金融工程 | 金工定期报告 金融工程 证券研究报告 量化择时周报：缓和预期仍存，调整空间或有限 缓和预期仍存，调整空间或有限 节前周报(20250928)认为：进入国庆长假，假期的不确定性或对市场风险 偏好有所压制；WIND 全 A 趋势线位于 6184 点附近，赚钱效应约为 0.65%， 仍然为正，在赚钱效应转负之前，建议耐心持有。考虑长假的不确定性， 可调仓红利板块应对。 WIND 全 A 上周下跌 0.36%，市值维度上，上周代 表小市值股票的中证 2000 下跌 0.06%，中盘股中证 500 下跌 0.19%，沪深 300 下跌 0.51%，上证 50 下跌 0.47%;上周中信一级行业中，表现较强行业 包括有色金属、煤炭，有色金属上涨 4.35%，传媒、消费者服务表现较弱， 传媒下跌 3.58%。上周成交活跃度上，煤炭、钢铁资金继续流入明显。 从择时体系来看，我们定义的用来区别市场整体环境的 wind 全 A 长期均 线（120 日）和短期均线（20 日）的距离继续缩小，最新数据显示 20 日 线收于 6237，120 日线收于 5525 点，短期均线继续位于长线均线之上， 两线差值由上周的 12 ...

证券研究报告 | 金融工程 2025 年 10 月 12 日 利率价量择时信号维持看多 ——利率市场趋势定量跟踪 20251011 利率市场结构变化 - 10 年期国债到期收益率录得 1.82%，相对上周下降 3.99BP。当前 利率水平、期限和凸性结构读数分别为 1.63%、0.45%、-0.03%， 从均值回归视角看，目前处于水平结构点位偏低、期限结构点位 中性偏低、凸性结构点位较低的状态。 利率价量周期择时信号：5 年期看多、10 年期看多、30 年期看多 美债价量周期择时信号：看多 - 基于美国市场 10 年期国债 YTM 数据判断的多周期择时信号为： 长周期向上突破、中周期向下突破、短周期向下突破。综合来看， 当前合计下行突破 2 票、上行突破 1 票，最终信号的综合评分结果 为看多。 国内利率价量多周期择时策略表现 - 自 2024 年底以来，基于 5/10/30 年期国债 YTM 价量趋势的交易策 略年化收益率分别为 1.86%、2.35%、2.98%，最大回撤为 0.79%、 1.06%、1.87%，收益回撤比为 3.15、4.07、3.26，相对业绩基准的 超额收益率为 0.86%、1.56 ...

主动量化策略周报 股票涨基金跌，成长稳健组合年内满仓上涨 62.19% 核心观点 金融工程周报 国信金工主动量化策略表现跟踪： 本周，优秀基金业绩增强组合绝对收益-0.98%，相对偏股混合型基金指数超 额收益 0.54%。本年，优秀基金业绩增强组合绝对收益 29.30%，相对偏股 混合型基金指数超额收益-4.01%。今年以来，优秀基金业绩增强组合在主动 股基中排名 54.63%分位点（1895/3469）。 本周，超预期精选组合绝对收益 0.22%，相对偏股混合型基金指数超额收益 1.74%。本年，超预期精选组合绝对收益 47.41%，相对偏股混合型基金指 数超额收益 14.10%。今年以来，超预期精选组合在主动股基中排名 21.71% 分位点（753/3469）。 本周，券商金股业绩增强组合绝对收益-1.51%，相对偏股混合型基金指数超 额收益 0.01%。本年，券商金股业绩增强组合绝对收益 34.07%，相对偏股 混合型基金指数超额收益 0.76%。今年以来，券商金股业绩增强组合在主动 股基中排名 44.42%分位点（1541/3469）。 本周，成长稳健组合绝对收益-0.08%，相对偏股混合型基金指数超 ...

