AI Investment Frontier AI Investment Frontier — Causal Separation Makes Portfolio AI Harder to Fake A new arXiv paper argues that portfolio AI should separate declared drivers from residual risk, turning Markowitz into a projected problem with clearer estimation and causal tests.
AI Investment Frontier AI 投资前沿：模型真正赢在少交易 本周两篇新论文说明，投资 AI 的边界正在从“更准的预测”转向“更稳的策略、更低的换手率、以及更像真实市场的状态估计”。
AI Investment Frontier AI Models Win by Trading Less, Not Just Forecasting Better Two new July papers suggest the edge in investment AI is shifting from better predictions to better policies, lower turnover, and cleaner state estimation.
AI Investment Frontier End-to-End Portfolio Policies Are Getting Real A new futures-timing paper shows when end-to-end AI policies beat simple rules, and why transaction costs still decide whether the edge survives.
AI Investment Frontier 基金数据正在变成投资 AI 的新人格层 Fund2Persona 提示我们：真正有用的投资 AI 不是会模仿语气，而是能被基金披露、持仓变化和经理评语约束住。
AI Investment Frontier Fund Data Is Becoming the New Persona Layer for Investment AI A new Fund2Persona paper shows how fund disclosures, holdings transitions, and manager commentary can ground investment personas instead of relying on generic prompts.
AI Investment Frontier ML Term-Structure Forecasts Should Feed Duration Policy A June 2026 preprint shows neural term-structure models can improve bond-curve forecasts, but the real value is in duration control and portfolio policy.
AI Investment Frontier 时间序列基础模型是先验，不是阿尔法引擎 一篇新的 arXiv 基准研究提醒投资 AI 团队：预训练时间序列基础模型能降低建模成本，但相对随机游走的优势仍然稀疏。
AI Investment Frontier Time-Series Foundation Models Are Priors, Not Alpha Engines A new arXiv benchmark finds pretrained time-series foundation models can reduce modeling work in return forecasting, but their gains over random walk remain sparse.
AI Investment Frontier 分位数任务需要组合策略，而不是一个预测 一篇 6 月修订的组合选择论文提醒投资 AI：下行保护、收益和成长任务，应当训练并评估不同的策略，而不是共用一个平均收益预测。
AI Investment Frontier Quantile Mandates Need Portfolio Policies, Not One Forecast A revised portfolio-choice paper shows why downside protection, income, and growth mandates should train different AI policies instead of sharing one return forecast.
AI Investment Frontier LLM Factor Forecasts Need Decision-Time Inputs A new leakage-aware benchmark shows why investment LLMs must separate forecast skill from information timing before entering factor allocation workflows.
AI Investment Frontier 金融 AI 需要确定性的生产内核 一篇关于 Mojo 与资本市场 AI 的论文提醒投资团队：速度只是生产化的一半，真正进入投研、风控和审计流程的 AI，还需要可复现的计算内核。
AI Investment Frontier Financial AI Needs Deterministic Production Kernels A new arXiv paper on Mojo for capital markets argues that speed is only half the production problem. Investment AI also needs reproducible kernels that survive audit, backtesting, and model handoff.
AI Investment Frontier 基本面研究智能体需要隔离式架构 FundaPod 论文提出：机构基本面研究不适合共识驱动型多智能体架构，而应让不同投资风格代理人独立推理、产出来源可追溯、分歧由基金经理通过知识图谱裁决。
AI Investment Frontier Fundamental Research Needs Independence-Preserving Agent Architectures FundaPod shows why institutional investment research needs isolated personas, knowledge-graph memory, and post-hoc disagreement surfacing rather than consensus-seeking agent teams.
AI Investment Frontier AI Exposure Needs a Cross-Asset Risk Map Apollo’s latest 60/40 warning turns AI from an equity theme into a cross-asset risk measurement problem for portfolio builders.
AI Investment Frontier 投资智能体需要信念状态验证 一篇新的 arXiv 论文把智能体式投资 AI 放进 POMDP 框架，提醒团队不要只看预测准确率，而要分别验证信念、预测、动作和效用。
AI Investment Frontier Agentic Investment AI Needs Belief-State Validation A new arXiv paper frames agentic AI as a partially observable decision process, giving investment teams a better validation target than forecast accuracy alone.
AI Investment Frontier AI 研究员需要投资经理接口 一家亚洲对冲基金把 AI 人才嵌入投资经理团队，提示投资 AI 的下一步不是更会聊天的模型，而是可审计、可复盘、能被 PM 接管的研究工作流。
AI Investment Frontier AI Research Fellows Need a Portfolio-Manager Interface A new hedge-fund AI fellowship points to a practical frontier: investment AI works best when builders are embedded beside portfolio managers, with evidence, controls, and workflow ownership.
AI Investment Frontier 执行模型需要“时间意外”遥测 一篇 2026 年 6 月的 arXiv 论文提醒投资 AI 构建者：执行系统不能只等滑点统计显著后才反应，而应记录成交后不利价格事件是否过快出现。
