Cluster Analysis of High-Pressing Types in K League 1 and 2 and Tactical Changes by Match Situation

This study profiles high pressing in the 2023 K League 1 and 2 using PPDA and compares tactical adjustments by match status (win, draw, loss). Team-season averages of status-specific PPDA (three variables) were z-standardized and clustered via K-means. Model selection combined Elbow, Silhouette, and Calinski–Harabasz (k=3 supported; K1 Silhouette=0.362; K2=0.375). Seed sweeps with a minimum cluster size ≥2 ensured solution stability. Three profiles emerged: (i)high-press maintenance, (ii)situational response (draw-phase relaxation), and (iii)conservative/relaxed pressing. Distributions were K1 [3,5,4] and K2 [4,7,2]. ANOVA indicated significant between-cluster differences; Games–Howell showed K1 C0–C1 differences in winning and losing (mean diff +5.10; 95% CI [1.19, 9.01]) and K2 C0–C1 in draws (−2.14; [−3.95, −0.33]). The league × cluster distribution was non-significant (χ²(2, N=25)=1.105; small effect). Overall, the draw phase functions as a decision pivot: K1 tends to mitigate risk (relax pressing), whereas K2 tends to maintain or intensify initiative. PPDA is treated as an operational, indirect indicator; we recommend standardizing state-transition SOPs and refining back-space coverage and pressing triggers.