Core Viewpoint - The article discusses the complexities of market behavior, emphasizing that while short-term fluctuations may seem irrational, the market ultimately reflects underlying realities over time [2][4][15]. Group 1: Market Dynamics - The market index returning to 3300 points is puzzling, suggesting that traditional pricing models may be incomplete, as other factors like policy and market sentiment also influence stock prices [4]. - The current market is characterized by a strong influence from government policies and market emotions, which may distort typical market reactions to economic indicators [4][15]. - The sentiment among investors is shifting, with many believing that government intervention will mitigate the impacts of trade tensions, leading to a more optimistic outlook [15]. Group 2: Long-term Perspectives - Historical trends indicate that markets can recover from downturns, as seen in the aftermath of the COVID-19 pandemic, which initially led to a bull market but was followed by a bear market from 2021 to 2024 [10]. - The resilience of the market is highlighted, suggesting that despite short-term challenges, the economy shows signs of strength and adaptability [16][17]. - Investors are encouraged to focus on long-term strategies rather than short-term market timing, as the market often rewards patience and informed decision-making [3][5]. Group 3: Investor Behavior - Many investors maintain a long-term position in the market, often holding 100% of their portfolio in equities, which reduces the concern of missing out on market movements [5]. - The article suggests that a significant portion of capital is flowing into the stock market as a safer alternative to traditional business ventures, driven by the perception of lower risk [12]. - The current market environment is described as a "patriotic bull market," indicating a collective sentiment among investors to support domestic growth despite external uncertainties [8].

Group 1 - The core opportunity for index-enhanced funds lies in their stable growth in scale and quantity, with 324 funds totaling 212.9 billion yuan as of March 31, 2025 [1][5][2] - The main challenge is the homogenization of multi-factor models, leading to alpha decay and increased drawdown risks in public index-enhanced products [1][5][12] Group 2 - The heuristic style classification method seeks to categorize stocks based on their representative styles, using a seed group as an anchor for clustering stock returns into growth, value, and balanced dimensions [3][28][54] - The essence of domain enhancement is to find commonalities among stocks and apply specialized selection methods for enhancement, which can be based on various dimensions such as investor structure and market style [28][29] Group 3 - Multi-strategy index-enhanced combinations have shown significant performance, with the multi-strategy CSI A500 index-enhanced combination achieving an annualized excess return of 18.22% since 2013 [4][24] - The multi-strategy CSI 300 index-enhanced combination has also performed well, with an annualized excess return of 18.86% since 2013 [4][24] Group 4 - The report highlights the importance of diversifying strategies to mitigate risks, with low correlation between different strategy excess returns, such as a correlation coefficient of 0.15 between growth and value styles [1][4][28] - The performance of various index-enhanced funds has been analyzed, showing that the excess return median and relative maximum drawdown have varied across different funds over the years [8][9][26]

证券研究报告：金融工程报告 研究所 分析师:肖承志 SAC 登记编号:S1340524090001 Email:xiaochengzhi@cnpsec.com 研究助理：金晓杰 SAC 登记编号:S1340124100010 Email:jinxiaojie@cnpsec.com 近期研究报告 小市值强势，动量风格占优——中邮因子周报 20250420 l 风格因子跟踪 本周估值、杠杆、动量因子的多空表现强势，市值、非线性市值、 流动性因子的空头表现较强。 《Meta LIama 4 开源，OpenAI 启动先 锋计划——AI 动态汇总 20250414》 - 2025.04.15 《小市值持续，高低波风格交替—— 中邮因子周报 20250413》 - 2025.04.14 《4 月是否还会有"最后一跌"？ ——微盘股指数周报 20250406》 - 2025.04.07 《"924"以来融资资金防守后均见到 行情低点，仍关注科技配置机会—— 行业轮动周报 20250330》 - 2025.03.31 《英伟达召开 GTC 2025 大会， Skywork-R1V、混元 T1 等推理模型接 连上线——AI 动 ...

