Use synthetic fixtures instead of article seed

Author: Pigbibi

README.md

@@ -4,9 +4,9 @@
 
 Research-only political event and public disclosure tracking for US equities.
 
-This repository asks whether the event-tracking effect described in the Longbridge article
-["国会山股神？戴尔暴涨 38% 的背后，是基本面还是……"](https://longbridge.com/zh-CN/topics/41260998.md)
-can be tracked in a repeatable, point-in-time way.
+This repository asks whether public disclosure, official remarks, policy capital,
+procurement, and other political/public events can be tracked in a repeatable,
+point-in-time way.
 
 ## Repository Role
 
@@ -24,9 +24,8 @@ The initial research scope is:
 
 ## Current Status
 
-The first committed dataset is a seed extracted from the Longbridge topic above.
-It is intentionally small and should be treated as a hypothesis registry, not as
-verified investment evidence.
+The committed CSV files under `examples/` are synthetic schema fixtures only.
+They are not investment evidence and are not derived from any article.
 
 Tracked event families:
 
@@ -43,16 +42,16 @@ Build the seed tracker:
 
 ```bash
 python scripts/build_tracker.py \
-  --watchlist data/seed/article_41260998_watchlist.csv \
-  --events data/seed/article_41260998_events.csv \
-  --output data/output/article_41260998_tracker.csv
+  --watchlist examples/political_watchlist.example.csv \
+  --events examples/political_events.example.csv \
+  --output data/output/political_tracker.example.csv
 ```
 
 Run the synthetic event study:
 
 ```bash
 python scripts/run_event_study.py \
-  --events data/seed/article_41260998_events.csv \
+  --events examples/political_events.example.csv \
   --prices examples/price_history.example.csv \
   --windows 1,2 \
   --output data/output/event_study.example.csv
@@ -75,7 +74,7 @@ event_id,event_date,symbol,event_type,direction,confidence,source_url,notes
 Watchlist input schema:
 
 ```text
-symbol,name,bucket,article_status,thesis,source_url
+symbol,name,bucket,research_status,thesis,source_url
 ```
 
 Price input schema:
@@ -106,7 +105,7 @@ This repo does not own:
 
 ## Next Work
 
-1. Replace article-derived seed events with official-source extraction.
+1. Add official-source extraction.
 2. Add a source adapter for OGE disclosure PDFs or normalized public datasets.
 3. Add public-remarks ingestion from White House pages and social-media exports.
 4. Backfill enough point-in-time events to evaluate hit rate, lag, and false

README.zh-CN.md

@@ -4,8 +4,6 @@
 
 QuantStrategyLab 的确定性研究仓库，用来验证“公开持仓/交易披露 + 公开讲话/社媒点名 + 政策资金事件”能否形成可追踪的美股事件线索。
 
-首个样本来自长桥文章：[《国会山股神？戴尔暴涨 38% 的背后，是基本面还是……》](https://longbridge.com/zh-CN/topics/41260998.md)。
-
 ## 仓库定位
 
 这是研究证据仓库，不是 AI 仓库，也不是交易执行仓库。
@@ -28,7 +26,7 @@ QuantStrategyLab 的确定性研究仓库，用来验证“公开持仓/交易
 
 ## 当前状态
 
-当前只提交了一份文章种子样本。它用于定义研究问题和跑通工具链，不应视为已经验证过的投资证据。
+当前提交的 `examples/` 数据是完全合成的 schema fixture，只用于跑通工具链，不是投资证据，也不是从任何文章抽取出来的样本。
 
 事件类型：
 
@@ -39,20 +37,20 @@ QuantStrategyLab 的确定性研究仓库，用来验证“公开持仓/交易
 
 ## 本地验证
 
-生成文章种子追踪表：
+生成合成示例追踪表：
 
 ```bash
 python scripts/build_tracker.py \
-  --watchlist data/seed/article_41260998_watchlist.csv \
-  --events data/seed/article_41260998_events.csv \
-  --output data/output/article_41260998_tracker.csv
+  --watchlist examples/political_watchlist.example.csv \
+  --events examples/political_events.example.csv \
+  --output data/output/political_tracker.example.csv
 ```
 
 用合成价格样本跑事件研究：
 
 ```bash
 python scripts/run_event_study.py \
-  --events data/seed/article_41260998_events.csv \
+  --events examples/political_events.example.csv \
   --prices examples/price_history.example.csv \
   --windows 1,2 \
   --output data/output/event_study.example.csv
@@ -66,10 +64,10 @@ python -m pytest -q
 
