Frag' FlorenceEvidenz. Klar. Anwendbar.
Uhr 7/8Sources Journal Tree
Easy Demo

Lokaler Crossref-Datenbestand · journal-article

Differentiating playing roles in professional Counter-Strike 2 by performance indicators: an observational study

Volodymyr Paskiv, Igor Zanevskyy, Volodymyr Koman

Health, sport, rehabilitation · 2026

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background and purpose. In professional Counter-Strike 2 a single composite rating still carries most of the weight in judging players, even though the game grants no mechanical class asymmetry, so playing roles arise as a codified yet largely unexamined division of labour. We therefore asked whether those roles can be recovered from the standard per-round indicators, and how their differentiation is organised. Material and methods. Under a single twelve-month filter we drew thirty-nine per-round indicators for 1052 players across 222 ranked teams. Role labels came from two external sources (n = 402; Cohen's κ = 0.771 for leadership and 0.893 for the sniper), with snipers serving as a positive control. Two side-specific analyses (n = 318, 313), a side-agnostic playstyle analysis (n = 457) and a leadership tag (n = 159) were examined with Kruskal–Wallis (ε², Holm-corrected), MANOVA, random-forest classification coupled with SHAP, and k-means. Results. On the sniping indicators the snipers stood apart (η² 0.841–0.883). Differentiation of the tactical roles ran mainly through the opening block (ε² 0.12–0.27; 35–38 of 39 indicators Holm-significant; for entry, opening kills +0.92 SD and opening deaths +1.00 SD). Every analysis yielded a significant MANOVA (omnibus Pillai 1.13–1.30, p < 0.001; tactical pairs 0.37–0.50). The random forest recovered role in 77–80 % of cases (macro-F1 0.78–0.82), a figure team-wise cross-validation left intact. Clustering the tactical players returned an adjusted Rand of 0.08, and 63 % of attacking-side lurkers held a site as anchors on defence. Conclusions. Across the externally annotated roles the indicators draw distinctions in layers: a near-definitional weapon axis, a substantive tactical axis of round direction, and an orthogonal leadership tag, which together form a graded rather than sharply discrete, side-specific structure. Players therefore ought not to be compared across roles or sides on a single composite, and the trainable tactical signal sits in a compact set of opening and survival indicators.

Bibliografischer Nachweis

Publikationsdaten

Autor:innen
Volodymyr Paskiv, Igor Zanevskyy, Volodymyr Koman
Quelle
Health, sport, rehabilitation
Publikation
2026-01-01
Band / Ausgabe
Nicht angegeben
Seiten
Nicht angegeben
ISSN / ISBN
2520-2685, 2520-2677
Zitationen
0 laut Crossref
Referenzen
0 hinterlegt

Zitieren

Zitierfähiger Nachweis

Volodymyr Paskiv, Igor Zanevskyy, Volodymyr Koman (2026). Differentiating playing roles in professional Counter-Strike 2 by performance indicators: an observational study. Health, sport, rehabilitation. https://doi.org/10.58962/hsr.1395
RIS BibTeX CSL-JSON

Kontext

Themen, Förderung und Nutzung

Lizenzhinweise: Lizenz 1