Investment Rating - The report suggests a positive outlook for C-end stock trading software companies and recommends focusing on companies like Zhinancai and Jiufang Zhituo Holdings, while also highlighting the potential recovery in IT investments for securities firms and banks, recommending attention to companies such as Hengsheng Electronics, Dingdian Software, and Yuxin Technology [1]. Core Viewpoints - The capital market IT sector showed signs of recovery in 2024, with a notable increase in A-share average daily trading volume by 21.2% year-on-year, leading to improved trading sentiment and revenue growth for C-end trading software companies [9][11]. - The banking IT sector faced challenges, with a decline in financial technology investments from major state-owned banks and a decrease in revenue for listed banking IT companies [24][26]. - The financial IT companies are expected to leverage new growth drivers such as domestic innovation (Xinchang), AI, and international expansion in 2025 [1][32]. Summary by Sections 1. Capital Market IT Performance - In 2024, the total revenue of 150 securities firms reached 451.2 billion yuan, a year-on-year increase of 11.2%, with net profit rising by 21.3% to 167.3 billion yuan [9][11]. - The average daily trading volume of A-shares was 1,063.4 billion yuan, reflecting a significant improvement in trading activity [9]. - C-end capital market IT companies performed well, with revenue growth exceeding 15% for companies like Zhinancai and Tonghuashun, while B-end companies faced revenue declines [22][23]. 2. Banking IT Sector Overview - The total revenue of 19 listed banking IT companies in 2024 was 67.459 billion yuan, down 4.55% year-on-year, indicating a slowdown in revenue growth [26][30]. - The overall IT investment from the six major state-owned banks slightly decreased to 124 billion yuan, with a year-on-year decline of 0.15% [25][24]. - The number of IT personnel in banking IT companies increased by 3.7% to 168,109, but revenue per employee decreased by 7.35% [27][30]. 3. Growth Drivers for Financial IT Companies - Financial IT companies are actively exploring growth opportunities in domestic innovation, AI, and international markets, with expectations for significant demand growth in 2025 [1][32]. - The report identifies three categories of AI deployment: providing integrated machines for rapid model deployment, enhancing existing product lines with AI capabilities, and developing code assistants to improve efficiency [38][39]. - The report highlights the ongoing progress in domestic innovation projects, with several companies achieving significant milestones in adapting their products to meet domestic standards [36][37].

PPI：油价强势反弹 生产：轮胎开工率降至年内次低 需求：集运运价指数止跌 CPI：猪价低位拉锯 统计口径误差。数据统计大多为抽样，恐与现实情况有些许出入。 敬请参阅最后一页特别声明 1 (1) 电厂日耗季节性下行。5 月 13 日，6 大发电集团的平均日耗为 74.7 万吨，较 5 月 6 日的 75.1 万吨下降 0.6%。5 月 7 日，南方八省电厂日耗为 173.7 万吨，较 4 月 28 日的 186.7 万吨下降 7.0%。 (2) 高炉开工率维持高位。5 月 9 日，全国高炉开工率 84.6%，较 5 月 2 日上升 0.3 个百分点；产能利用率 92.1%， 较 5 月 2 日上升 0.1 个百分点。5 月 9 日，唐山钢厂高炉开工率 94.3%，较 5 月 2 日持平。 (3) 轮胎开工率降至年内次低。5 月 8 日，汽车全钢胎（用于卡车）开工率 44.8%，较 5 月 1 日下降 11.5 个百分点； 汽车半钢胎（用于轿车）开工率 58.4%，较 5 月 1 日下降 14.1 个百分点。 (4) 江浙地区织机开工率小幅回升。5 月 8 日，江浙地区涤纶长丝开工率 92.0%，较 5 月 1 ...

