Comparative analysis of clustering methods in the evaluation of home advantage in handball: Implications for sport performance

Abstract

This study aimed to compare four team classification methods—hierarchical clustering (HC), final ranking (FR), quartiles (Q), and median split (MS)—to analyze the Home Advantage (HA) in handball. The objective was to evaluate the effectiveness of each method in segmenting teams based on performance level, with particular attention to sex-based differences. Data were collected from the first divisions of the Spanish men's and women's handball leagues over ten seasons (2013–2024), excluding the suspended season due to COVID-19. HA was calculated as the percentage of points earned at home out of total season points. Teams were grouped into high-level teams (HLT), medium-level teams (MLT), and low-level teams (LLT) according to their win percentage using each of the four methods. Each classification method presented distinct strengths and limitations. HC allowed flexible, data-driven grouping and revealed subtle differences between teams, but was sensitive to sample variation. FR, commonly used in sports performance analysis, offered intuitive understanding but lacked statistical granularity. Q provided a detailed segmentation suitable for heterogeneous leagues but sometimes obscured the practical interpretation. MS was simple and efficient, but it risked oversimplifying performance differences. Results showed that HC and Q were more effective at identifying performance patterns, especially when analyzing differences by sex. However, no single method proved to be universally superior. The choice of grouping strategy should be aligned with the specific objectives of the analysis. These findings provide methodological guidance for researchers and practitioners in sport science, with implications for tactical planning, talent identification, and competition analysis in team sports.