns and evaluate the effectiveness of quantitative strategies [2][3]. Group 1: Excess Return and Its Calculation - Excess return (Alpha) is defined as the return of an investment portfolio relative to a benchmark, reflecting the ability to outperform passive benchmarks through active management [3]. - The calculation of excess return varies based on the chosen strategy and benchmark, with a core formula being: Excess Return = Portfolio Return - Benchmark Return [3]. - An example illustrates that if a quantitative strategy has a return of 25% while the benchmark (e.g., CSI 300) returns 10%, the simple excess return is 15% [3]. Group 2: Sources of Excess Return - Excess return can be categorized into three components: Pure Alpha, Smart Beta, and Beta, each with different characteristics and risk profiles [3]. - The performance of excess return is influenced by external market factors and the comprehensive investment capabilities of the institution, which are critical for assessing a fund's sustainability of returns [3]. Group 3: Brinson Attribution Model - The Brinson attribution model is a widely used method for performance attribution, breaking down excess return into allocation effect, selection effect, and interaction effect [4]. - The model requires detailed portfolio holding data to accurately assess the contributions of asset allocation and stock selection to excess returns [4]. Group 4: Performance Attribution Example - An example using the Brinson model shows a fund outperforming the CSI 300 by 4.2%, with contributions from asset allocation and stock selection analyzed to determine the sources of excess return [9]. - The analysis reveals that stock selection contributes significantly to excess return, indicating a strong capability in identifying high-performing stocks [9]. Group 5: Barra Risk Model - The Barra risk model is utilized for post-performance analysis, helping to identify risk exposures and optimize investment strategies [10][11]. - The model decomposes risk into various factors, allowing for a detailed understanding of how different risk factors contribute to overall portfolio volatility [13]. Group 6: Risk Management and Optimization - The article discusses the importance of managing risk while maintaining return potential, with specific strategies for adjusting factor exposures to enhance performance [15][16]. - It highlights the need for continuous strategy iteration and adaptation to market conditions to mitigate risks associated with excess returns [17].

After 43+ years working for one investment research company or another, I finally retired. So now, I’m completely independent. And for the first time on Seeking Alpha, I won’t be working based on anybody else’s product agenda. I have only one goal now… to give you the best actionable investment insights I can.I have long specialized in rules/factor-based equity investing strategies. But I’m different from others who share such backgrounds. I don’t serve the numbers. Instead, the numbers serve me… to inspire ...

"最终通过月度更新的动态加权模块合成综合因子，在不失长期稳定性的前提下增强对短期市场变化的 适应能力。该模型的突出优势在于将长周期有效因子与动态短期信息相结合，形成复合AI信号，从而 构建出兼顾稳定性与响应速度的投资组合，进一步提升组合管理的科学性与实战效能。"韩羽辰表示。 中证报中证网讯(记者魏昭宇)9月25日晚间，路博迈基金基金经理韩羽辰在中国证券报"中证点金汇"直 播间表示，目前市场中的量化团队主要可以分为两类：传统量化体系和AI驱动的量化投研体系。而路 博迈基金所使用的是AI驱动的量化投研体系，经过不断迭代升级，目前已经发展到路博迈量化3.5模 型。 韩羽辰表示，路博迈量化3.5模型在原有基础上进行了系统化升级，其核心区别在于对训练目标的重新 定位与应用场景的针对性优化。"在模型训练目标上，我们的模型将重心置于中低频信号的捕捉与提 取，依托深度神经网络融合时序与截面数据，从而识别更具长期稳健性的市场规律。" "数据方面，除了传统量价数据和基本面数据之外，我们还将蕴含丰富日内信息的高频数据、刻画股票 间相互关系的产业链数据、描述分析师预期和舆情信息的文本数据等另类数据都纳入到机器学习模型体 系中，为模型 ...

