- Home
- Skills
- Research & Analysis
- Event-Driven ETF Study Builder
More screenshots
Works with the AI Agents you already use
Event-Driven ETF Study Builder
Turns "how did X stocks react to Y news?" into a reproducible study: a tiered concept basket, cap- and equal-weighted indices anchored pre-event, benchmark-relative results, dashboard and report.
$5
Event-Driven ETF Study Builder
Example session with this skill installed
New domestic GPU export controls were announced on 2026-09-14. How did Chinese AI-chip concept stocks react? Build me the study.
- Read your context and instructions
- Compiled the event-driven etf study
EVENT STUDY — built by the skill's protocol. Event pinned: export controls announced 2026-09-14 (market-local, before open) → event day 2026-09-14. Study spec frozen before results were inspected: universe from three concept listings snapshotted with dates, 23 constituents tiered T1 8 / T2 9 / T3 6 with a one-line exposure rationale each; window −5 to +10 trading days; weights from 2026-09-11 market caps; NAV = 100 pre-event; benchmark CSI 300; suspensions forward-filled and counted.
Results: T1 basket +18.3% cumulative over the window vs. CSI 300 +1.2% — an excess of roughly 17 points, so this is an event effect, not a market move. Equal-weighted full basket +9.7% against the cap-weighted figure: the move concentrated in the large core names rather than spreading evenly. Per-stock bars show 5 of the 8 T1 names driving 80% of the move. Event-day volume ran 3.4× the 20-day average. Pattern per the interpretation table: cap-weighted ≫ equal-weighted with T1 strongest — a real effect, concentrated.
Deliverables: runnable script with frozen parameters, price/NAV/summary data files plus reproducibility manifest, HTML dashboard (NAV curves, per-stock impact bars, weight breakdown, constituent table), and the written report.
Known limitations, disclosed in the report: two T1 names were suspended for one day (forward-filled); the concept list is current-vintage, so names delisted since the window are missing — survivorship bias noted, direction: likely understates the downside tail.
Connects securely to your tools. The creator never sees your data.
About this skill
Event questions about stocks usually get answered with vibes and a cherry-picked chart. This skill answers them with a reproducible artifact. You define the event precisely — what happened, the exact announcement date, and the first trading day it could move prices (with the after-close rule applied). You build the concept universe from snapshotted concept listings and tier every name by exposure: T1 core, T2 adjacent, T3 thematic — because T3 dilution washes out real T1 signals, and the T1-only basket is always reported alongside the full one. The index is constructed the honest way: weights from market cap on the trading day before the event (never after — that bakes the reaction into the weights), NAV anchored at 100, and both a cap-weighted version (what a real ETF would look like) and an equal-weighted twin (which reveals whether a move is broad or driven by giants). The analysis reports cumulative and excess returns per tier against the broad market over the identical window, per-constituent moves and concentration, drawdown path, and volume context — with an interpretation table for the classic patterns (cap ≫ equal means concentrated and real; basket ≈ benchmark means no event effect, whatever the headline return says). Deliverables are a runnable script with frozen parameters, standard data files with a reproducibility manifest, an interactive HTML dashboard (NAV curves, per-stock impact bars, weights, constituents), and a written report whose Assumptions and Known Limitations sections are mandatory. The skill is blunt about the pitfalls that invalidate studies: look-ahead weights, cherry-picked windows, missing benchmarks, and survivorship bias from applying today's concept list to yesterday's event.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.
- Results are only as good as the price data and the universe snapshot; a current-vintage concept list applied to a past event carries survivorship bias, which the report must disclose rather than eliminate.
- Attribution gets muddy when the benchmark itself moves on unrelated news inside the window; the method narrows the claim instead of pretending otherwise.
- This is historical measurement, not prediction: an event study describes how stocks reacted, and says nothing about what they will do next.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
Permissions
No special permissions declared or detected