- The high-frequency skewness factor showed strong performance with long-short returns of 0.9%, 4.93%, and 22.69% for the past week, September, and 2025, respectively[5][9] - The intraday downside volatility proportion factor had long-short returns of 0.77%, 5.18%, and 18.23% for the past week, September, and 2025, respectively[5][9] - The post-open buying intention proportion factor had long-short returns of 1.11%, 3.65%, and 19.98% for the past week, September, and 2025, respectively[5][9] - The post-open buying intention intensity factor had long-short returns of 1.62%, 3.28%, and 25.81% for the past week, September, and 2025, respectively[5][9] - The post-open large order net buying proportion factor had long-short returns of 0.34%, 1.51%, and 20.7% for the past week, September, and 2025, respectively[5][9] - The post-open large order net buying intensity factor had long-short returns of 0.38%, 1.51%, and 12.86% for the past week, September, and 2025, respectively[5][9] - The intraday return factor had long-short returns of 0.98%, 1.26%, and 20.66% for the past week, September, and 2025, respectively[5][9] - The end-of-day trading proportion factor had long-short returns of 1.25%, 4.18%, and 17.74% for the past week, September, and 2025, respectively[5][9] - The average single outflow amount proportion factor had long-short returns of 0.29%, 0.26%, and -0.54% for the past week, September, and 2025, respectively[5][9] - The large order-driven price increase factor had long-short returns of 0.09%, 2.88%, and 8.88% for the past week, September, and 2025, respectively[5][9] - The GRU(10,2)+NN(10) deep learning factor had long-short returns of 1.33%, 8.73%, and 41.75% for the past week, September, and 2025, respectively, with long-only excess returns of 0.71%, 3.42%, and 8.08%[5][9] - The GRU(50,2)+NN(10) deep learning factor had long-short returns of 1%, 7.98%, and 42.75% for the past week, September, and 2025, respectively, with long-only excess returns of 0.63%, 2.99%, and 7.91%[5][9] - The multi-granularity model (5-day label) factor had long-short returns of 0.99%, 6.15%, and 53.09% for the past week, September, and 2025, respectively, with long-only excess returns of 0.5%, 2.56%, and 19.48%[5][9] - The multi-granularity model (10-day label) factor had long-short returns of 0.81%, 5.2%, and 49.1% for the past week, September, and 2025, respectively, with long-only excess returns of 0.37%, 2.97%, and 20.1%[5][9] - The weekly rebalanced CSI 500 AI enhanced wide constraint portfolio had excess returns of -0.99%, -4.8%, and -0.06% for the past week, September, and 2025, respectively[5][11] - The weekly rebalanced CSI 500 AI enhanced strict constraint portfolio had excess returns of -1%, -2.32%, and 2.66% for the past week, September, and 2025, respectively[5][11] - The weekly rebalanced CSI 1000 AI enhanced wide constraint portfolio had excess returns of -1.48%, -1.06%, and 7.53% for the past week, September, and 2025, respectively[5][11] - The weekly rebalanced CSI 1000 AI enhanced strict constraint portfolio had excess returns of -0.79%, -0.12%, and 13.11% for the past week, September, and 2025, respectively[5][11]

每周 一 景：湖南衡阳衡山 点击右上角菜单，收听朗读版 【金工】大类资产及择时观点月报 大类资产 4季度配置信号： 根据 2025年9月底的最新数据，信用利差和期限利差均发出收窄信号，Q4宏观环境预测结果为Imnflation。 行业复合趋势因子组合表现及信号： 2015年1月至2025年9月，行业复合趋势因子组合的累积收益为122.66%,超额收益为48.42%上月（2025年9月）因子 信号为正向，Wind全A当月收益率为2.80%。根据2025年9月底的最新数据，行业复合趋势因子为-0.30出现下滑，但依旧维持正向信号。 风险提示： 模型失效风险、因于失效风险、海外市场波动风险， 【电子】自强，先进制程设备的突破是核心 投资建议： 美国众议院"中美战略竞争特别委员会"出具一份对中国半导体战略的系统性"围堵"建议书，核心逻辑是中国半导体产业的崛起威胁美国国家安全 与全球技术主导地位，报告建议通过出口管制、技术封锁、产业补贴等手段，确保美国及其盟友在全球半导体产业链中的主导地位，从而遏制半导体崛起。我 们认为半导体产业的全球化仍然是不变的追求，但美国政府不断打压限制我国集成电路产业的发展，本土优秀的半导体装 ...