Group 1: ETF Rotation Strategy Tracking - The ETF rotation strategy, constructed using GBDT+NN machine learning factors, has shown strong performance in out-of-sample testing, with an annualized excess return of 11.90% and a maximum drawdown of 17.31% [2][12][17] - Recent performance indicates a weekly excess return of 0.77% and a monthly excess return of 1.10%, while the year-to-date excess return stands at -0.19% [20][24] - The strategy's information ratio is 0.68, reflecting its effectiveness in generating excess returns relative to risk [24] Group 2: High-Frequency Factor Overview - High-frequency factors have demonstrated overall strong performance, with the price range factor yielding a year-to-date excess return of 4.79% and the price-volume divergence factor achieving 10.08% [3][20] - The regret avoidance factor has underperformed with a year-to-date excess return of -0.56%, while the slope convexity factor has shown a year-to-date excess return of -3.64% [3][20] - The high-frequency "gold" combination strategy has an annualized excess return of 10.69% and a maximum drawdown of 6.04% [5][60] Group 3: High-Frequency Factor Performance Tracking - The price range factor measures the activity level of stocks within different price ranges, showing strong predictive power and stable performance this year [4][28] - The price-volume divergence factor assesses the correlation between stock price and trading volume, with recent performance indicating a mixed stability [4][39] - The regret avoidance factor reflects investor behavior, showing stable out-of-sample excess returns, while the slope convexity factor illustrates the impact of order book elasticity on expected returns [4][51] Group 4: Combined Strategies Performance - The high-frequency and fundamental resonance combination strategy has an annualized excess return of 14.98% and a maximum drawdown of 4.52% [5][64] - Recent performance for this combined strategy includes a weekly excess return of 0.63% and a monthly excess return of 2.00%, with a year-to-date excess return of 4.69% [67]

一、本周指数增强组合表现 沪深300指数增强组合本周超额收益0.79%，本年超额收益2.38%。 中证500指数增强组合本周超额收益0.58%，本年超额收益2.73%。 中证1000指数增强组合本周超额收益1.17%，本年超额收益4.33%。 二、本周选股因子表现跟踪 沪深300成分股中预期BP、预期净利润环比、BP等因子表现较好。 中证500成分股中单季净利同比增速、标准化预期外盈利、单季超预期幅度 等因子表现较好。 中证1000成分股中预期PEG、标准化预期外收入、BP等因子表现较好。 公募基金重仓股中预期净利润环比、标准化预期外盈利、单季营利同比增速 等因子表现较好。 三、本周公募基金指数增强产品表现跟踪 沪深300指数增强产品本周超额收益最高0.76%，最低-0.44%，中位数 0.17%。 中证500指数增强产品本周超额收益最高1.12%，最低-0.44%，中位数 0.48%。 中证1000指数增强产品本周超额收益最高1.40%，最低-0.30%，中位数 0.55%。 主 要 结 论 一 国信金工指数增强组合表现跟踪 国信金工指数增强组合的构建流程主要包括收益预测、风险控制和组合优化三块，我们分别以 ...

本周估值风格显著，规模因子表现出色 ——因子周报 20250418 金融工程 1. 主要市场指数与风格表现回顾 本周主要宽基指数涨跌不一。北证 50 上涨 3.48%，上证指数上涨 1.19%，中证 2000 上涨 0.75%，沪深 300 上涨 0.59%，中证 800 上涨 0.34%；中证 500 下跌 0.37%，中证 1000 下跌 0.52%，深证成指下跌 0.54%，创业板指下跌 0.64%。 从行业来看，本周银行、房地产、煤炭、综合、石油石化等行业表现居 前；国防军工、农林牧渔、计算机、消费者服务、电子等行业表现居后。 从风格因子来看，最近一周估值因子、规模因子和非线性市值因子的表现 尤为突出，因子多空收益分别为 2.06%、-2.87%和-0.89%。 2. 选股因子表现跟踪 沪深 300 股票池中，本周 120 日成交比率、单季度 ROA 同比、BP 因子 表现较好。中证 500 股票池中，本周标准化预期外盈利、流动比率、单季度营 业收入同比增速因子表现较好。中证 800 股票池中，本周单季度 ROE 同比、 单季度净利润同比增速、单季度营业利润同比增速因子表现较好。中证 1000 股票池 ...

