This study develops an early forecasting framework for Supplemental Coverage Option (SCO) and Enhanced Coverage Option (ECO) crop insurance payouts using preliminary county-level loss information. Using data from 2011–2025 for non-irrigated corn in Kansas, the analysis combines USDA RMA yield and price information, claims-related variables, and PRISM weather data. Four modeling approaches, linear regression, nonlinear regression, XGBoost, and Random Forest, are evaluated using out-of-sample validation, and predicted county yields are translated into SCO/ECO trigger predictions. The Random Forest model performs best, with an out-of-sample R-Squared of 0.88, RMSE of 14.24, and MAE of 10.83, and correctly identifies approximately 89% of actual trigger events. The results demonstrate the potential for preliminary loss data to provide earlier information on likely SCO/ECO payouts, supporting financial planning by producers, insurers, and agricultural lenders.