- The report proposes an iterative factor selection strategy to compress the "factor zoo" by systematically evaluating the contribution of new factors to the remaining alpha of the factors using the GRS statistic[2][3][4] - The iterative factor selection process starts with the CAPM model and adds one factor at a time that maximally reduces the remaining alpha of the factors, measured by the decrease in the GRS statistic[3][25][26] - The process stops when the added factor no longer makes the remaining alpha of the factors statistically significant from zero[3][25][26] - The study finds that only 10 to 20 carefully selected factors are needed to effectively explain the performance of 153 factors in the US market, indicating high redundancy among factors[4][17][19] - The selected factors come from 8 out of 13 factor style categories, showing the heterogeneity of the factor set[17][19] - The iterative factor model outperforms common academic models by selecting alternative definitions of value, profitability, investment, or momentum factors, or including alternative factor style categories such as seasonality or short-term reversal[17][19] - The study also confirms that equal-weighted factors exhibit stronger and more diverse alpha, requiring more than 30 factors to cover the factor zoo[4][64][69] - The effectiveness of the method is validated using global data, showing similar core factor sets across different regions, but with the global model explaining US factors better than non-US factors[4][71][75] - The iterative factor selection strategy provides a practical framework for investors to streamline their models by focusing on the most relevant factors[2][3][4] Factor Selection Process Results - The iterative factor selection process results in a model that starts with the CAPM model, which leaves 105 significant alphas (t>2) and 86 significant alphas (t>3) with a GRS statistic of 4.36 and a p-value of 0.00[39][40] - Adding the cash-based operating profits-to-book assets (cop_at) factor reduces the GRS statistic to 3.54, leaving 101 significant alphas (t>2) and 78 significant alphas (t>3) with an average absolute alpha of 3.94%[39][40] - The process continues by adding factors such as change in net operating assets (noa_grla), sales growth (saleq_gr1), and intrinsic value-to-market (ival_me), among others, until the remaining significant alphas are reduced to zero[39][40][41] - The final model includes 15 to 18 factors, depending on the significance threshold, effectively explaining the factor zoo[39][42][43] Comparison with Common Academic Models - The iterative factor model leaves fewer significant alphas compared to common academic models such as the Fama and French five-factor and six-factor models, the q-factor model, and the mispricing model[43][44] - The Barillas et al. (2020) revised six-factor model performs better than other academic models but still leaves 33 significant alphas, while the iterative factor model leaves only 10 significant alphas with four factors and 14 significant alphas with five factors[43][44] Global Factor Analysis - The global factor analysis shows that 11 global factors are needed to cover the global factor zoo at the t>3 threshold, and around 20 factors at the t>2 threshold[73][74] - The global factor model explains US factors better than non-US factors, indicating that international factors have higher and more diverse alpha potential[75][76][77] Rolling Window Analysis - The rolling window analysis shows that the number of factors needed to cover the factor zoo decreases over time, with around 8 factors needed in recent years compared to 15 factors in the early sample period[59][60][61] - The most relevant factor styles over time include low volatility, seasonality, investment, and quality, while the relevance of momentum, short-term reversal, and value has decreased in recent years[59][60][61] Robustness to Alternative Weighting Schemes - The robustness analysis shows that equal-weighted factors require more than 30 factors to cover the factor zoo, while cap-weighted and value-weighted factors require 18 and 19 factors, respectively[64][65][69] - The equal-weighted factor model exhibits higher and more diverse alpha potential, indicating the need for more factors to cover the equal-weighted factor zoo[64][65][69]