Report Industry Investment Rating No relevant content provided. Core Viewpoints - The yield of ultra-long credit bonds continues to decline, and the number of outstanding ultra-long credit bonds with a yield below 2.2% has increased to 261 compared to last week [2][13]. - The subscription enthusiasm for ultra-long credit bonds has reached a high level. The supply of new ultra-long credit bonds in the first week after the holiday was still slow, with a weekly issuance volume of less than 9 billion. The average issuance rate of ultra-long industrial bonds has decreased marginally to 2.35%, while the average coupon rate of ultra-long urban investment bonds has continued to rise to 2.78% [3][21]. - The growth of the ultra-long credit bond index is weak. After the holiday, the ultra-long credit bond index had a slight catch-up increase, but the growth was still conservative, with the weekly increase of AA+ credit bonds over 10 years being only 0.37%. The scenario of rushing to buy ultra-long credit bonds is unlikely to occur, and the willingness of the market to extend the duration of credit bonds is weak [4][30]. Summary by Directory 1. Stock Market Characteristics - The yield of ultra-long credit bonds continues to decline. Due to the early implementation of interest rate cuts and reserve requirement ratio cuts, long-term interest rates have rapidly declined to a low level, further pushing down the yield of ultra-long credit bonds. The number of outstanding ultra-long credit bonds with a yield below 2.2% has increased to 261 compared to last week [2][13]. 2. Primary Issuance Situation - The subscription enthusiasm for ultra-long credit bonds has reached a high level. The supply of new ultra-long credit bonds in the first week after the holiday was still slow, with a weekly issuance volume of less than 9 billion. The average issuance rate of ultra-long industrial bonds has decreased marginally to 2.35%, while the average coupon rate of ultra-long urban investment bonds has continued to rise to 2.78%. Driven by loose expectations and the low new issuance scale this week, the subscription enthusiasm for new ultra-long credit bonds has risen to a high level of over 70% since 2024 [3][21]. 3. Secondary Trading Performance - The growth of the ultra-long credit bond index is weak. After the holiday, the ultra-long credit bond index had a slight catch-up increase, but the growth was still conservative, with the weekly increase of AA+ credit bonds over 10 years being only 0.37% [30]. - The scenario of rushing to buy ultra-long credit bonds is unlikely to occur. As the yields of government bonds over 10 years and medium- and short-term coupon assets approach the lowest levels of the year, investors should theoretically shift part of their bond allocation focus to long-term credit bonds. However, judging from the trend of trading volume, the willingness of the market to extend the duration of credit bonds is weak. Although the weekly trading volume of ultra-long industrial bonds in the mainstream 7 - 10-year maturity has increased, the figure is still lower than that in late March, and the trading sentiment is not sufficient to support the ultra-long bond market [4][33]. - The proportion of new ultra-long credit bond trading in May has continued to rise compared to the previous period, with the figure exceeding 40% in the latest week. The low valuation deviation of ultra-long credit bonds in the latest week is also relatively conservative, and the increase in holdings is restricted by the low spread protection. The proportion of TKN transactions in ultra-long credit bonds this week has also dropped below 70% [4][37]. - In terms of investor structure, public funds have increased their holdings of 5 - 10-year credit bonds by more than 2.4 billion in a single week, and wealth management products have also increased their allocation of general long-term bonds within 10 years. However, insurance companies did not show any buying volume for ultra-long credit bonds this week, possibly shifting some positions to chase equity assets [4][40].

Group 1: Dify Overview - Dify is a specialized workflow application development tool designed to address the challenges of implementing large models in financial research, such as long processing times and high data dependency[2]. - Dify offers high flexibility, modularity, and low-code features, significantly improving efficiency compared to previous frameworks like LangChain[2]. - Dify allows integration with specialized knowledge bases like RAGFlow, enhancing its application capabilities[2]. Group 2: Key Features - Dify supports diverse tool sources, enabling users to connect various tools, including custom-built workflows, enhancing reusability[3]. - The platform allows for low-code application development, making it accessible for non-technical users to build complex workflows[3]. - Dify's Agent feature enables models to actively engage in tasks, such as querying databases or retrieving information, rather than just generating text[53]. Group 3: Application Cases - The "Time Extraction" tool can automatically identify and resolve ambiguous time references in user queries, enhancing data accuracy[4]. - A specialized "Financial Analyst" Agent categorizes user inquiries into stock, industry, concept, or macroeconomic questions, facilitating targeted analysis[4]. - The "Geographic Information Retrieval" Agent can locate addresses in annual reports and search for nearby relevant locations, showcasing Dify's integration capabilities[4]. Group 4: Competitive Advantage - Compared to competitors like FastGPT and Coze, Dify offers a more flexible and customizable platform, suitable for users with specific needs in workflow and knowledge base integration[26]. - Dify's knowledge base supports various data sources and allows for structured processing, enhancing the accuracy and relevance of responses[27].

