Description: Reserving and ratemaking calculations are vital to how insurance companies perform financially. Current-day practices for predicting future loss ratio or pure premiums for reserving and ratemaking rates are not based on low prediction errors—mean squared error (MSE) criterion. This report provides mathematically optimal weights that lead to minimum MSE prediction errors, and the formulas are validated using retro-testing with real-world datasets. In this report, two main theorems are discussed, each addressing prediction of future loss ratio or pure premium and improving upon current practices for how rates and reserves are calculated.