中金公司 景气跃迁：量化视角下的盈利预测与"预期差" 挖掘 20250710 摘要 量化景气投资侧重广度，通过预测股票利润增长排名而非具体数值，以 实现投资收益。理想化测试表明，准确预测 ROE 变化并持仓排名靠前的 股票能带来超额收益，验证了该方法的可行性。 基于财务信息，当期业绩高增长的股票有较高概率延续高增长，但存在 业绩变脸风险。通过引入加速度概念，即增速的变化，可以优化模型， 提高预测准确性并降低风险。 二次趋势外推模型通过考虑利润增速和加速度，在预测胜率（72%）和 假阳性率（13%）方面均优于线性外推和分析师一致预期，显著改善了 盈利预测效果。 "成长趋势共振选股策略"结合优化后的盈利预测模型、分析师预期、估 值和现金流等因子，选取约 30 支股票，自 2009 年回测以来表现优异， 并在样本外跟踪中持续实现超额收益。 引入机器学习方法，特别是 XGBoost 和 LightGBM 等树模型，能处理 更多维度数据并捕捉非线性关系，显著提升盈利预测的准确性，胜率可 达 85%，假阳性率降至 4.7%。 Q&A 景气型投资的传统思路是什么？ 景气型投资是一种主流的机构化投资思路，传统上更多依赖于基本 ...

今天为大家整理了几十篇具身相关的综述，设计数据集、评测、VLA、VLN、强化学习、基础模 型、DP等方向，为大家一览具身发展的路线， 内容出自具身智能之心知识星球。 A Survey on Vision-Language-Action Models: An Action Tokenization Perspective.2025 论文链接：https://arxiv.org/pdf/2507.01925 A Survey on Vision-Language-Action Models for Autonomous Driving.2025 论文链接：https://arxiv.org/pdf/2506.24044 Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes.2025 论文链接：https://www.arxiv.org/abs/2408.03539 A Survey on Diffusion Policy for Robotic Manipulation: Taxonomy, Analysis, and ...

下大气力抓落实、抓执行，自上而下的学习热潮迅速在四川国资国企系统内掀起。 健全工作机制，强化统筹指导。17户省属监管企业的组织部长均被纳入省国资委工作专班联络指导 组，建立周调度、月总结机制，综合运用工作推演、定期调度、随机调研等方式，动态掌握企业开展学 习教育进展情况，及时发现并纠正偏差和问题。 紧扣主题主线，一体推进学查改。集中学习、常态学习、以训助学……多形式开展学习的同时，四 川国资国企系统全面深入查摆问题，省国资委领导班子通过深入调研、广泛征求意见、认真对照检查等 方式，带头全面查摆5个方面问题，建立把关备案管理机制，确保问题查摆精准、深入，同时扎实推进 整改落实，持续深化督导指导。 日前，正在建设中的泸石高速传来喜讯——泸定段45公里主线贯通，为实现今年泸定段分段式通车 奠定坚实基础。为保证重点项目"加码提速"，蜀道集团开展"旗帜领航当先锋·红色堡垒促发展"专项行 动，推动党员、干部砥砺作风、担当作为。 这是四川国资国企系统推进深入贯彻中央八项规定精神学习教育的一个切片。自学习教育启动以 来，四川国资国企系统提高政治站位，深刻领会党中央重大政治考量，把思想和行动统一到党中央和省 委决策部署上来，把 ...

Core Insights - The rapid development of AI Agents has ignited a trend of "everything can be an Agent," particularly evident in the competitive landscape of model development and application [1][2][10] - Major companies like OpenAI, Google, and Alibaba are heavily investing in the Agent space, with new products emerging that enhance user interaction and decision-making capabilities [2][7][8] - The evolution of AI applications is categorized into three phases: prompt-based interactions, workflow-based systems, and the current phase of AI Agents, which emphasize autonomous decision-making and tool usage [17][19] Group 1: Model Development - The AI sector has entered a "arms race" for model development, with significant advancements marked by the release of models like DeepSeek, o3 Pro, and Gemini 2.5 Pro [5][6][14] - The introduction of DeepSeek has demonstrated that there is no significant gap between domestic and international model technologies, prompting major players to accelerate their model strategies [6][10] - The focus has shifted from "pre-training" to "post-training" methods, utilizing reinforcement learning to enhance model performance even with limited labeled data [11][13] Group 2: Application Development - The launch of OpenAI's Operator and Deep Research has marked 2025 as the "Year of AI Agents," with a surge in applications that leverage these capabilities [7][8] - Companies are exploring various applications of AI Agents, with notable examples including Cursor and Windsurf, which have validated product-market fit in the programming domain [9][21] - The ability of Agents to use tools effectively has been a significant breakthrough, allowing for enhanced information retrieval and interaction with external systems [20][21] Group 3: Challenges and Opportunities - Despite advancements, AI Agents face challenges such as context management, memory mechanisms, and interaction with complex software systems [39][40] - The future of Agent applications may involve evolving business models, potentially shifting from subscription-based to usage-based or outcome-based payment structures [40][41] - The industry is witnessing a competitive landscape where vertical-specific Agents may offer more value due to their specialized knowledge and closer user relationships [42][46]

