Trade Filings
管理人 股票 申报文件 超级投资者
信号 机构买入 机构卖出 内部人活动 13D & 13G 监控 板块轮动
工具 股票筛选器 管理人筛选器 对比管理人 共识篮子 所有研究工具 数据 API
学习与 AI 学习 数据提问 AI 智能体
新闻 ★ 已保存
关于 关于我们 方法论 联系我们 免责声明
阅读器
数据方案

历史数据下载——随账户功能上线。

API 网关

对本站缓存数据提供免费只读JSON访问。

深色模式

🧭 引导视图
初次接触市场——对价格、收益率、年初至今涨幅、市值等概念感到陌生?我们在您浏览时以浅显易懂的方式解释每个术语,数据不变,帮助随时在线。

⚡ 专家视角
您已了解市场。只看数据——干净、快速、简洁,无多余解释。这是默认视图。

界面语言
方法论 · 变动引擎

Change classification & persistence

How NEW / INCREASED / REDUCED / EXITED and the streak measures are computed.


Changes are computed per manager, per quarter, on the common-equity grain: rows with share-type SH and no put/call flag, aggregated per (manager, period, ticker). Option positions and unmapped CUSIPs are excluded from change analytics (they remain visible in the as-reported holdings tables).

Portfolio weight = position value ÷ the manager's reported common-equity value that quarter.

q_held = consecutive quarters the position has been reported within our data window (window: three quarters for tracked managers, nine for curated managers — displayed as “N+” at the window edge, since the true streak may extend further back).

streak = consecutive quarters of share increases (+N) or reductions (−N) for the manager-security pair.

Stock splits. Where a stock split falls between two reporting periods, quarter-over-quarter classifications and share changes are computed against the split-adjusted prior share count (a pure split reads UNCHANGED, never “+400%”); the originally reported figures are always preserved and shown, and affected rows carry an “adjusted for split” note. Split data: Financial Modeling Prep splits calendar.

局限性

Share counts can change for non-trading reasons (corporate actions, in-kind transfers, reporting-entity changes). Classifications describe reported share counts, not confirmed open-market trades.


← 所有方法论页面 数据来源