event-etf-study

IMPORTANT: Output-Language Lock

• The final conversation reply and every deliverable (dashboard / charts / tables / custom_html) must follow the language of the user's latest query, not the market
• If the prompt is in English and the symbols are China / Hong Kong stocks, both the reply and the deliverables must stay in English; stock references should default to ticker code such as 600519.SH / 0700.HK
• If the prompt is in Chinese, both the reply and the deliverables must stay in Chinese; when a Chinese stock name is known, prefer the Chinese name
• Do not make this mistake: the HTML is in English but the actual conversation reply switches back to Chinese
• If the English stock name is uncertain, use the ticker code instead of a Chinese stock name

Event Study ETF

Workflow

1. Read the pitfalls: read references/common_pitfalls.md in full, then self-check against the checklist at the end before delivery.
2. Freeze reproducibility metadata: hard-code query, language, event_date_source, generated_at, price_adjustment, market, data_source, and constituent_snapshot in the code configuration block. Resolve language to a concrete "zh" or "en" string from the query text (CJK detection) before hard-coding it. Do not let reruns of the same study update these values automatically.
3. Identify concept stocks: search concept stocks across Tonghuashun (10jqka), Xueqiu, and East Money -> save a source snapshot CSV -> take the union as constituent candidates -> validate with mshtools/ifind -> assign T1/T2/T3 tiers by relevance. See references/concept_research.md for methodology.
4. Fetch data: use MCP ifind to fetch forward-adjusted daily prices plus total shares -> save raw returns/previews under raw/ -> compute daily market cap.
   • Set the window length exactly to the user's request: if the user asks for "buy after the event and hold for one week", use 3-5 trading days before the event plus 1-2 weeks after the event (about 10-15 trading days).
   • General rule: start_date = 3-5 trading days before the reference date; end_date = 2-3 trading days after the user's focus window.
5. Build the ETF: use market cap on the pre-event reference date to calculate weights, then generate both market-cap-weighted NAV and equal-weighted NAV.
6. Export standard files: call references/export_event_results.py to produce 3 standard data files plus 1 reproducibility manifest. Always pass market ("china_a" or "us") and generated_at.
7. Generate the dashboard: call references/render_event_dashboard.py to read the standard files and produce an HTML dashboard. Use assets/dashboard_template.html as the shell template. See "Dashboard Chart Selection" below for choosing modules.
8. Static charts: use Matplotlib to generate standalone PNG files in the cwd.
9. Report: write report.md; it must include ## Assumptions and ## Known Limitations.
10. Self-check: trial run -> 4 standard files written -> run references/validate_event_outputs.py -> reconcile numbers -> complete the pitfalls checklist.
11. Deliver: runnable code + 4 standard files + report.md + PNG files + HTML dashboard.