stock-correlation

This skill analyzes stock correlations to find related companies and trading pairs, using historical price data from Yahoo Finance via the yfinance library. It solves the problem of discovering which stocks move together (or inversely) — sector peers, sympathy plays, supply-chain partners, and hedging pairs — and quantifying those relationships. The documentation is explicit that it is for research and educational purposes only and is not financial advice.

The skill first checks whether yfinance, pandas, and numpy are available and installs them via pip only if missing, then routes the request to one of four sub-skills. Co-movement Discovery takes a single ticker, dynamically builds a 15-30 candidate peer universe using yfinance screeners (`yf.screen()` with `yf.EquityQuery`, broadened by sector and adjacent industries rather than hardcoded lists), computes daily log-return correlations, and presents a ranked table with company names, sectors, and a plain-English rationale for each link. Return Correlation deep-dives two or more tickers, reporting Pearson correlation, beta, R-squared, rolling 60-day correlation statistics, and a z-scored log-price spread for mean-reversion. Sector Clustering groups a set of tickers into a correlation matrix or clusters, and Realized Correlation handles time-varying and conditional (for example downside) correlation. Sensible defaults are documented: a one-year daily lookback, Pearson method, a 0.60 minimum correlation threshold, top-10 results, log returns, and a 60-day rolling window.

Target users are retail and quantitative investors, analysts, and researchers exploring co-movement, pair trading, hedging, and sector structure. All data access is read-only market data; the only system action is installing standard Python data-analysis packages when they are absent.