- The micro-cap stock index rotation signal indicates that the micro-cap stock index relative net value crossed above the annual line on October 14, 2024, and this trend has continued since then[29] - The 20-day closing price slopes of both the micro-cap stock index and the Moutai index are positive, suggesting a favorable outlook for the micro-cap index's relative performance[29] - In the micro-cap timing model, the volatility congestion indicator, which reflects market trading sentiment, fell below the threshold on October 15, 2024, lifting the risk warning signal[29] - The interest rate year-on-year indicator, which is more fundamental, was -20.45%, not triggering the interest rate risk control threshold of 0.3[29] - The micro-cap timing model has not triggered any risk controls, so investors who prefer to hold micro-cap stocks long-term are advised to continue holding[29] Quantitative Factors and Construction Methods 1. Factor Name: Size Factor; Construction Idea: Measures the impact of company size on stock returns; Construction Process: Calculated as the logarithm of the market capitalization (LN_MktCap)[40][51] 2. Factor Name: Value Factor; Construction Idea: Evaluates the stock's valuation; Construction Process: Includes metrics like book-to-price ratio (BP_LR), earnings-to-price ratio (EP_FTTM), and sales-to-enterprise value ratio (Sales2EV)[40][51] 3. Factor Name: Growth Factor; Construction Idea: Assesses the company's growth potential; Construction Process: Includes metrics like single-quarter operating income growth (OperatingIncome_SQ_Chg1Y) and single-quarter revenue growth (Revenues_SQ_Chg1Y)[40][51] 4. Factor Name: Quality Factor; Construction Idea: Measures the company's financial health; Construction Process: Includes metrics like operating cash flow to current debt (OCF2CurrentDebt) and gross margin (GrossMargin_TTM)[40][51] 5. Factor Name: Consensus Expectation Factor; Construction Idea: Reflects market expectations; Construction Process: Includes metrics like expected return on equity change over three months (ROE_FTTM_Chg3M) and target return over 180 days (TargetReturn_180D)[40][51] 6. Factor Name: Technical Factor; Construction Idea: Based on stock price and volume trends; Construction Process: Includes metrics like 20-day turnover mean (Turnover_Mean_20D) and 60-day return volatility (Volatility_60D)[40][51] 7. Factor Name: Volatility Factor; Construction Idea: Measures stock price volatility; Construction Process: Includes metrics like CAPM model residual volatility (IV_CAPM) and Fama-French three-factor model residual volatility (IV_FF)[40][51] 8. Factor Name: Reversal Factor; Construction Idea: Captures the tendency of stock prices to revert to the mean; Construction Process: Includes metrics like 40-day return (Price_Chg40D) and 60-day return (Price_Chg60D)[40][51] Factor Backtesting Results 1. Size Factor, IC Mean: 35.37% (All A-shares), -9.73% (CSI 300), 13.63% (CSI 500), 17.13% (CSI 1000); Long-Short Return: 8.09% (All A-shares), -3.21% (CSI 300), 3.16% (CSI 500), 4.01% (CSI 1000)[42][43] 2. Value Factor, IC Mean: 2.40% (All A-shares), 14.21% (CSI 300), 15.32% (CSI 500), 8.40% (CSI 1000); Long-Short Return: 3.53% (All A-shares), 1.74% (CSI 300), 3.00% (CSI 500), 3.33% (CSI 1000)[42][44] 3. Growth Factor, IC Mean: -4.85% (All A-shares), 0.05% (CSI 300), -5.97% (CSI 500), -17.08% (CSI 1000); Long-Short Return: 2.23% (All A-shares), 4.13% (CSI 300), 1.21% (CSI 500), 0.56% (CSI 1000)[42][43] 4. Quality Factor, IC Mean: -8.43% (All A-shares), -10.63% (CSI 300), -8.60% (CSI 500), -11.51% (CSI 1000); Long-Short Return: 2.07% (All A-shares), 1.47% (CSI 300), 2.50% (CSI 500), 0.71% (CSI 1000)[42][44] 5. Consensus Expectation Factor, IC Mean: -0.33% (All A-shares), 12.78% (CSI 300), -4.34% (CSI 500), 0.08% (CSI 1000); Long-Short Return: 2.76% (All A-shares), 4.51% (CSI 300), 4.52% (CSI 500), 3.72% (CSI 1000)[42][43] 6. Technical Factor, IC Mean: 4.13% (All A-shares), 2.37% (CSI 300), 3.13% (CSI 500), 11.63% (CSI 1000); Long-Short Return: 8.36% (All A-shares), 4.02% (CSI 300), 6.42% (CSI 500), 8.53% (CSI 1000)[42][44] 7. Volatility Factor, IC Mean: 2.14% (All A-shares), 5.04% (CSI 300), 16.85% (CSI 500), 15.33% (CSI 1000); Long-Short Return: 8.14% (All A-shares), 3.35% (CSI 300), 6.02% (CSI 500), 8.12% (CSI 1000)[42][44] 8. Reversal Factor, IC Mean: 11.17% (All A-shares), 17.06% (CSI 300), 30.80% (CSI 500), 31.65% (CSI 1000); Long-Short Return: 6.60% (All A-shares), 2.81% (CSI 300), 6.37% (CSI 500), 7.58% (CSI 1000)[42][43]