2025 年 4 月 19 日 ——量化组合跟踪周报 20250419 要点 量化市场跟踪 大类因子表现：本周全市场股票池中，动量因子获取正收益 0.69%；市场表现 出动量效应；非线性市值因子、残差波动率因子和市值因子分别获取负收益 -0.58%、-0.64%和-1.02%，市场小市值风格显著；其余风格因子表现一般。 单因子表现：沪深 300 股票池中，本周表现较好的因子有下行波动率占比 (0.92%)、小单净流入(0.55%)、成交量的 5 日指数移动平均 (0.49%)。表现较 差的因子有 5 日反转(-3.17%)、5 分钟收益率偏度(-2.40%)、总资产增长率 (-1.57%)。 中证 500 股票池中，本周表现较好的因子有标准化预期外收入(1.46%)、下行波 动率占比(1.41%)、单季度净利润同比增长率 (1.18%)。表现较差的因子有 5 日 反转(-1.78%)、单季度 EPS (-1.29%)、毛利率 TTM (-0.92%)。 总量研究 流动性 1500 股票池中，本周表现较好的因子有 6 日成交金额的移动平均值 (1.94%)、对数市值因子(1.77%)、6 日成交金额的标准差 ( ...

证券研究报告 | 2025年4月17日 DeepSeek辅助识别财务瑕疵 ——基于财报文本的情感语调的分析 策略研究 · 策略专题 证券分析师：陈凯畅 021-60375429 chengkaichang@guosen.com.cn S0980523090002 证券分析师：王 开 021-60933132 wangkai8@guosen.com.cn S0980521030001 请务必阅读正文之后的免责声明及其项下所有内容 核心观点 请务必阅读正文之后的免责声明及其项下所有内容 ➢ 财务造假样本分析：从CSMAR数据库"财务违规表"筛选2010-2021年样本，通讯服务行业造假占比最高，金融和公用事业最低。2010-2018年造假公 司数量及占比上升，2019年后下降，且约58.3%的造假行为在1-2年内暴露或终止。信息披露违规成为主流，虚构利润和虚列资产减少。 ➢ 特征池构建：基于上市公司定期财务报告，从8个维度构建378个比率型指标，经筛选处理后保留100个指标，形成特征池，包含5483个财务造假样本 和42046个控制样本。 ➢ 情感语调因子构建：利用DeepSeek R1模型分析财报文本情感语调， ...

Report Industry Investment Rating - No relevant content provided Core Viewpoints of the Report - The styrene industry chain starts from crude oil, goes through the production of benzene and ethylene, then to the production of styrene and its derivatives, and is finally applied in multiple fields such as packaging, automotive, electronics, and construction. It is an important part of the petrochemical industry, with characteristics of high dependence on crude oil, a long chain, and wide - ranging demand. The factors affecting styrene futures prices are complex. Fundamental quantitative factors cover 9 aspects, and volume - price factors include 7 aspects. By back - testing and screening, setting parameters such as back - testing time, handling fees, and leverage, and combining factors in a simple equal - weight addition way, a trend strength signal can be output [3]. - The fundamental multi - factor portfolio has an annualized return of 50.7% and a Sharpe ratio of 2.85 since 2019. The volume - price multi - factor portfolio has an annualized return of 35.3% and a Sharpe ratio of 2.14 since 2019. In the comprehensive model, all single factors are equally weighted, with an annualized return of 32.2% and a Sharpe ratio of 1.86 since 2019. Fundamental factors and volume - price factors have a low correlation. Investors can adjust the proportion of fundamental and volume - price factors in the comprehensive model according to their target returns and risk requirements [4]. Summary According to the Directory 1. Styrene Single - Commodity Timing Factor Framework - Styrene is an important organic chemical raw material with a clear upstream - downstream industrial chain. The model divides factors into two categories: fundamental quantitative factors and volume - price factors. Fundamental factors are constructed from dimensions such as inventory, basis, upstream inventory, profit, and overseas prices. Volume - price factors are constructed from dimensions such as momentum, moving averages, and technical indicators based on daily - frequency market data. As of the writing of the report, the model includes 9 fundamental quantitative factors and 7 volume - price factors [8][10]. - When back - testing and screening factors, the back - testing time for most fundamental factors and volume - price factors starts from October 2019, with the out - of - sample back - testing starting from January 2023 and ending in December 2024. The handling fee is set at a bilateral rate of 0.03%, and the leverage is 1x. Other settings such as cumulative return calculation, factor value mapping, and signal update rules are also specified [11][12][13] 2. Introduction and Back - Testing Results of Styrene Fundamental Quantitative Factors 2.1 Styrene Weekly Shipment Volume - A significant increase in styrene weekly shipment volume may lead to an oversupply situation if downstream demand does not increase synchronously, causing price decline. The data used is from the East China region, Jiangsu Province, China, and is published every Monday. Since 2019, its back - testing performance shows an annualized return of 30.3%, a Sharpe ratio of 1.68, a Calmar ratio of 1.23, a win rate of 51.0%, an average holding period of 13.7 days, and a maximum drawdown of 24.6% [19]. 