 ## 研究判断
 
-文章里的“追踪效果”可以拆成三个可验证问题：
+这类“追踪效果”可以拆成三个可验证问题：
 
 1. **能不能第一时间知道谁进入观察池**：需要结构化公开披露和政策/持仓来源。
 2. **能不能捕捉公开点名**：需要按时间记录白宫讲话、采访、社媒和新闻文本。
-3. **点名后是否有可交易的统计优势**：需要事件研究和样本外验证，不能只看 DELL、MU 等少数案例。
+3. **点名后是否有可交易的统计优势**：需要事件研究和样本外验证，不能只看少数轶事案例。
 
 本仓库先解决前两步的数据结构和复盘框架；第三步需要更多点位和真实行情输入。后续如果需要 LLM 处理长文本，只能作为可替换的抽取工具，不能把模型判断结果写成核心信号合同。

data/seed/article_41260998_events.csv

@@ -4,7 +4,7 @@
 
 验证公开人物持仓披露、公开点名、政策资金事件与美股短中期收益之间是否存在可复现关系。
 
-首个样本来自长桥文章 `41260998`，其中主要假设是：
+这类研究思路的主要假设是：
 
 - 先出现于公开持仓或交易披露
 - 再出现于公开讲话、采访、白宫材料或社媒文本
@@ -25,7 +25,7 @@
 
 ## 事件分层
 
-- `watchlist_only`：只在公开持仓或文章观察池中出现，尚未触发点名
+- `watchlist_only`：只在公开持仓或研究观察池中出现，尚未触发点名
 - `disclosure_buy`：公开披露中出现买入或新增持仓
 - `public_mention`：公开讲话、采访、社媒、白宫材料中被点名
 - `policy_capital`：政策资金、政府采购、入股、产业政策支持
@@ -37,13 +37,13 @@
 2. 官方讲话：WhiteHouse.gov、官方转录、正式新闻稿。
 3. 可审计社媒：Truth Social、X 的可复核 URL 或导出。
 4. 市场数据：Longbridge、Yahoo、Polygon、Nasdaq Data Link 或 QuantStrategyLab 已有价格输入。
-5. 二级媒体和社区文章：只做线索，不直接作为高置信事件。
+5. 二级媒体和社区文章：只做灵感或线索，不直接作为事件证据入库。
 
 ## 初版成功标准
 
 初版不判断能否实盘，只判断是否具备研究可行性：
 
-- 能从同一 schema 载入文章种子事件。
+- 能从同一 schema 载入合成示例和官方来源事件。
 - 能生成候选追踪表，区分已触发点名和仅观察池标的。
 - 能在本地价格 CSV 上计算事件后收益。
 - 能记录数据来源、置信度和未验证事项。
@@ -60,9 +60,9 @@
 
 ## 当前限制
 
-- 长桥文章是线索来源，不是最终数据源。
+- 社区文章只能作为灵感来源，不是默认数据源。
 - 公开披露通常有滞后，无法假设“披露日”等于“买入日可交易发现日”。
-- 文章中的部分时间线需要官方材料逐条复核。
+- 非官方材料中的时间线需要官方材料逐条复核后才能入库。
 - 单个总统或单一产业周期可能产生强样本偏差。
 
 ## 与 AI 影子信号仓库的边界

data/seed/article_41260998_watchlist.csv

@@ -4,9 +4,11 @@ This file records source families for future ingestion. A source listed here is
 not automatically trusted; every event still carries its own `source_url` and
 `confidence`.
 
-## Seed Source
+## Inspiration Sources
 
-- Longbridge topic 41260998:
+- The Longbridge topic below inspired the initial research question. It is not
+  committed as an event seed and should not be treated as evidence without
+  official-source verification:
   <https://longbridge.com/zh-CN/topics/41260998.md>
 
 ## Candidate Official Sources
@@ -28,4 +30,3 @@ not automatically trusted; every event still carries its own `source_url` and
 - `medium`: reputable secondary source with enough details to locate official
   source later.
 - `low`: article, community post, or manually transcribed lead pending review.
-

docs/research_plan.zh-CN.md

@@ -0,0 +1,6 @@
+event_id,event_date,symbol,event_type,direction,confidence,source_url,notes
+demo-evt1-disclosure-buy,2026-01-10,EVT1,disclosure_buy,bullish,medium,https://example.invalid/official-filing/evt1-2026-01-10,"Synthetic official filing event for schema validation."
+demo-evt1-public-mention,2026-01-20,EVT1,public_mention,bullish,medium,https://example.invalid/official-remarks/evt1-2026-01-20,"Synthetic public mention event for schema validation."
+demo-evt2-policy-capital,2026-02-05,EVT2,policy_capital,bullish,high,https://example.invalid/policy-capital/evt2-2026-02-05,"Synthetic policy capital event for schema validation."
+demo-evt3-market-reaction,2026-03-15,EVT3,market_reaction,bullish,medium,https://example.invalid/issuer-news/evt3-2026-03-15,"Synthetic market reaction marker for schema validation."
+