中美日内瓦经贸会谈达成重要突破，双方同意将中国对美反制关税从 125%降至 10%，美国对华加征关税从 145%降至 30%，其中 24%关税设 90 天缓冲期。这一结果远超市场预期，反映出美国在谈判中处于更被动立场，其急于取得成果 源于国内经济压力和政治反噬风险，而中国通过与东盟、欧盟等非美国家深化合作争取了谈判主动权。 谈判成功的关键在于中美双方的现实考量。美国面临关税持久战下的供应链中断风险和盟友拖延策略压力，中国则需 避免高关税对出口的长期冲击，同时利用关税战窗口期完成与非美经济体的战略布局。双方在相互经济压力下选择阶 段性妥协，既缓解了短期经贸摩擦，也为后续磋商机制奠定基础。 协议落地后，预计中国 6-7 月出口将因前期抢出口透支需求而回落，但降幅较 145%关税情景明显缓和。政策层面关注 财政对冲工具，市场层面电子产品、纺织等出口链企业压力减轻，国产替代和出海逻辑仍具长期价值。全球风险偏好 回升利好美股，黄金短期波动加剧，人民币贬值压力有所缓解。 风险提示 中美贸易博弈磋商多次反复；美国经济下行加速超预期；国内出口放缓超预期。 敬请参阅最后一页特别声明 1 宏观经济点评 中美关税超预期降级。5 月 ...