Core Viewpoint - The meeting focused on advancing the rectification of issues identified in the central ecological environment protection inspection, emphasizing the importance of implementing the "Ecological Environment Protection Inspection Work Regulations" to enhance ecological governance and promote sustainable development in Jiangsu [1][2]. Group 1: Key Actions and Strategies - The provincial leadership stressed the need for a pragmatic approach to ensure thorough rectification of inspection issues, with a focus on systematic and comprehensive measures to address problems effectively [2]. - Emphasis was placed on the importance of accountability, with a clear delineation of responsibilities among party and government officials to ensure effective implementation of rectification measures [2][3]. - Continuous monitoring and emergency response mechanisms are to be strengthened to combat pollution and ensure the safety of water bodies, particularly Lake Tai, during the summer [3]. Group 2: Regulatory Framework and Governance - The "Ecological Environment Protection Inspection Work Regulations" were highlighted as a crucial framework for standardizing inspection processes and enhancing the effectiveness of ecological governance [1][4]. - The meeting included discussions on the progress of rectification efforts across various cities and departments, indicating a collaborative approach to addressing ecological issues [4]. - The need for innovation and reform in ecological governance was emphasized, with a focus on integrating technology to improve monitoring and problem-solving capabilities [3].

习近平总书记指出："制定实施中央八项规定，是我们党在新时代的徙木立信之举，必须常抓不懈、久 久为功。"自今年3月在全党开展深入贯彻中央八项规定精神学习教育以来，金融机构聚焦主题、注重成 效，在一体推进学查改上再深化，引导广大从业人员锲而不舍落实好中央八项规定，为推动金融行业高 质量发展营造良好氛围。在上级党委的指导下，浙商期货党总支扛起主体责任，坚持问题导向，不断创 新学习形式，以更高要求、更严标准开展全面学、重点查、持续改，切实把学习成果转化成推动发展、 服务群众的实际成效，为做好金融工作、践履金融职责筑牢作风根基。 抓学习，坚持深学细悟筑根基 在全党开展深入贯彻中央八项规定精神学习教育，是今年党建工作的重点任务。在3月接到上级党委通 知后，浙商期货党总支迅速响应，积极落实，第一时间组织召开主题学习会，对学习教育进行系统部 署，有序推进，以深学、严查、长改的形式持续推动学习教育走深走实。 强化思想，突出成效。浙商期货党总支坚持把学习教育作为工作主线，采取书记带学、党支部领学、党 员群众共学的形式，通过"三会一课"、主题党日、线上学习等方式，深入学习研讨，扎实贯彻落实习近 平总书记重要讲话和重要指示精神，在对 ...

去年有一个双非的同学找到我们，情况是：老师没有人带，但自己想申请博士，想咨询有没有快速发表论文的 渠道。在分析这位同学的基础和硬件资源后，我们为他快速制定了一个研究方向，并匹配到了相关的老师！经 过近10个月的沟通、实验、写作，最终成功投出到了CVPR25，并被录取。成为学院首个发CVPR的硕士研究 生。 SCI一区~四区； 中科院1区，2区，3区，4区； 谈到这个，归咎于2点。没人指导不可怕，可怕的是自己不行动，主动出击才有胜算。如果当时没有主动找老 师辅导，也许CVPR对他来说只是一个梦。还有就是同学性格很主动、肯吃苦，经常分析到凌晨。遇到问题不 逃避，敢于直面！ EI/中文核心； 毕设论文/申博/比赛等； 如果你缺乏指导、身边没有老师带着科研，欢迎联系具身智能之心！我们提供从idea->实验->写作->投稿一站 式服务。 辅导方向：大模型、VLA、视觉语言导航、端到端、强化学习、Diffusion Policy、sim2real、具身交互、抓取 点预测与位姿估计、机器人决策规划、运动规划、3DGS、SLAM、触觉感知、双足/四足机器人、遥控操作、 零样本学习等方向，如果您有任意论文发表需求，支持带课题/ ...

点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近15个 方向 学习 路线 端到端自动驾驶 - 下一代智能驾驶量产核心算法 端到端自动驾驶（End-to-End Autonomous Driving）作为目前智驾量产的核心算法，可以分为一段式端到端、二段式端到端两个大的技术方向。自UniAD获得 CVPR Best Paper以来，正式拉开了国内新一轮的智驾军备竞赛。 2024年理想汽车更是宣布E2E+VLM的双系统架构量产！ 端到端自动驾驶通过传感器数据输入 （视觉/Lidar等）直接输出自车规划或控制信息，是目前智能驾驶最具代表性的方向。 目前VLM/VLA也是招聘的刚需，3-5年就能冲击百万年薪！ 而随着学术界和工业界的目光投向端到端这个技术领域，我们发现了很多问题。UniAD是端到端的最终解吗？显然不是！一系列算法如雨后春笋般冒出： 技术栈多？入门困难？ 去年我们推出了《首个面向工业级的端到端算法与实战教程》，今年很多小伙伴反馈技术发展太快了，先前的技术方案已经不适合当下的大环境。端到端目前发 展出多个领域技术的方向，需要掌握多模态大模型、BEV感知、强化学习、视觉Trans ...

