Math anxiety and academic distress, two interrelated forms of psychological stress, are pervasive problems for undergraduate mathematics students. Most of the research in this area has taken a broad view of the impact this stress has on students' learning across an entire course or, more broadly, the entire curriculum for their degree. To complement this existing body of research, this paper serves a dual purpose: First, to provide insight into undergraduate students' day-to-day classroom stress, and secondly, to explore fitness trackers as a tool for in situ detection of student anxiety or stress in the classroom. To accomplish the first objective, we conduct a phenomenographic analysis to explore the variation in 29 students' reported experiences of psychological stress within a session on roots of polynomial functions. We identified 7 categories of description. For the second objective, we recorded the students' heart rate during the same lesson. We clustered students heart rate variability plots according to our categories of description and employed a logistic regression model to estimate the probability that a student will report experiencing psychological stress given their heart rate variability. Our results show that fitness trackers can produce measures that serve as a predictor of self-reported emotional change.