In the current era, one of the key challenges in utilizing the Holt-Winters Method of forecasting is the accurate selection of smoothing coefficients. To address this issue, researchers have explored an optimization approach that aims to minimize forecasting errors, such as Mean Squared Errors (MSE) or Mean Absolute Deviation (MAD). This paper presents a novel methodology that employs a Genetic Algorithm (GA) to optimize the forecasting error by determining the optimal smoothing coefficients for the Holt-Winters Method. The objective value of the optimization problem is the Mean Square Error (MSE), which serves as a measure of the accuracy of the forecast. To evaluate the effectiveness of the proposed approach, actual test cases based on rice stock commercial commodity in the Philippines during the COVID-19 pandemic were utilized. The paper examines different variants of the Holt-Winters Method and assesses their suitability for capturing the characteristics of the rice stock data. The findings indicate that an additive seasonal effect is more appropriate for modeling the seasonal patterns observed in the rice stock data. Furthermore, the performance of the proposed GA-based approach is compared to other forecasting models to ascertain its efficacy. The results demonstrate promising outcomes, suggesting that the GA-based optimization approach for determining the smoothing coefficients in the Holt-Winters Method improves the accuracy of rice stock forecasting during the COVID-19 pandemic.