Stock Correlation AnalysisFinance Skill

A stock correlation analysis skill using yfinance, pandas, and numpy. 3-step process: (1) ensure dependencies (auto-install yfinance/pandas/numpy via pip if missing); (2) route to correct sub-skill based on user intent; (3) present results with caveats. Four sub-skills: A — Co-movement Discovery: build dynamic peer universe (15-30 candidates) using yf.screen() + yf.EquityQuery (same-industry stocks, broaden to sector, add thematic/adjacent industries), compute daily log returns, Pearson correlation, rank by absolute correlation, present top 10 + notable negative correlations; B — Return Correlation: pairwise analysis with Pearson correlation, beta (cov/var), R-squared, 60-day rolling correlation (mean/std/min/max), spread Z-score for mean-reversion, interpretation guide (>0.80 strong, 0.50-0.80 moderate, <0.50 weak, spread Z > |2| unusual divergence); C — Sector Clustering: NxN correlation matrix, hierarchical clustering via scipy (Ward linkage, falls back to sorting if scipy unavailable), identify clusters/outliers/strongest-weakest pairs; D — Realized Correlation: rolling windows (20/60/120-day), regime-conditional correlation (all days, up days, down days, high vol top 25%, low vol bottom 25%, large drawdown < -2%), highlight correlation increases during sell-offs. Defaults: 1y lookback, daily interval, Pearson method, 0.60 min threshold, top 10 results, 60-day rolling window. Includes reference file for dynamic peer universe construction using yfinance Screener API. Uses yfinance (required), pandas, numpy, scipy (optional). Python 3.8+. Works on all platforms.

Snapshot quality

88/100

Snapshot risk

Medium

Audit conclusion

Needs review