etf-premium

The etf-premium skill calculates the premium or discount of an ETF's market price relative to its Net Asset Value (NAV) using data from Yahoo Finance via the yfinance library, and goes further by decomposing single-day price surges into NAV-driven versus structural components such as gamma squeeze, dealer hedging, and blocked authorized-participant arbitrage. It solves the problem of understanding when an ETF's traded price has diverged from the value of its underlying holdings — a divergence that is tiny for liquid US equity ETFs but can be significant for bond, international, leveraged/inverse, commodity, and crypto ETFs, especially under market stress.

The skill first checks for and installs its Python dependencies (yfinance, pandas, numpy), then routes the user's request to one of several sub-skills: a single-ETF snapshot, a multi-ETF comparison, a premium screener for finding extreme discounts, a deep dive with context, or a premium-surge decomposition that estimates how much of a move is dealer-hedging-driven. Its reference material explains the creation/redemption arbitrage mechanism and why it breaks down, documents the key yfinance data fields and their limitations (no historical NAV, end-of-day NAV timing, delayed prices), and provides category-specific 'normal' premium/discount benchmark ranges, stress-scenario examples, curated ETF universes for screening, peer-comparison groups, and a bid-ask-spread reality check.

The target users are retail and professional investors, analysts, and quant-curious traders who want to evaluate ETF pricing efficiency or investigate why an ETF jumped. The skill is explicitly framed as for research and educational purposes only and not financial advice. It performs read-only market-data fetches and computation; the only notable behavior is that it may run pip to install its data dependencies, which is a standard, benign setup step.