Even in a warming world, extreme winter weather events like cold air outbreaks still occur and have high socioeconomic costs, but they remain a challenge to forecast skillfully in the subseasonal-to-seasonal timeframe (i.e., 2-8 weeks). In this talk, I will discuss recent work by the Applied Climate Dynamics Group at the University of Oklahoma to improve our dynamical understanding of the evolution of cold air outbreaks in the central United States and then use this knowledge to improve skill in subseasonal predictions of those events. We use a combination of observations and numerical modeling experiments to show why we should care about intraseasonal variability in the stratosphere to improve our forecast skill, including considering different types of stratosphere-troposphere coupling outside of the “classical” paradigm. Additionally, we will highlight particular precursor flow patterns in the upper troposphere (so-called blocking patterns) which were key for skillful 3+ week forecasts of the February 2021 North American Cold Air Outbreak and could be exploited for long-lead predictions of future cold air outbreaks.