2.2 Styrene Overseas Price - An increase in overseas styrene prices may push up domestic prices, while a decrease may suppress domestic prices. This factor mainly considers prices in the US Gulf, Rotterdam, and South Korea. The data is published with a one - day lag. Since 2016, its back - testing performance shows an annualized return of 19.6%, a Sharpe ratio of 0.99, a Calmar ratio of 0.73, a win rate of 52.5%, an average holding period of 19.1 days, and a maximum drawdown of 27% [21]. 2.3 Styrene Basis - When the market supply is tight, the basis widens; when the supply is excessive, the basis narrows. The data is from the Guojun Futures database and is published daily. Since 2019, its back - testing performance shows an annualized return of 27.7%, a Sharpe ratio of 1.12, a Calmar ratio of 0.71, a win rate of 51.9%, an average holding period of 2.6 days, and a maximum drawdown of 39.1% [23]. 2.4 Pure Benzene: Port Inventory - A low level of pure benzene port inventory may increase the production cost of styrene. The data is from the East China region and is published every Friday. Since 2019, its back - testing performance shows an annualized return of 15.8%, a Sharpe ratio of 0.67, a Calmar ratio of 0.42, a win rate of 50.8%, an average holding period of 38.2 days, and a maximum drawdown of 37.3% [25]. 2.5 Styrene: Non - Integrated Plant: Production Gross Margin - A high production gross margin of non - integrated styrene plants may encourage enterprises to increase production, leading to an increase in market supply. The data is from the Steel Union and is published after the market closes. Since 2019, its back - testing performance shows an annualized return of 12.5%, a Sharpe ratio of 0.46, a Calmar ratio of 0.3, a win rate of 50.4%, an average holding period of 11.3 days, and a maximum drawdown of 34.4% [27]. 2.6 Styrene Capacity Utilization Rate - An increase in styrene capacity utilization rate may lead to an oversupply situation and price decline. The data is from the Steel Union and is published weekly. Since 2019, its back - testing performance shows an annualized return of 16.5%, a Sharpe ratio of 0.91, a Calmar ratio of 0.85, a win rate of 50%, an average holding period of 28.6 days, and a maximum drawdown of 19.4% [27]. 2.7 Styrene Warehouse Receipts - An increase in warehouse receipts indicates sufficient market supply, while a decrease indicates tight supply. The data is from Flush and is published after the market closes. Since 2020, its back - testing performance shows an annualized return of 22.6%, a Sharpe ratio of 1.34, a Calmar ratio of 1.04, a win rate of 50.2%, an average holding period of 9.6 days, and a maximum drawdown of 21.7% [30]. 2.8 Styrene Arbitrage Spread - The internal - external spread has a mean - reversion characteristic. This factor considers styrene prices in Europe, Asia, and the Americas. The data is from the Steel Union and is updated with a one - day lag. Since 2019, its back - testing performance shows an annualized return of 33.8%, a Sharpe ratio of 1.68, a Calmar ratio of 0.93, a win rate of 53.2%, an average holding period of 12.5 days, and a maximum drawdown of 36.4% [32]. 2.9 Styrene: Spot Inventory - High inventory usually means sufficient or excessive market supply, while low inventory may indicate tight supply. The data is from the Steel Union and is updated every Monday. Since 2019, its back - testing performance shows an annualized return of 25.9%, a Sharpe ratio of 1.45, a Calmar ratio of 1.52, a win rate of 52.3%, an average holding period of 114.6 days, and a maximum drawdown of 17.1% [35]. 