docs/source_registry.md

@@ -0,0 +1,6 @@
+symbol,name,bucket,research_status,thesis,source_url
+EVT1,Example Catalyst One,named_mentioned,triggered,"Synthetic disclosure plus mention candidate used only to validate tracker output.",https://example.invalid/watchlist/evt1
+EVT2,Example Policy Two,policy_capital,triggered,"Synthetic policy capital candidate used only to validate tracker output.",https://example.invalid/watchlist/evt2
+EVT3,Example Quality Three,disclosed_holding,watchlist,"Synthetic quality review candidate used only to validate tracker output.",https://example.invalid/watchlist/evt3
+EVT4,Example Monitor Four,drone_policy_watchlist,watchlist,"Synthetic watch-only candidate used only to validate tracker output.",https://example.invalid/watchlist/evt4
+

examples/political_events.example.csv

@@ -1,41 +1,23 @@
 date,symbol,close
-2026-02-10,DELL,100.00
-2026-02-11,DELL,101.00
-2026-02-17,DELL,105.00
-2026-03-10,AAPL,180.00
-2026-03-11,AAPL,181.00
-2026-03-12,AAPL,183.00
-2026-03-18,AAPL,188.00
-2026-03-10,TMO,520.00
-2026-03-11,TMO,522.00
-2026-03-12,TMO,525.00
-2026-03-18,TMO,535.00
-2026-03-25,MU,95.00
-2026-03-26,MU,98.00
-2026-04-01,MU,104.00
-2026-05-08,DELL,111.00
-2026-05-11,DELL,115.00
-2026-05-15,DELL,123.00
-2026-05-22,MU,110.00
-2026-05-26,MU,116.00
-2026-05-29,MU,120.00
-2026-05-28,DELL,130.00
-2026-05-29,DELL,179.40
-2026-06-04,DELL,182.00
-2026-02-10,SPY,510.00
-2026-02-11,SPY,511.00
-2026-02-17,SPY,515.00
-2026-03-11,SPY,520.00
-2026-03-12,SPY,521.00
-2026-03-18,SPY,525.00
-2026-03-25,SPY,530.00
-2026-03-26,SPY,531.00
-2026-04-01,SPY,534.00
-2026-05-08,SPY,545.00
-2026-05-11,SPY,546.00
-2026-05-15,SPY,548.00
-2026-05-22,SPY,550.00
-2026-05-26,SPY,552.00
-2026-05-28,SPY,553.00
-2026-05-29,SPY,554.00
-2026-06-04,SPY,556.00
+2026-01-10,EVT1,100.00
+2026-01-12,EVT1,101.00
+2026-01-20,EVT1,104.00
+2026-01-21,EVT1,107.00
+2026-01-27,EVT1,111.00
+2026-02-05,EVT2,50.00
+2026-02-06,EVT2,52.00
+2026-02-12,EVT2,54.00
+2026-03-15,EVT3,80.00
+2026-03-16,EVT3,84.00
+2026-03-23,EVT3,83.00
+2026-01-10,SPY,500.00
+2026-01-12,SPY,501.00
+2026-01-20,SPY,505.00
+2026-01-21,SPY,506.00
+2026-01-27,SPY,508.00
+2026-02-05,SPY,510.00
+2026-02-06,SPY,511.00
+2026-02-12,SPY,514.00
+2026-03-15,SPY,520.00
+2026-03-16,SPY,521.00
+2026-03-23,SPY,522.00

examples/political_watchlist.example.csv

@@ -36,7 +36,7 @@ class WatchlistItem:
     symbol: str
     name: str
     bucket: str
-    article_status: str
+    research_status: str
     thesis: str
     source_url: str
 
@@ -49,7 +49,7 @@ def load_watchlist(path: str | Path) -> list[WatchlistItem]:
                 symbol=row["symbol"].upper(),
                 name=row.get("name", ""),
                 bucket=row.get("bucket", ""),
-                article_status=row.get("article_status", ""),
+                research_status=row.get("research_status") or row.get("article_status", ""),
                 thesis=row.get("thesis", ""),
                 source_url=row.get("source_url", ""),
             )
@@ -81,7 +81,7 @@ def trigger_status(item: WatchlistItem, events: list[Event]) -> str:
         return "mentioned"
     if "disclosure_buy" in event_types:
         return "disclosed"
-    return item.article_status or "watchlist"
+    return item.research_status or "watchlist"
 
 
 def build_tracker_rows(items: list[WatchlistItem], events: list[Event]) -> list[dict[str, object]]:
@@ -152,4 +152,3 @@ def main(argv: list[str] | None = None) -> None:
 
 if __name__ == "__main__":
     main()
-