Quantitative Models and Construction Methods 1. Model Name: Weekly Rebalancing AI-Enhanced CSI 500 Wide Constraint Portfolio - **Model Construction Idea**: This model aims to enhance the CSI 500 index performance by leveraging AI-based factors while applying wide constraints on portfolio construction [72][73] - **Model Construction Process**: - The model uses deep learning factors (e.g., multi-granularity model with 10-day labels) as the basis for stock selection [72] - Constraints include: - Stock weight: 1% - Industry weight: 1% - Market cap weight: 0.3 - Turnover rate constraint: 0.3 - The optimization objective is to maximize expected returns, represented by the formula: $$ max \sum \mu_{i}w_{i} $$ where \( w_{i} \) is the weight of stock \( i \) in the portfolio, and \( \mu_{i} \) is the expected excess return of stock \( i \) [73][74] - **Model Evaluation**: The model demonstrates moderate performance under wide constraints, with cumulative excess returns shown over time [75][77] 2. Model Name: Weekly Rebalancing AI-Enhanced CSI 500 Strict Constraint Portfolio - **Model Construction Idea**: Similar to the wide constraint model but applies stricter constraints to control risk and enhance robustness [72][73] - **Model Construction Process**: - Constraints include: - Stock weight: 1% - Industry weight: 1% - Market cap weight: 0.1 - Additional constraints: - Market cap squared: 0.1 - ROE: 0.3 - SUE: 0.3 - Volatility: 0.3 - Component stock constraint: 0.8 - Optimization objective remains the same as the wide constraint model [73][74] - **Model Evaluation**: The stricter constraints result in a more stable performance, with cumulative excess returns displayed over time [76][80] 3. Model Name: Weekly Rebalancing AI-Enhanced CSI 1000 Wide Constraint Portfolio - **Model Construction Idea**: This model targets the CSI 1000 index, applying wide constraints while leveraging AI-based factors for enhanced returns [72][73] - **Model Construction Process**: - Constraints are similar to the CSI 500 wide constraint model, with a focus on smaller-cap stocks [73] - **Model Evaluation**: The model shows significant cumulative excess returns, particularly in recent years [79][86] 4. Model Name: Weekly Rebalancing AI-Enhanced CSI 1000 Strict Constraint Portfolio - **Model Construction Idea**: Similar to the wide constraint model but applies stricter constraints to manage risk and improve consistency [72][73] - **Model Construction Process**: - Constraints are similar to the CSI 500 strict constraint model, tailored for the CSI 1000 index [73] - **Model Evaluation**: The model demonstrates strong performance under strict constraints, with cumulative excess returns highlighted [85][87] --- Model Backtesting Results 1. Weekly Rebalancing AI-Enhanced CSI 500 Wide Constraint Portfolio - **Weekly Excess Return**: -1.36% (last week), -3.85% (September), 0.94% (YTD 2025) [13][78] - **Weekly Win Rate**: 23/39 weeks [13] 2. Weekly Rebalancing AI-Enhanced CSI 500 Strict Constraint Portfolio - **Weekly Excess Return**: -1.35% (last week), -1.33% (September), 3.70% (YTD 2025) [13][81] - **Weekly Win Rate**: 24/39 weeks [13] 3. Weekly Rebalancing AI-Enhanced CSI 1000 Wide Constraint Portfolio - **Weekly Excess Return**: 0.40% (last week), 0.42% (September), 9.15% (YTD 2025) [13][83] - **Weekly Win Rate**: 26/39 weeks [13] 4. Weekly Rebalancing AI-Enhanced CSI 1000 Strict Constraint Portfolio - **Weekly Excess Return**: -0.19% (last week), 0.67% (September), 14.01% (YTD 2025) [13][90] - **Weekly Win Rate**: 25/39 weeks [13] --- Quantitative Factors and Construction Methods 1. Factor Name: Intraday Skewness Factor - **Factor Construction Idea**: Captures the skewness of intraday stock returns to identify potential outperformers [6][8] - **Factor Construction Process**: Referenced in the report "Stock Selection Factor Series Research (19)" [13] - **Factor Evaluation**: Demonstrates strong performance with IC values of 0.027 (historical) and 0.042 (2025) [9][10] 2. Factor Name: Downside Volatility Proportion Factor - **Factor Construction Idea**: Measures the proportion of downside volatility in realized volatility to assess risk-adjusted returns [6][8] - **Factor Construction Process**: Referenced in the report "Stock Selection Factor Series Research (25)" [18][20] - **Factor Evaluation**: Shows moderate performance with IC values of 0.025 (historical) and 0.036 (2025) [9][10] 3. Factor Name: Post-Open Buying Intensity Factor - **Factor Construction Idea**: Quantifies the intensity of buying activity after market open to identify short-term momentum [6][8] - **Factor Construction Process**: Referenced in the report "Stock Selection Factor Series Research (64)" [22][26] - **Factor Evaluation**: Displays stable performance with IC values of 0.035 (historical) and 0.030 (2025) [9][10] 4. Factor Name: Deep Learning Factor (Improved GRU(50,2)+NN(10)) - **Factor Construction Idea**: Utilizes a gated recurrent unit (GRU) and neural network (NN) architecture to predict stock returns [6][8] - **Factor Construction Process**: Combines GRU with NN to capture temporal dependencies in high-frequency data [61][62] - **Factor Evaluation**: Strong performance with IC values of 0.066 (historical) and 0.050 (2025) [12][61] --- Factor Backtesting Results 1. Intraday Skewness Factor - **IC**: 0.027 (historical), 0.042 (2025) [9][10] - **Multi-Long-Short Return**: 3.82% (September), 16.22% (YTD 2025) [9][10] 2. Downside Volatility Proportion Factor - **IC**: 0.025 (historical), 0.036 (2025) [9][10] - **Multi-Long-Short Return**: 2.86% (September), 13.58% (YTD 2025) [9][10] 3. Post-Open Buying Intensity Factor - **IC**: 0.035 (historical), 0.030 (2025) [9][10] - **Multi-Long-Short Return**: 0.65% (September), 11.29% (YTD 2025) [9][10] 4. Deep Learning Factor (Improved GRU(50,2)+NN(10)) - **IC**: 0.066 (historical), 0.050 (2025) [12][61] - **Multi-Long-Short Return**: 2.13% (September), 7.40% (YTD 2025) [12][61]

