Predicting the onset of acute performance decline in esports

Abstract

Video game playing and esports have witnessed remarkable growth and popularity on a global scale, becoming one of the fastest-growing forms of leisure and competitive activity. However, despite its exponential rise, there is still a notable lack of systematic knowledge regarding the impacts of prolonged gameplay on human performance, physiology, and cognition. Within the dynamic realm of esports, two pivotal player states, frequently debated for their profound impact on performance outcomes, stand out: tilt and flow. Tilt refers to a state of emotional distress that impairs performance, while flow indicates a state of complete immersion and peak performance. This study focuses on detecting and predicting tilt using machine learning models trained on physiological, cognitive, and behavioral data collected from 45 players in prolonged gaming sessions across League of Legends, Call of Duty, and Valorant. The model identifies distinct response patterns preceding tilt, enabling early detection to improve game performance. Understanding the triggers and manifestations of tilt is essential for optimizing player performance and well-being. This research represents a significant step in real-time esports performance monitoring, using simple devices and software to create interventions that enhance esports players' and teams’ performance.