在大模型底层创新趋近平稳之际，有百亿量化私募助力AI应用闯进了名片场景。 近日，灵均投资发起了"点点同行，乡乡同兴——AI点亮乡村振兴"公益行动，并推出"点点AI名片"的垂 类应用。以灵均投资董事长蔡枚杰作为创始人的点富科技，参与了这一应用的技术研发和平台的搭建。 "能对话、有科普"是点点名片的特色。据了解，使用者可以在点点名片中构建自己的"AI名片"，也可以 通过AI技术与其他人的名片进行互动交流。 21世纪经济报道记者 黎雨辰 北京报道 当前，中国AI技术发展正处于快车道，制度与市场的双重优势为技术落地提供了良好环境。政策层 面，"新型举国体制"下的算力统筹与科技专项支持加速技术转化；市场层面，2.5 亿生成式AI用户推动 着AI与垂类领域的产业升级、造福社会。 而在其中，国内头部量化私募的资金实力、算力资源以及人才储备，使其具备了进入AI领域的潜力。 从投研的角度出发，事实上2017年以来，AI就已开始走入量化行业，并在提升数据处理分析能力、优 化模型构建与策略创新等方面大显身手。 "AI支持量化行业发展的根源在于对算力的提升。"一位头部量化人士向记者指出，"在策略的制定方 面，AI推动了数据处理分析深 ...

行业研究 上周量化私募超额有所回升，中性策略正收益。上周样本量化私募收益及超额如下： 1）300 增强策略周/月/年初以来平均收益分别为-0.4%/-1.3%/-1.8%，周/月/年初以来平均 超额分别为+0.1%/+0.9%/+3.3%；2）500 增强策略周/月/年初以来平均收益分别为+0.4%/- 1.7%/+2.3%，周/月/年初以来平均超额分别为+0.3%/+2%/+7.1%；3）A500 增强策略周/月 /年初以来平均收益分别为-0.2%/+1.5%/+5.6%，周/月/年初以来平均超额分别为- 0.1%/+4.1%/+11%；4）1000 增强策略周/月/年初以来平均收益分别为+0.7%/-1.2%/+5.4%， 周/月/年初以来平均超额分别为+0.5%/+3%/+9.2%；5）空气指增策略周/月/年初以来平均 收益分别为+0.8%/-0.2%/+7.8%；6）市场中性策略周/月/年初以来平均收益分别为 +0.2%/+0.8%/+4.4%。 指数收益情况：1）沪深 300 相对中证 500 周/月/年初以来超额收益分别为-0.4%/- 0.2%/+1.9%；2）中证 1000 相对中证 500 ...

Group 1 - The core initiative "AI Lights Up Rural Revitalization" aims to connect top resources to rural areas through AI technology, targeting 832 poverty-stricken counties in China [1][2] - The project involves collaboration between Lingjun Investment, Beijing Normal University, and Dianfu Technology, with Olympic champion Zhang Yufei serving as the public ambassador [1] - The AI card integrated into the initiative includes expert resources in four key areas: learning, health, psychology, and finance, designed to enhance knowledge and professional skills for rural residents [2] Group 2 - Lingjun Investment emphasizes the importance of AI technology in empowering education and contributing to social welfare, aligning with the national strategy for educational development [1][2] - The initiative is part of a broader effort to improve the quality and balance of education in underdeveloped regions, leveraging new technologies to innovate educational methods [2] - Zhang Yufei highlights the transformative potential of AI technology, describing it as a warm connection that can unify educational efforts and societal support for rural revitalization [2]

Core Insights - The launch of the first domestic CSI A500 quantitative index enhancement product has gained significant traction, with major quantitative private equity firms achieving stable excess returns over the past six months [1][2] - High-net-worth clients have shown increased recognition and enthusiasm for the CSI A500 quantitative index enhancement strategy, leading to higher subscription rates than previously expected [2][3] - The competitive landscape among institutions has intensified, with firms focusing on diverse factors, technical innovation, and refined operations to differentiate their offerings [4][5] Performance and Market Trends - Leading quantitative private equity firms like Blackwing Asset and Siyue Investment have reported notable excess returns, with Blackwing achieving particularly high rates [2][3] - The overall performance of CSI A500 quantitative index enhancement products has been strong, with most products outperforming benchmark indices [2][3] - The market for CSI A500-related products has seen a surge in registrations and fundraising, indicating high investor interest and confidence in the strategy [3][6] Future Outlook - The growth trend of CSI A500 index-linked products is expected to continue, potentially bringing more incremental capital into the A-share market [6][7] - The balanced industry distribution of CSI A500 index components makes it suitable for quantitative strategies, likely attracting further investment [7] - As more quantitative institutions enter this space, competition will intensify, necessitating continuous improvement in core competencies and innovative strategies [7]

这几天到处都能刷到关于deepseek的话题，这是国产的一个ai，经过测试目前已经达到了开源模型里的一线水平，除了高性能，它还有一个标签就是经济 实惠，它的训练费用不到600万美元，运算成本比同行低了90%以上，人送外号"ai拼多多"。 便宜又厉害的模型是怎么做出来的呢？我稀里糊涂看了不少文章，过于专业，我没看懂，这题pass。 关于deepseek的第二个新闻点就有意思了，出资研发它的竟然是国内著名量化私募机构幻方，幻方的老板梁文峰持有deepseek公司83%的股权。也正是由 于这一层关系，这几天圈子里有不少人在哀嚎，说怪不得量化私募持续不断的在a股收割，原来人家早就用ai炒股，用机枪扫射还在用长矛弓箭的散户。 对此幻方倒是回应过，说deepseek团队就是专门做通用大模型的，并没有用它来炒股。话是这么说，但研究deepseek的资金和人才很多都是从幻方量化输 送过去的，能研究出成功ai模型的团队，在股票模型的开发上肯定也不弱。毕竟ai模型现在没有一个挣钱，但炒股是真挣钱。 坊间一直传闻幻方量化曾对超算集群系统投入 10 亿元，搭载了超 1 万张英伟达 A100 显卡。 然后就是今天又爆出一个热搜，说雷军 ...