Group 1 - The article discusses a series of upcoming online meetings focused on various sectors, including renewable energy, AI, and the electronic industry [1][22]. - The first meeting on September 29 will cover "Renewable Energy + AI" led by Liu Qiang, the Chief Analyst of the Electric New Industry [1][22]. - The second meeting on the same day will focus on "Industry Configuration Model Review and Update Series" led by Liu Xiaofeng, the Chief Analyst of Financial Engineering [1][22]. Group 2 - An electronic industry investment outlook meeting is scheduled for September 30, featuring Zhang Shijie, the Chief Analyst of the Electronic Industry [1][12]. - A deep dive into the new stock of "Laoxiangji" will take place on October 10, presented by Guo Mengjie and Lin Xuxi, analysts in the food and beverage sector [1][20].

Core Viewpoint - The articles present a comprehensive collection of trading strategies and research reports focused on index futures and options, emphasizing quantitative methods and market timing techniques [2][3]. Group 1: Index Futures Trading Strategies - The series includes various strategies such as noise trend trading based on chaos theory, trend-following strategies using polynomial fitting, and day trading systems based on intraday volatility extremes [2]. - Additional strategies cover genetic programming methods for intelligent trading, statistical language models for timing trades, and deep learning approaches for intraday trading [2][3]. - The reports also explore cross-variety arbitrage strategies and high-frequency trading techniques, indicating a focus on both theoretical and practical applications in the futures market [3]. Group 2: Derivatives and Options Strategies - The derivatives series provides foundational knowledge on options, including dynamic hedging strategies and volatility arbitrage [3]. - It discusses the impact of options on the underlying assets and market dynamics, highlighting the importance of options in institutional investment strategies [3]. - The reports also analyze the development of global individual stock options markets and their implications for market participants [3].

Group 1 - The report indicates that the overall asset allocation strategy has experienced fluctuations due to domestic equity asset volatility, with various models recording different degrees of decline [1][4][7] - The performance of major asset classes from September 15 to September 19, 2025, shows that the S&P 500, Hang Seng Index, and other indices recorded gains, while convertible bonds and gold experienced declines [7][10] - The domestic asset BL model 1 and model 2 both reported a weekly return of -0.04%, while the global asset BL models had slightly better performance with a return of -0.01% for model 1 and -0.03% for model 2 [15][17] Group 2 - The Black-Litterman (BL) model is highlighted as an improvement over traditional mean-variance models, integrating subjective views with quantitative models to optimize asset allocation [12][13] - The domestic asset risk parity model achieved a return of -0.02% for the week, while the global asset risk parity model recorded a positive return of 0.05% [21][22] - The macro factor-based asset allocation strategy reported a weekly return of -0.1%, with a year-to-date return of 3.25%, indicating its performance amidst changing economic conditions [27][28]