This project studies correlation model risk in basket option pricing and hedging. I compare a constant-correlation multi-asset Black-Scholes model with a calm/stress regime-switching correlation model for a European basket call. In the synthetic setting, time-0 pricing differences are usually modest, but hedging differences are more meaningful, especially in downside P&L tails under harsher stress scenarios and for larger baskets. I then extend the setup so the regime is hidden from the hedger and must be filtered from past returns. The filtered hedge does not match the oracle regime hedge, but it still preserves a meaningful share of the oracle improvement in severe downside outcomes.