Pick a date range and click ▶ RUN. We'll fetch grouped daily closes from Polygon for the start and end of the window, compute each constituent's return, weight by current market cap, and group by GICS sector. The result is a Brinson-style attribution: which sectors drove the index's return over the window, and by how many percentage points.

Pick an anchor (e.g. NVDA), a correlation lookback (trading days), and a hypothetical shock%. This ranks thematic subsectors by their correlation to the anchor, picks each subsector's leader (the member most correlated to the anchor), then propagates the shock two hops — anchor → leader → the rest of each subsector — via correlation-scaled betas (β = ρ·σ_y/σ_x). Each name shows its correlation, R² (confidence), and predicted move. The STABILITY window flags leaders whose correlation drifts between the two lookbacks.