The war in Ukraine and the global crisis caused by the coronavirus pandemic highlighted the importance of real-time assessments of the macroeconomic situation as a basis for conducting appropriate stabilization policies. However, this is a challenging task, with a major complicating factor being the lagged publication of key economic variables. In order to facilitate economic decision-makers, there has been an increased focus on producing faster statistics that better reflect the current state of the economy in order to improve the decision-making basis for stabilization policy. An important method for this is current forecasting of key variables such as gross domestic product (GDP), which is currently carried out regularly by various forecasting institutes such as the National Institute of Economic Research. Our contribution to the growing interest in baseline forecasting is to incorporate methods from the machine learning literature. These have advantages over previously used methods because they can handle large data sets and model non-linear relationships between economic variables. This project aims to evaluate and further develop these methods. The evaluation is carried out through simulation methods and applications where Swedish GDP is forecast at present. The aim is to improve the NIER's current GDP forecasting methods. Our focus is thus on a small open economy, which will characterize the design of the simulation study.