Core Insights - Quantitative investment has gained popularity among investors due to its rational, scientific, and emotionally stable characteristics. Jiukun Investment, one of the earliest quantitative private equity firms in China, has won over 150 industry awards and maintains competitive performance and scale [2][6]. Performance Metrics - As of the end of August, Jiukun Investment has 13 products with reported performance, achieving an average return of ***% this year. The "Jiukun Day Enjoy CSI 1000 Index Enhanced No. 1" product ranks first in returns this year, with a five-year excess return of ***% and a cumulative return of ***% since inception, showcasing Jiukun's strong long-term investment capabilities in the index enhancement sector [2][4]. AI Integration - With the rapid development of artificial intelligence, Jiukun Investment has positioned itself as a technology company from its inception, establishing an AI team early on and launching an internal lab in 2020. Over 90% of the researchers hired in the past five years have an AI research background, enabling comprehensive AI capability coverage in the investment research team [5][12]. Investment Principles - Jiukun Investment adheres to three core investment principles: "rationality, long-term focus, and scientific approach," which empower its quantitative investment strategies to create long-term value for investors [6][11]. Talent and Organizational Structure - The company emphasizes the importance of talent and organizational structure as core assets in the quantitative field. Jiukun has a diverse team of experts in mathematics, physics, and computer science, and it fosters a culture of free research and collaboration to enhance its research capabilities [10][15]. Long-term Strategy and Market Position - Jiukun Investment has been deeply involved in the quantitative field for over 13 years, accumulating extensive historical data and practical experience. This foundation allows the firm to quickly address new market challenges and develop robust index enhancement strategies, such as the recently launched A500 index enhancement product [16][17].

看着手机里不断弹出的基金限购公告，我不禁想起十年前刚入行时的困惑。那时的我和大多数散户一样，以为这些公告只是例行公事。直到后来通过量化数 据分析，才发现这背后隐藏着机构资金的精妙布局。 格雷厄姆曾说"牛市是普通投资者亏损的主要原因"，这句话在我十年的投资生涯中不断得到验证。牛市中的暴跌往往让散户措手不及，而机构却能精准把握 进出时机。 以2024年9月24日后的行情为例，表面上指数震荡不前，但个股表现却天差地别。亚辉龙和汇金科技就是最好的对比案例。 汇安基金、湘财久盈等机构的限购措施看似是为了保护投资者利益，实则暗含更深层的市场逻辑。作为一名量化投资者，我发现这些动作往往与机构资金的 季节性调仓密切相关。 记得2015年那轮牛市后的惨痛教训：大多数散户在牛市中亏损高达60%，而同期机构却赚得盆满钵满。这让我深刻认识到，决定股价走势的不是表面的涨 跌，而是背后资金的真实流向。 通过长期跟踪量化数据，我发现一个有趣的现象：A股市场早已告别了齐涨共跌的时代。2016年至2019年间，即便在没有牛市的背景下，个人投资者平均收 益仍为负值，而机构投资者却实现了可观的盈利。 | 同版 | 总收益 | 擇时收益 | 造股收 ...

Group 1 - The core viewpoint of the article highlights the dangers of retail investors falling into typical traps during a bullish market, despite the apparent market excitement and significant stock price increases [1][7] - The article emphasizes that the recent surge in the ChiNext index and the market's overall performance are primarily driven by capital flow, rather than genuine value increases in stocks [7][8] - It points out that many retail investors tend to react slowly to market movements, often entering positions after significant price changes have already occurred, which can lead to losses [7][8] Group 2 - The article discusses the importance of understanding institutional behavior and capital flow rather than relying on price movements alone, suggesting that successful investment requires a data-driven approach [7][9] - It mentions specific examples of stocks, such as the dramatic rise of Shangwei New Materials and the decline of Jingyuan Environmental Protection, to illustrate the risks of following market trends without understanding underlying data [1][5] - The article concludes with a call for investors to develop their own quantitative observation systems to track institutional buying and selling activities, reinforcing the idea that knowing "who is buying" is more critical than "what to buy" [8][9]

证券研究报告 | 2025年09月25日 AI 赋能资产配置（十七） AI 盯盘： "9·24"行情案例 策略研究·策略解读 | 证券分析师： | 王开 | 021-60933132 | wangkai8@guosen.com.cn | 执证编码：S0980521030001 | | --- | --- | --- | --- | --- | | 证券分析师： | 陈凯畅 | 021-60375429 | chenkaichang@guosen.com.cn | 执证编码：S0980523090002 | 事项： 金融市场中的短期内快速上涨行情往往因情绪驱动而非基本面改善，容易导致阶段性追高。传统技术分析 （如 KDJ、RSI、MACD、均线体系、成交量、换手率及估值水平等单一指标）虽能提供部分洞察，但其 信号纷杂、滞后性强且受主观经验影响较大，难以有效预警此类脉冲式行情的风险。 为系统性地解决这一问题，本研究旨在构建一个多维度、量化、由人工智能驱动的综合研判框架。研究首 先从趋势、动量、资金流向、估值四个核心维度出发，构建了十二个关键原始指标，形成一个全面刻画市 场状态的多因子体系，并初步判断和市场趋势的关 ...