2.10 Fundamental Multi - Factor - By equally weighting the above 9 fundamental single factors to form a long - short timing model, since 2019, the back - testing shows an annualized return of 50.7%, a Sharpe ratio of 2.85, a Calmar ratio of 2.08, a win rate of 52.6%, an average holding period of 6 days, and a maximum drawdown of 24.4% [37]. 3. Introduction and Back - Testing Results of Styrene Volume - Price Factors 3.1 Intraday Momentum - Intraday momentum is defined as the average of the daily high and low prices divided by the opening price. A larger value indicates a faster price increase. Since 2020, its back - testing performance shows an annualized return of 27.6%, a Sharpe ratio of 1.51, a Calmar ratio of 1.7, a win rate of 47.2%, an average holding period of 3.7 days, and a maximum drawdown of 16.2% [40]. 3.2 Median Double Moving Averages - Similar to double moving averages, but the price for calculating the moving average is the median of the daily high and low prices. Since 2019, its back - testing performance shows an annualized return of 18%, a Sharpe ratio of 0.81, a Calmar ratio of 0.56, a win rate of 51.6%, an average holding period of 8.5 days, and a maximum drawdown of 32.4% [42]. 3.3 Kaufman Adaptive Moving Average (KAMA) - Calculated through steps such as efficiency coefficient (ER) and smoothing constant (SC). Since 2019, its back - testing performance shows an annualized return of 21.1%, a Sharpe ratio of 1.23, a Calmar ratio of 1.19, a win rate of 48.8%, an average holding period of 30.6 days, and a maximum drawdown of 17.8% [45]. 3.4 On - Balance Volume (OBV) - Calculated based on the relationship between daily closing prices and trading volumes, and a long - short double moving average strategy is constructed. Since 2020, its back - testing performance shows an annualized return of 21.2%, a Sharpe ratio of 1.17, a Calmar ratio of 1.28, a win rate of 50.4%, an average holding period of 72.4 days, and a maximum drawdown of 16.6% [49]. 3.5 Commodity Channel Index (CCI) - When CCI breaks through + 100, it is a potential selling signal; when it breaks through - 100, it is a potential buying signal. Since 2019, its back - testing performance shows an annualized return of 28.9%, a Sharpe ratio of 1.72, a Calmar ratio of 1.98, a win rate of 51.0%, an average holding period of 29.9 days, and a maximum drawdown of 12.0% [53]. 3.6 TRIX - Defined through exponential moving averages and a long - short double moving average strategy is constructed based on its daily change rate. Since 2019, its back - testing performance shows an annualized return of 28.9%, a Sharpe ratio of 1.72, a Calmar ratio of 1.98, a win rate of 51.0%, an average holding period of 29.9 days, and a maximum drawdown of 14.6% [55]. 3.7 MESA Adaptive Moving Average - Hilbert transform is used to process price data. MAMA and FAMA lines are calculated, and a double moving average strategy is constructed for timing. Since 2019, its back - testing performance shows an annualized return of 20.5%, a Sharpe ratio of 1.11, a Calmar ratio of 1.11, a win rate of 49.8%, an average holding period of 29.3 days, and a maximum drawdown of 18.5% [55]. 3.8 Volume - Price Multi - Factor - By equally weighting the above 7 volume - price single factors to form a long - short timing model, since 2019, the back - testing shows an annualized return of 35.3%, a Sharpe ratio of 2.14, a Calmar ratio of 2.41, a win rate of 51.5%, an average holding period of 10.3 days, and a maximum drawdown of 14.7% [59]. 4. Fundamental Quantitative and Volume - Price Multi - Factor Comprehensive Model 4.1 All - Factor Portfolio Long - Short Model - By equally weighting all 16 single factors to form a long - short timing model, since 2019, the back - testing shows an annualized return of 32.2%, a Sharpe ratio of 1.86, a Calmar ratio of 2.07, a win rate of 46.6%, an average holding period of 5.1 days, and a maximum drawdown of 15.6% [61]. 