Forecasting is the process of estimating future events based on past information. In this study, the Support Vector Regression (SVR) method with the grid search time series cross-validation algorithm was used to analyze time series data. SVR is an extension of Support Vector Machine (SVM) for regression. This research aims to obtain the best model for predicting and forecasting the daily stock time series data of DSS company in Indonesia. The study compares four types of kernels—linear, polynomial, RBF, and sigmoid—to determine the best model. Model accuracy evaluation was conducted using RMSE, MSE, MAPE, and R-squared, where the model with the lowest error value was considered the best. The results show that SVR with a linear kernel, parameter C = 100, and epsilon = 0.01 produced an RMSE of 0.0583, MSE of 0.0034, MAPE of 10.53%, and R-squared of 0.99. Based on the MAPE value, this model is considered suitable for forecasting DSS stock, showing a downward trend in predictions