- The report tracks the performance of various enhanced index funds, including Huashang Quantitative Selected CSI 300 Enhanced A, which achieved an excess return of 0.66% last week relative to its benchmark[4][37][40] - Among CSI 500 Enhanced Index Funds, Haitong CSI 500 Enhanced A delivered the best performance last week with an excess return of 1.06%[4][37][40] - For CSI 1000 Enhanced Index Funds, Guojin CSI 1000 Enhanced A achieved an excess return of 0.97% last week, while over the past year, Bosera CSI 1000 Enhanced A led with an excess return of 15.32%[4][37][38] - In the CSI 2000 Enhanced Index Funds category, Wanjia CSI 2000 Enhanced A performed best last week with an excess return of 0.66%, while Huixintong CSI 2000 Enhanced A achieved the highest one-year excess return of 20.64%[4][37][38]

Group 1: Bond Performance Insights - The "24 铁道 MTN006B" bond shows a significant valuation price deviation of -0.35% with a net price of 108.70 and a yield of 2.12%[4] - The "21 万科02" bond has a positive deviation of 0.75% with a net price of 96.31 and a yield of 9.50%[5] - The "24 齐商银行永续债 01" bond has a slight positive deviation of 0.07% with a net price of 99.63 and a yield of 3.40%[6] Group 2: Market Trends - The majority of non-financial credit bonds have a transaction maturity concentrated in the 2 to 3-year range, with 3 to 4-year bonds showing the highest discount transaction ratio[2] - Real estate bonds rank high among those with transaction yields exceeding 10%[8] - The construction materials sector exhibits the largest average valuation price deviation among industries[2] Group 3: Risk Considerations - There is a risk of statistical data bias or omission, which may lead to discrepancies in reported bond performance[3] - Bonds with significant valuation price deviations may face credit risk due to changes in issuer qualifications[18]

投资逻辑 计算机行业观点：我们处在一个大的变革期，一方面，上游环节因为大国竞争的愈来愈烈，科技制裁加剧不可逆，而 呈现出更强的国产替代、自主可控的诉求。DS 为代表的高性价比低成本开源模型，证明了中国科技行业的创新引领 能力，也让国产芯片有了更多适配的机会。我们认为，特朗普关税政策对计算机公司基本面影响相对较小，国产替代 /自主可控板块有望受益。1）基本面角度看，计算机下游市场需求中，海外收入占比约 10%，且征税主要针对需要清 关的实体商品，对于软件影响不大，因此预计对于板块总体的业绩端影响较小。少数美国收入占比较高的硬件/软硬结 合产品的公司，在过去几年通过本地化生产、安全库存等策略进行不同程度的规避。且考虑到关税落地后，国内可能 继续出台稳增长的政策，对于以国内需求为主的计算机板块影响正面。2）估值角度看，计算机板块受市场风险偏好 和情绪影响大，短期美股和海外资本市场乃至大类资产的回撤预计一定程度也将影响到中国资本市场，进而带来计算 机板块估值体系的压力，这一点不可忽视。3）受益的角度看，对等制裁背景下，预计国产替代/自主可控迫切度提升， 成为市场关注点，包括信创、华为产业链、工业软件、军工信息化、CI ...

投资逻辑 计算机行业观点：我们处在一个大的变革期，一方面，上游环节因为大国竞争的愈来愈烈，科技制裁加剧不可逆，而 呈现出更强的国产替代、自主可控的诉求。DS 为代表的高性价比低成本开源模型，证明了中国科技行业的创新引领 能力，也让国产芯片有了更多适配的机会。我们认为，特朗普关税政策对计算机公司基本面影响相对较小，国产替代 /自主可控板块有望受益。1）基本面角度看，计算机下游市场需求中，海外收入占比约 10%，且征税主要针对需要清 关的实体商品，对于软件影响不大，因此预计对于板块总体的业绩端影响较小。少数美国收入占比较高的硬件/软硬结 合产品的公司，在过去几年通过本地化生产、安全库存等策略进行不同程度的规避。且考虑到关税落地后，国内可能 继续出台稳增长的政策，对于以国内需求为主的计算机板块影响正面。2）估值角度看，计算机板块受市场风险偏好 和情绪影响大，短期美股和海外资本市场乃至大类资产的回撤预计一定程度也将影响到中国资本市场，进而带来计算 机板块估值体系的压力，这一点不可忽视。3）受益的角度看，对等制裁背景下，预计国产替代/自主可控迫切度提升， 成为市场关注点，包括信创、华为产业链、工业软件、军工信息化、CI ...