4.2 Only - Long Model - Fundamental only - long model: By equally weighting the first 9 single factors, when a short - selling signal is generated, it is regarded as closing the existing long position or staying in cash; when a long - buying signal is triggered, open a long position or hold the existing long contract. Since 2019, the back - testing shows an annualized return of 29.6%, a Sharpe ratio of 1.89, a Calmar ratio of 1.31, an average holding period of 6.7 days, and a maximum drawdown of 22.6%. - Volume - price only - long model: By equally weighting the latter 7 single factors, with similar signal - handling rules. Since 2019, the back - testing shows an annualized return of 22.1%, a Sharpe ratio of 1.57, a Calmar ratio of 1.68, an average holding period of 10.6 days, and a maximum drawdown of 13.1%. - All - factor comprehensive only - long model: By equally weighting all 16 single factors, with similar signal - handling rules. Since 2019, the back - testing shows an annualized return of 20.0%, a Sharpe ratio of 1.32, a Calmar ratio of 1.27, an average holding period of 7.6 days, and a maximum drawdown of 15.8% [64][67][69]. 4.3 Only - Short Model - Fundamental only - short model: By equally weighting the first 9 single factors, when a long - buying signal is generated, it is regarded as closing the existing short position or staying in cash; when a short - selling signal is triggered, open a short position or hold the existing short contract. Since 2019, the back - testing shows an annualized return of 20.0%, a Sharpe ratio of 1.48, a Calmar ratio of 1.28, an average holding period of 6.3 days, and a maximum drawdown of 15.7%. - Volume - price only - short model: By equally weighting the latter 7 single factors, with similar signal - handling rules. Since 2019, the back - testing shows an annualized return of 12.5%, a Sharpe ratio of 0.87, a Calmar ratio of 0.9, an average holding period of 16.7 days, and a maximum drawdown of 13.9%. - All - factor comprehensive only - short model: By equally weighting all 16 single factors, with similar signal - handling rules. Since 2019, the back - testing shows an annualized return of 11.8%, a Sharpe ratio of 0.87, a Calmar ratio of 0.82, an average holding period of 9 days, and a maximum drawdown of 14.4% [72][75][76].

Group 1 - The A-share market showed mixed performance with the Shanghai Composite Index relative to the growth index declining by 0.45%, closing at 2675.86 points and a trading volume of 171.84 billion [1] - The Shanghai Composite Index relative to the growth index has decreased by 5.59% over the past month, increased by 0.79% over the past three months, and has declined by 1.15% year-to-date [1] - The Shanghai Composite Index style index series is based on the Shanghai Composite Index, calculating style scores based on growth and value factors, selecting the top 150 listed companies for the growth and value indices [1] Group 2 - The top ten holdings of the Shanghai Composite Index relative to the growth index include Kweichow Moutai (12.65%), Zijin Mining (3.77%), and others, with the Shanghai Stock Exchange accounting for 100% of the holdings [2] - The industry composition of the holdings in the Shanghai Composite Index relative to the growth index includes Information Technology (19.80%), Industrials (19.31%), Consumer Staples (18.67%), and others [2] Group 3 - The index sample is adjusted every six months, with adjustments occurring on the next trading day after the second Friday of June and December, with a sample adjustment ratio generally not exceeding 20% [3] - In special circumstances, the index may undergo temporary adjustments, and companies that are delisted will be removed from the index sample [3]