Quantitative Models and Construction Methods 1. Model Name: IPCA Factor Model - **Model Construction Idea**: The IPCA factor model is designed to explain the returns of 46 option strategies, aiming to capture 80% of their returns while minimizing abnormal monthly returns to near zero[22] - **Model Construction Process**: The model integrates factors such as transaction costs and heterogeneous risk aversion to optimize derivative pricing. It also addresses the absence of reliable credit or liquidity premiums in pre-WWI corporate bond returns[25] - **Model Evaluation**: The model demonstrates strong explanatory power for option strategy returns and highlights the role of transaction costs in driving return volatility[22][25] 2. Model Name: Neural Functionally Generated Portfolios (NFGP) - **Model Construction Idea**: NFGP combines Transformer and diffusion models to enhance probabilistic time-series forecasting accuracy and improve decision reliability[35] - **Model Construction Process**: The model reduces forecasting errors by 42% compared to benchmarks and introduces dual uncertainty indicators to optimize portfolio decisions[35] - **Model Evaluation**: The model outperforms traditional approaches in terms of predictive accuracy and robustness in decision-making[35] --- Model Backtesting Results 1. IPCA Factor Model - **Explanatory Power**: 80% of option strategy returns explained[22] - **Abnormal Monthly Returns**: Approaching zero[22] 2. Neural Functionally Generated Portfolios (NFGP) - **Forecasting Error Reduction**: 42% compared to benchmarks[35] --- Quantitative Factors and Construction Methods 1. Factor Name: "Betting Against (Bad) Beta" (BABB) - **Factor Construction Idea**: The BABB factor improves the "Betting Against Beta" (BAB) strategy by managing transaction costs and isolating bad beta components[15] - **Factor Construction Process**: The factor is constructed using double sorting to isolate bad beta components. It achieves an annualized alpha exceeding 6%, independent of traditional sentiment indicators[15] - **Factor Evaluation**: The factor demonstrates strong performance in low-risk investment strategies, with significant alpha generation[15] 2. Factor Name: High-Speed Rail Network Centrality - **Factor Construction Idea**: This factor captures the impact of high-speed rail network centrality on corporate bond spreads by improving the information environment and regional trust[25] - **Factor Construction Process**: The factor is derived from the centrality of high-speed rail networks, showing a significant reduction in corporate bond spreads, particularly for non-state-owned enterprises and non-central cities[25] - **Factor Evaluation**: The factor effectively highlights the role of infrastructure in reducing financing costs and improving capital allocation efficiency[25] 3. Factor Name: Residual-Based Structural Change Detection - **Factor Construction Idea**: This factor robustly detects structural changes in factor models, accommodating over-specified factor numbers and error correlations[17] - **Factor Construction Process**: The factor employs residual-based tests to identify smooth or abrupt structural changes in factor models, enhancing robustness in model evaluation[17] - **Factor Evaluation**: The factor is highly effective in detecting structural changes and improving the robustness of factor model evaluations[17] --- Factor Backtesting Results 1. "Betting Against (Bad) Beta" (BABB) - **Annualized Alpha**: >6%[15] 2. High-Speed Rail Network Centrality - **Corporate Bond Spread Reduction**: Significant, especially for non-state-owned enterprises and non-central cities[25] 3. Residual-Based Structural Change Detection - **Robustness**: Effective in detecting both smooth and abrupt structural changes[17]

机器之心报道 编辑：笑寒、陈陈 这或许是庞若鸣（Ruoming Pang）在苹果参与的最后一篇论文。 庞若鸣 ，苹果基础模型团队负责人、杰出工程师，即将成为 Meta 新成立的超级智能团队的最新成员。他本科毕业于上海交通大学，在谷歌工作了 15 年，此后加 入苹果。另据彭博社最新消息，Meta 更是开出了 2 亿美金的天价来邀请庞若鸣加入。 虽然即将跨入另一段人生旅程，但庞若鸣还在为苹果站好最后一班岗。 7 月 9 日，庞若鸣在 X 上宣传了自己参与的一项研究《 AXLearn: Modular Large Model Training on Heterogeneous Infrastructure 》，据了解，这项研究是构建 Apple Foundation 模型的基础代码库。 具体而言，本文设计并实现了 AXLearn ，一个用于大规模深度学习模型训练的生产级系统，其具备良好的可扩展性和高性能。与其他先进的深度学习系统相比， AXLearn 具有独特的优势： 高度模块化和对异构硬件基础设施的全面支持 。 AXLearn 内部的软件组件接口遵循严格的封装原则，使得不同组件能够灵活组合，从而在异构计算环境中快 ...